36 AG11 Abstracts

IP1 IP5 Algebraic Statistics for Social Network Models: Advances in Elliptic-Curve Cryptography Recent Results and Challenges The first part of this talk presents results on attacking In collaboration with Petrovic and Rinaldo, I have devel- elliptic-curve cryptography, in particular an ongoing effort oped algebraic statistical tools for the study of some dyadic to break the Certicom challenge ECC2K-130 and a detailed random graph models, including Markov bases, that have study on the correct use of the negation map in the Pollard- important implications for the existence of maximum like- rho method. The second part presents a signature scheme lihood estimation and other statistical problems. These which on a 390 USD mass-market quad-core 2.4GHz Intel tools do not extend in a simple fashion to more complex Westmere (Xeon E5620) CPU can create 108000 signatures models in the class of exponential random graph models. In per second and verify 71000 signatures per second on an this presentation, I explain why there are difficulties as we elliptic curve at a 2128 security level. Public keys are 32 move away from dyadic models and I describe some of the bytes, and signatures are 64 bytes. These performance fig- challenges for algebraic statistics in this area of research. ures include strong defenses against software side-channel attacks: there is no data flow from secret keys to array in- Stephen Fienberg dices, and there is no data flow from secret keys to branch Carnegie Mellon University conditions. fi[email protected] Tanja Lange Technische Universiteit Eindhoven IP2 Eindhoven Institute for the Protection of Systems and Complexity of the Separation of Variables and Re- Inform lated Observations [email protected]

Abstract not available at time of publication. IP6 Leonid Gurvits B-splines: A Fundamental Tool for Analysis and Los Alamos National Laboratory Computation [email protected] Starting from their inception in approximation theory 60 years ago, the use of B-splines has blossomed in diverse IP3 areas in science, engineering and electronic arts. These in- Calculus and Constructible Sheaves clude animation, computational geometry, computer-aided design, computer-aided manufacturing, computer graphics, This talk describes an ingenious integral calculus based on control theory, geometric design, image analysis, medical Euler characteristic, stemming from work on constructible visualization, optimization, partial differential equations, sheaves due to MacPherson and Kashiwara in the 1970s, robotics, and statistics. More recently B-splines have been and connecting back further to classical integral geometry. assuming a fundamental role in the isogeometric analysis, The talk will emphasize (1) its novel utility in data man- an ambitious effort to unify shape representation and en- agement, particularly in aggregation of redundant data and gineering analysis. Honoring its interesting historical de- inverse problems over sensor networks; and (2) how issues velopment, this talk gives a mathematical introduction to of numerical computation, inspired by applications, leads splines and B-splines in one and several space dimensions to fascinating connections with Morse theory and compu- and also considers various generalizations. tational topology. Tom Lyche Robert W. Ghrist University of Oslo University of Pennsylvania Department of Informatics and CMA [email protected] tom@ifi.uio.no

IP4 IP7 The Geometry of Cell Structures in Foams and Symbolic-numeric Algorithms for Computing the Metals Singular Solution of Polynomial Systems Networks of crystals in a metal and networks of bubbles In this talk we will describe some recent progress in in a foam are both examples cell complexes that arise in symbolic-numeric algorithms for computing the singular nature. Both systems evolve over time according to mathe- solution of polynomial systems. Using the local dual struc- matical equations. We believe that, for generic initial con- ture of the isolated singular solution with limited accuracy, ditions, they evolve towards a statistically universal state. we present modifications of Newton’s method to restore The topology and geometry of this universal state is great quadratic convergence. In particular, when the corank of interest in applications. This talk will present both math- the Jacobian matrix at the singular solution is one, based ematical results and computer investigations on these sys- on the regularized Newton iteration and the computation tems in two and three space dimensions. of partial differential conditions satisfied approximately at Robert MacPherson the singular solution, we develop a new approach to com- Institute for Advanced Study pute a proper direction and step size of the Newton itera- School of Mathematics tion to ensure the quadratic convergence. Finally, we will [email protected] show recent results on computing verified error bounds for singular solutions of polynomial systems based on Rump’s verification method.

Lihong Zhi AG11 Abstracts 37

Academia Sinica CP1 [email protected] Extracting Topological Properties Using Manifold Learning Techniques IP8 High-dimensional, unordered data, is often difficult to an- Multiview Geometry alyze or visualize. In such cases, manifold learning tech- niques - like PCA, Isomap, Hessian LLE - are quite helpful The study of two-dimensional images of three-dimensional and efficient in extracting the low-dimensional representa- scenes is a foundational subject for computer vision, known tion of the data. When used individually, these methods as multiview geometry. We present recent work with Chris just give the low-dimensional representation of the data. Aholt and Rekha Thomas on the polynomials defining im- But owing to the complementary nature of the limitations ages taken by n cameras. Our varieties are threefolds that of these methods, performing a hierarchy of tests and com- vary in a family of dimension 11n-15 when the cameras are paring the outputs can reveal useful topological features moving. We use toric geometry and multigraded Hilbert like the presence of holes and (degree of) non-linearity of schemes to characterize degenerations of camera positions. the data. We develop and showcase such a hierarchical strategy on various material science data to provide insight into process-property-structure relationships. Bernd Sturmfels University of California , Berkeley Sai Kiranmayee Samudrala [email protected] Department of Mechanical Engineering, Ames, Iowa -50011 CP1 [email protected] Discrete Bounded-curvature Paths and Path Plan- ning Baskar Ganapathysubramanian In 1957 Dubins classified shortest bounded curvature paths Iowa State University in the plane into two classes. We generalize this result to [email protected] a new class of discrete polygonal paths. The properties and construction of discrete Dubins paths are discussed. CP1 Properties of discrete bounded-curvature motion are anal- ysed and applied to prove properties of smooth bounded- Lower Bounds on Stochastic Games curvature paths. In particular, the classification of discrete Shapley’s discounted stochastic games are classical models Dubins paths is used to obtain Dubins seminal result as a of game theory describing two-player zero-sum games of limiting case. Related to these problems the question of potentially infinite duration. We show, based on a gener- defining curvature for a polygonal path is addressed. Po- alization of Eisenstein criterion, that there exists a game tential research questions, such as generalizing Cauchy’s with N positions, m actions for each player in each posi- arm lemma, are discussed. tion, such that its value is an algebraic number of degree mN−1 Sylvester Eriksson-Bique . This strengthens a result of Etessami and Yan- nakakis who considered the case of m =2andproveda University of Helsinki Ω(N) New York University 2 lower bound. [email protected] Kristoffer A. Hansen Aarhus University, Denmark David Kirkpatrick [email protected] University of British Columbia, Canada [email protected] Michal Koucky Czech Academy of Sciences Valentin Polishchuck [email protected] University of Helsinki [email protected].fi Niels Lauritzen, Peter Bro Miltersen Aarhus University, Denmark CP1 [email protected], [email protected] On Middle Universal Weak and Cross Inverse Property Loops With Equal Length of Inverse Cy- Elias Tsigaridas cles University of Aarhus [email protected] This study presents an isotopism under which the weak inverse property(WIP) is isotopic invariant in loops. It is shown that under this isotopism, whenever n is a positive MS1 even integer, a finite WIPL has an inverse cycle of length Symmetric Determinantal Representations of Poly- n if and only if its isotope is a finite WIPL with an in- nomials verse cycle of length n. Explanations and procedures are givenonhowtheseresultscanbeusedtoapplyCIPLsto In a seminal paper (STOC 1979), Valiant expressed the cryptography. polynomial computed by an arithmetic formula as the de- terminant of a matrix whose entries are constants or vari- Temitope G. Jaiyeola ables. This result was then extended by Malod and Toda Obafemi Awolowo University to weakly-skew circuits. We are interested here in express- [email protected] ing formulas and weakly-skew circuits by determinants of symmetric matrices. In my talk, I will sketch some of these 38 AG11 Abstracts

constructions for fields of characteristic different from 2. In tion which by definition accepts closed sets only. Also, we characteristic 2, the picture is quite different: First we can explore which problems in NP can be reduced to the com- prove an impossibility result, showing that some polynomi- pact feasibility problem using an effective finiteness Theo- als have no symmetric determinantal representation. On rem involving Lojasiewicz’ inequality. the other hand, we answer a question by B¨urgisser about the VNP-completeness of the partial permanent. This talk Saugata Basu is based on joint works with E. L. Kaltofen, P. Koiran and Purdue University N. Portier, and to a smaller extent with T. Monteil and S. [email protected] Thomass´e. Peter Scheiblechner Bruno Grenet Department of Mathematics, Purdue University LIP, ENS Lyon [email protected] [email protected]

MS1 MS1 On the Complexity of Computing with Zero- Solving the Generalized MinRank Problem with dimensional Triangular Sets Groebner Bases We study the complexity of some fundamental operations Let K be a field and M be a p × k matrix with entries in for triangular sets in dimension zero. Using Las-Vegas al- K[X1,...,Xn]ofdegreeD and 0

INRIA, Paris-Rocquencourt secant variety of the Veronese embedding of original pro- [email protected] jective space and the linear equations. This reduces the question of finding the equations of the secant variety of Pierre-Jean Spaenlehauer a (sufficiently amply embedded) variety to the finding of UPMC, Univ. Paris 6 the equations of the secant variety of a Veronese embed- [email protected] ding of a projective space. Time permitting, context and other results will be mentioned. This is joint work with J. Buczynski and J.M. Landsberg. MS2 Secant Varieties of Segre-Veronese Varieties Adam Ginensky WH Trading In general, the secant varieties of Segre-Veronese varieties [email protected] can be difficult to describe explicitly. In many cases, the dimensions are not even known. I will explain the following theorem: for r at most 5, the rth secant variety of P 2 × MS2 P n embedded by O(∞, ∈) is defined by the minors of the Border Bases, Hilbert Scheme and Tensor Decom- flattening and the Pfaffian of the exterior flattening. position Dustin Cartwright The commutativity is a natural property that we expect in Yale University the context of algebraic geometry. Surprisingly, this sim- [email protected] ple property is also enough to characterize the solution of several problems. In this talk, we will show how it is in- Daniel Erman trinsically related to the description of the Hilbert scheme Stanford University of points and to the tensor decomposition problem. We [email protected] will analyze the correspondence between these two prob- lems and detailed different characterisation of ranks of a tensor. Revisiting an approach of J.J.Sylvester for the de- Luke Oeding composition of a binary forms, we will describe an algo- University of California Berkeley rithm to decompose general tensors, based on techniques Universit`a degli Studi di Firenze related to border basis computation and truncated moment [email protected] problems. Some examples will illustrate the approach. Bernard Mourrain MS2 GALAAD Tensor Complexes INRIA Sophia Antipolis [email protected] The most fundamental complexes of free modules over a commutative ring are the Koszul complex, which is con- structed from a vector (i.e., a 1-tensor), and the Eagon– J´erˆome Brachat Northcott and the Buchsbaum–Rim complexes, which are GALAAD, INRIA Sophia Antipolis constructed from a matrix (i.e., a 2-tensor). I will discuss [email protected] a multilinear generalization of these complexes, which we construct from an arbitrary higher tensor. Our construc- MS2 tion provides detailed new examples of minimal free reso- lutions, as well as a unifying view on several well-known Decomposition of Tensors of Small Rank examples. We discuss some algorithms to decompose a tensor in A ⊗ B ⊗ C Berkesch Christine as a sum of decomposable tensors. The technique Duke University is connected to the equations of the secant varieties, which [email protected] are known only when the rank is small. The results are better in the symmetric case. Some sufficient conditions which guarantee the uniqueness of the decomposition will Daniel Erman be presented. Stanford University [email protected] Giorgio Ottaviani University of Firenze Manoj Kummini [email protected]fi.it Purdue University [email protected] MS3 Steven Sam Polynomials and Optimization in Computer Vision MIT [email protected] Multiview geometry is the study of planar images of points in space, an essential basis for computer vision. I will de- scribe the algebraic objects in this field. These include MS2 the multiview variety associated to a fixed set of cameras Determinantal Equations for Secant Varieties – along with an explicit universal Groebner basis for its ideal – and also a Hilbert scheme parametrizing all such We show that if a smooth variety X is re-embedded by multiview ideals. This data feeds into the polynomial opti- a sufficiently large Veronese embedding of its original em- mization problem of triangulating a point from noisy image bedding then, set theoretically, the equations of the rth observations. secant variety of X, are just the equations defining the rth Chris Aholt 40 AG11 Abstracts

University of Washington involve the comparison of certain invariants arising from [email protected] persistent topology constructions. The computation of these lower bounds can be done in polynomial time. Bernd Sturmfels University of California , Berkeley Facundo Memoli [email protected] Stanford University, Mathematics Department [email protected] Rekha Thomas University of Washington MS4 [email protected] Toric Ideals of Hypergraph Models This talk is about models parametrized by edges of a hy- MS3 pergraph. Such models are toric by definition, and the goal Lossless Representation of Graphs using Distribu- is to obtain a Markov basis for the model. Equivalently, tions we are interested in the defining ideal of the edge subring of a hypergraph. The ideal of the edge subring of a graph Complex structures such as composite objects or networks gives a Markov bases for the p1 model for random graphs are often represented by weighted graphs. Since many sig- (networks). The algebraic and combinatorial constructions nal processing and analysis techniques require the input to used for these toric ideals give insight into the geometry take the form of a vector, it is desirable to find ways to rep- of the model, existence of maximum likelihood estimators, resent such weighted graphs as vectors. For reasons that and a better way to generate Markov moves. The ideal the- will be explained in the talk, it is important to make sure ory of graphs has a rich history, including some results that that any invariance of the structure under a relabelings of are well-known in algebraic statistics. A more complicated the node carry over to the vector representation scheme family of algebraic models motivates the generalization of chosen. In this talk, I will discuss how one can use distri- this construction to hypergraphs. The talk will outline the butions of certain functions of the weights of the graph to motivation and the main results. represent the ”vast majority” of graphs without losing any information about the structure of the graph. Sonja Petrovic University of Illinois, Chicago Mireille Boutin [email protected] School of ECE Purdue University [email protected] Despina Stasi Mathematics, Statistics, and Computer Science University of Ilinois at Chicago MS3 [email protected] Title Not Available at Time of Publication

Abstract not available at time of publication. MS4 The Polytope of Degree Sequences and Maximum Gunnar E. Carlsson Likelihood Estimation Stanford University [email protected] In this talk, we show how the polytope of degree sequences can be used to characterize nonexistence of the maximum likelihood estimator (MLE) in two statistical models: the MS3 beta model for random undirected graphs with fixed degree A Categorical Approach to Agent Networks sequences and the Rasch model from item response theory. We derive new asymptotic results about the existence of In this research, we continue the work of Lawvere and Giry the MLE in the random graph case and describe numeri- in developing a categorical approach to conditional prob- cally feasible ways for checking for existence of the MLE abilities. Specifically, we investigate categorical properties and for computing the associated facial sets. of the Kleisli category of the Giry monad with the aim of constructing relations. Our goal is to then use this cate- Alessandro Rinaldo gorical framework to model sensor networks in an abstract, Carnegie Mellon University relational and probabilistic way. [email protected]

Jared Culbertson Sonja Petrovic Air Force Research Laboratory University of Illinois, Chicago [email protected] [email protected]

MS3 Stephen Fienberg Carnegie Mellon University Lower Bounds for the Gromov-Hausdorff distance fi[email protected] Using Persistent Topology

Viewing shapes as metric spaces (or metric measure spaces) MS4 permits dealing with different problems in object matching under deformations. The Gromov-Hausdorff distance pro- An Algebraic Statistics Approach to the Ecological vides one possible measure of dissimilarity between shapes, Inference Problems but its computation leads to NP-hard problems. We will We present an algebraic computational framework that discuss a family of lower-bounds for the GH distance that handles special cases of missing data problems. We focus AG11 Abstracts 41

on the ecological inference problem where we use aggregate of polynomial, piecewise polynomial, and rational curves (ecological) data to infer discrete individual-level relation- and surfaces to industrial design and manufacturing. The ships of interest when individual-level data are unavailable. purpose of this talk is to elucidate the role that Algebraic The proposed method can handle multi-dimensional con- Geometry plays in Geometric Modeling. We will discuss tingency tables, deterministically respects bounds and can several topics of current research, including elimination incorporate information from many different sources. We theory, syzygies, and the analysis of singularities. illustrate how the analysis can be done with the recently developed R-4ti2 package. Ron Goldman Department of Computer Science Aleksandra Slavkovic,VisheshKarwa Rice University Penn State University [email protected] [email protected], [email protected] MS5 MS4 Wachspress Varieties Estimating the Number of Binary Multi-way Ta- bles via Sequential Importance Sampling We will examine an algebraic variety parametrized by barycentric coordinates of a polygon. We will make use In 2005, Chen et al introduced a sequential importance of the combinatorics of the polygon to understand the ge- sampling (SIS) procedure to analyze binary two-way tables ometry of this variety. In addition, we will observe that with given fixed marginal sums (row and column sums) via this variety can be seen as the blowup of certain points in the conditional Poisson (CP) distribution. They showed the projective plane. that compared with Markov chain Monte Carlo (MCMC)- based approaches, their importance sampling method is Corey Irving more efficient in terms of running time and it also provides Department of Mathematics an easy and accurate estimate of the total number of con- Texas A&M University tingency tables with fixed marginal sums. In 2007, then, [email protected] Chen showed also that we can generalize it to binary two- way tables with structural zeros. In this talk we will review theirresultsandthenwewillshowthatwecanextendtheir MS5 result to binary multi-way (d-way, d ≥ 2) contingency ta- Splines, Spectral Sequences, and Polyhedral Com- bles under the no d-way interaction model, i.e., with fixed plexes d − 1 marginal sums. Piecewise polynomial functions on a polyhedral complex Jing Xi (splines) are fundamental objects in mathematics, with ap- Department of Statistics plications ranging from approximation theory and numer- University of Kentucky ical analysis to algebraic geometry, where they appear as [email protected] the equivariant Chow ring of a toric variety. In joint work with T. McDonald [Advances in Applied Math, 2009], we Ruriko Yoshida found an (asymptotic) version of the Alfeld-Schumaker di- University of Kentucky mension formula for planar splines. I will also describe [email protected] recent work [Trans. A.M.S, to appear] which uses the Cartan-Eilenberg spectral sequence to obtain results on splines on polyhedral complexes of dimension greater than MS5 two. Computing Vanishing Ideals with the Help of Mul- Hal Schenck tivariate Polynomial Interpolation Mathematics Department In this paper, the problem of computing vanishing ideals University of Illinois is investigated with the help of multivariate polynomial [email protected] interpolation. We generate certain subsets from the given point set whose associated interpolation bases are obtained MS5 subsequently in theory, with which the processes of known algorithms for vanishing ideals can be accelerated to some Multivariate Interpolation and Algebraic Geome- extent, according to the sizes of the subsets. Algorithms try are implemented and experimental timings are given. I will describe connections between multivariate interpo- Tian Dong lation and algebraic geometry and present two counterex- School of Mathematics amples to problems in the field. First is to a conjecture of JilinUniversity Carl de Boor regarding the existence of certain error for- [email protected] mula for multivariate interpolation. Second is to a conjec- ture of Tomas Sauer regarding restrictions of interpolation projector to polynomials of lesser degree. MS5 Boris Shekhtman The Role of Algebraic Geometry in Geometric Department of Mathematics and Statistics Modeling University of South Florida Algebraic Geometry and Geometric Modeling both deal [email protected] with curves and surfaces generated by polynomial equa- tions. Algebraic Geometry studies theory, algorithms, and computation; Geometry Modeling investigates applications 42 AG11 Abstracts

MS6 2-dimensional domain. We seek to compute the probability Persistence Based Signatures for Metric Spaces that a set of sensors fails to cover given only non-metric, local (who is talking to whom) information and a probabil- We introduce a family of signatures for compact metric ity distribution of failure of each node. This builds off the spaces, possibly endowed with real valued functions, based work of de Silva and Ghrist who analyzed this problem in on the persistence diagrams of suitable filtrations built on the deterministic situation. We first show that the problem top of these spaces. We prove the stability of these signa- as stated is #P-complete, and thus fast algorithms likely tures under Gromov-Hausdorff perturbations of the spaces. do not exist unless P=NP. We then give a deterministic We illustrate their use through an application in shape clas- algorithm which is feasible in the case of a small set of sen- sification. sors or at least a set of sensors which is not very dense, and discuss other methods to approximate the probability Frederic Chazal of failure. These algorithms build on the theory of topo- INRIA logical persistence. [email protected] Elizabeth Munch Duke University MS6 [email protected] Rigid Rips Complexes and Topological Data Anal- ysis MS6 In recent years, persistence diagrams have become an inte- Topology of Spaces of Micro-images and Applica- gral part of topological data analysis, providing a tool to tions measure the scale of topological features. From the theo- retical point of view, persistence diagrams can be seen as One of the most celebrated success stories in topological effectively computable invariants of metric spaces. A clas- data analysis is perhaps that of the Klein bottle, as a model sical way to construct a persistence diagram is by way of space for relevant 3×3 patches coming from natural images. the so-called Rips complexes. In this paper we define a I will explore in this talk how the Klein bottle model can be 1-parameter family of simplicial complexes which we call extended to include other meaningful patches, and describe rigid Rips complexes. This family includes the regular Rips an application to texture discrimination. complex at infinity and yields a 1-parameter family of per- sistence diagrams - the data which we call rigid persistence. Jose Perea In addition to measuring the scale of topological features, Stanford University this family provides a tool to analyze the evolution of these [email protected] features when the rigidity changes. While rigid persistence contains the regular persistence, we give an example show- ing that the reverse of that statement is not true, and that MS7 rigid persistence contains strictly more information that Riemann Theta Functions: An Applied Introduc- the regular one. (joint with E.-M. Feichtner, J. Lehmann) tion Dmitry Feichtner-Kozlov Riemann’s theta function is the fundamental building block University of Bremen for function theory on compact Riemann surfaces and [email protected] their Jacobians. In this talk I will present an incomplete overview of different applications where theta function play an important role. MS6 Topology, Geometry and Statistics: Merging Bernard Deconinck Methods for Data Analysis University of Washington [email protected] Dimension reduction and shape description for scientific data sets are difficult problems, ones that continue to grow in importance within the statistical, mathematical and MS7 computer science communities. Powerful new methods of Tropical Curves, Tropical Theta Functions and In- Topological Data Analysis and Diffusion Geometry have tegrable Systems emerged in the last 10 years, and these have added signif- icantly to the data analysis toolbox. In this talk we will Tropical geometry is the combinatorial algebraic geome- give an overview of these methods and describe some early try over the min-plus algebra. For the last dozen years, efforts to make them work together with Statistics. In tropical geometry has been rapidly developed and widely particular we will discuss how random walk methods give applied not only in complex algebraic geometry, but also in distance functions that enhance topological features, and mathematical physics. The aim of this talk is to introduce how one goes about defining the mean and variance of a an application of tropical curve theory to an integrable collection of persistence diagrams. piecewise-linear map which originates from the Toda lattice equation. First we recall basic notions of tropical curves, John Harer tropical Jacobian and tropical theta functions by follow- Duke University ing the work of Mikhalkin and Zharkov in 2006. Second, [email protected] a tropical version of Fay’s trisecant identity is formulated. As a special case of this identity, we finally obtain a general solution to the piecewise-linear map. (This talk is based MS6 on the collaborative works with T. Takenawa.) Using Persistent Homology to Compute Probabal- istic Failure of a Sensor Network Rei Inoue Faculty of Science We consider the question of sensor network coverage for a Chiba University AG11 Abstracts 43

[email protected] BenGurionUniversity [email protected]

MS7 Algebraic Curves and Riemann Surfaces in Matlab MS8 Closed Form Solutions of Difference Equations We present a fully numerical approach to Riemann surfaces defined via plane algebraic curves. The code in Matlab In this talk we present an algorithm that finds closed computes for a given algebraic equation in two variables form solutions for homogeneous linear recurrence equa- the branch points and singularities, the holomorphic dif- tions. The key idea is transforming an input operator Linp ferentials and a base of the homology. The monodromy to an operator Lg with known solutions. The main prob- group for the surface is determined via analytic continu- lem of this idea is how to find a solved equation Lg to which ation of the roots of the algebraic equation on a set of Linp can be reduced. To solve this problem, we use local contours forming the generators of the fundamental group. data of a difference operator, that is invariant under the The periods of the holomorphic differentials are computed transformation. with Gauss-Legendre integration along these contours. The Abel map is obtained in a similar way. The performance of Yongjae Cha the code is illustrated for many examples. As an applica- RISC Linz tion we study quasi-periodic solutions to certain integrable [email protected] partial differential equations.

Christian Klein MS8 Institut de Math´ematiques de Bourgogne On the Structure of Compatible Rational Functions 9 avenue Alain Savary, BP 47870 21078 DIJON Cedex [email protected] A finite number of rational functions are compatible if they satisfy the compatibility conditions of a first-order linear functional system involving differential, shift and q-shift MS7 operators. We present a theorem that describes the struc- Thetanulls of Algebraic Curves of Genus 3 (A Pre- ture of compatible rational functions. The theorem en- liminary Report) ables us to decompose a solution of such a system as a product of a rational function, several symbolic powers, a In this talk we will discuss thetanulls of genus 3 hyperel- hyperexponential function, a hypergeometric term, and a liptic curves. The case of genus 3 hyperelliptic curves has q-hypergeometric term. We outline an algorithm for com- been studied by Shaska/Wijesiri and relations among them puting this product, and present an application. have been determined for cyclic curves of small genus. In this talk the more general case for genus 3 will be discussed. Shaoshi Chen Research Institute for Symbolic Computation Tanush Shaska Johannes Kepler University Linz Department of Mathematics [email protected] Oakland University [email protected] Ruyong Feng, Guofeng Fu, Ziming Li Key Laboratory of Mathematics Mechanization AMSS, Chinese Academy of Sciences MS7 [email protected], [email protected], Determinantal representations, Cauchy kernels, [email protected] and theta functions

Determinantal representations of plane algebraic curves MS8 were the object of numerous studies during the last cen- The Q=0 Expansion of Q-Holonomic Sequences tury and a half. They became important for optimiza- tion in the last decade since representing a convex set as I will discuss recent results on the degree, leading term and a feasibility set for semidefinite programming boils down q=0 expansion of a q-holonomic sequence of polynomials. to constructing a positive symmetric determinantal repre- The results are of interest in Knot Theory and Quantum sentations (for certain multiples of the Zariski closure of Topology, as well as the physics of fivebranes. the boundary). One can obtain explicit formulae for de- terminantal representations using theta functions. These Stavros Garoufalidis formulae are closely related to the famous trisecant iden- Georgia Tech tity of Fay. Similarly to Fay’s identity, these formulae can [email protected] be easily derived tracing the poles and residues of various expressions involving Cauchy kernels for line bundles on a compact Riemann surface. Here the Cauchy kernel is a MS8 section in two arguments and is holomorphic except for a Restricted Lattice Walks and Creative Telescoping simple pole with residue one along the diagonal; it can be easily written down using theta functions (and the prime We start with a combinatorial problem: determining the form). The purpose of the talk is to describe the various generating functions of certain restricted lattice walks. interrelations between the different objects: determinan- This can be done with support from computer algebra. tal representations, theta functions, and Cauchy kernels. The most efficient techniques are however not ’rigorous’ Many of the results are based on joint work with Daniel (joint work with A. Bostan; FPSAC 2009). It is possible Alpay, Joe Ball, Bill Helton, and Dmitry Kerner. to make them rigorous by evaluating certain definite inte- grals via Zeilberger’s method of creative telescoping. But Victor Vinnikov the computational cost for doing so is quite high (ongoing Department of Mathematics joint work with A. Bostan, F. Chyzak, M. van Hoeij). This 44 AG11 Abstracts

leads to the general question whether creative telescoping valued tensors of border rank 4 at most is given by poly- can be done in a more efficient way. In the end, we sketch nomial equations of degree 5, (the Strassen commutative an idea for minimizing the cost for creative telescoping by conditions), of degree 6, (the Landsberg-Manivel polyno- using a Cylindrical Decomposition computation as a pre- mials), and of degree 9, (the symmetrization conditions). processing step (ongoing joint work with S. Chen).

Manuel Kauers Shmuel Friedland RISC Linz Department of Mathematics [email protected] University of Illinois, Chicago [email protected] MS8 Computing Belyi Maps MS9 Tensor Decomposition and Dimensions of Secant Let L(y)=0 be a Heun equation with no Liouvillian solu- Varieties of Segre Varieties tions. We consider the question of deciding if there exists a rational function f for which L can be solved in terms I will discuss joint work with M.V. Catalisano (Genoa) and of 2F1(a,b;c;f). Here a,b,c are constants, and 2F1 is the A. Gimigliano (Bologna) giving the dimensions of ALL the hypergeometric function. Computationally, the most non- secant varieties of the Segre Varieties P 1 ×···×P 1.This trivial case occurs when f is a Belyi map. In a joint work solves the problem of finding the number of summands in with R. Vidunas, we have computed a complete table with the decomposition of a general tensor in V ⊗···⊗V (any all Belyi maps f that can occur in this context. It con- number of factors) when dim V = 2. Similar results for the tains 48 + 366 Belyi maps (48 parametric cases and 366 cases when V has higher dimension will also be mentioned. non-parametric cases). The degrees in this table reach up to degree 60, finding these Belyi maps and verifying com- pleteness required new techniques. Anthony Geramita Queen’s University Mark van Hoeij [email protected] Florida State University [email protected] MS9 Representations with Finitely Many Orbits MS9 Report on the Recent Progress on the Study of In this talk I will report on the joint work with Witold Secant Defectivity of Segre-Veronese Varieties Kraskiewicz on singularities of orbit closures in the repre- sentations of reductive groups with finitely many orbits. In this talk, I report the recent progress on the secant de- We are interested in Cohen-Macaulay, Gorenstein and ra- fectivity of Segre-Veronese varieties. The primary goal of tional singularities properties as well as in defining ideals the talk is to present the conjecturally complete list of de- of the orbit closures. fective secant varieties of Segre-Veronese varieties with two factors. If time permits, I will discuss defective secants to Jerzy Weyman Segre-Veronese varieties with three or more factors. This talk is based on joint work with Chiara Brambilla. [email protected] Hirotachi Abo University of Idaho MS10 [email protected] Computing Explicit Optimal Value Functions by a Symbolic-numeric Cylindrical Algebraic Decompo- Chiara Brambilla sition Dip. di Scienze Matematiche Universita‘ Politecnica delle Marche, Ancona, Italy We consider to solve parametric optimization problems. [email protected]fi.it Our purpose is to compute optimal values as a function of the parameters, which is called an optimal value func- tion. This is very useful for solving reactive/online op- MS9 timizations, dynamic programming, hierarchical optimiza- Kruskal’s Uniqueness Inequality is Sharp tions and so on. Thus constructing explicit optimal value functions is very important and has many applications in

Kruskal proved that a tensor in V1 ⊗V2 ⊗···⊗Vm of rank r engineering. We present en efficient method, which is based has a unique decomposition as a sum of r pure tensors if a on a specialized cylindrical algebraic decomposition using certain inequality is satisfied. We will show the uniqueness symbolic-numeric computation, to construct optimal value fails if the inequality is weakened. functions explicitly for polynomial parametric optimization problems. Moreover, we also show several practical appli- Harm Derksen cation examples in control. University of Michigan [email protected] Hirokazy Anai Kyushu University [email protected] MS9 A Proof of the Set-theoretic Version of a Salmon Conjecture MS10 Cylindrical Algebraic Decomposition without Aug- We show that the irreducible variety of 4 × 4 × 4complex AG11 Abstracts 45

mented Projection will be given also.

Cylindrical algebraic decomposition is a fundamental tool Mohab Safey El Din in computational real algebraic geometry. It takes a set of University Pierre and Marie Curie multivariate polynomials and decomposes the real space [email protected] into cylindrically arranged cells which can be described by boolean expressions of polynomial equations and in- equalities. It begins by repeated projection (elimination of MS10 variables) and follows by repeated lifting. During the lift- A Divide-and-conquer Algorithm for Roadmap ing, the descriptions of cells are constructed using Thoms Computation lemma. This necessitates the so-called augmented projec- tion (the derivatives are also considered). It often makes We consider the problem of constructing roadmaps of real the projection time-consuming and in turn increases the algebraic sets. This problem was introduced by Canny to number of cells. In this talk, we show how to avoid aug- answer connectivity questions and solve motion planning mented projection by describing cells using Sturm-Habicht problems. Given s polynomials with rational coefficients, of degree D in n variables, Canny’s algorithm has a Monte theorem (instead of Thoms lemma). 2 Carlo cost of sn log(s)DO(n ) operations in Q.Anim- Hoon Hong provement by Basu, Pollack and Roy has a deterministic North Carolina State University ( ) ( 2) cost sO n DO n , for the more general problem of comput- [email protected] ing roadmaps of a semi-algebraic set. We recently obtained 1.5 a probabilistic algorithm of cost (nD)O(n ) for the prob- MS10 lem of computing a roadmap of a compact and smooth D Composing Hybrid Automata hypersurface of degree . In this talk, we show how to im- prove this algorithm using divide-and-conquer techniques, We address the problem concerning the decidability of cer- and extend it to compact and smooth algebraic sets. This tain hybrid automata that may be built hierarchically as requires us to introduce, and control the degree of, multi- is the case in many exemplar systems, be it natural or en- homogeneous systems corresponding to the iterated intro- gineered. Parallel composition can be considered a funda- duction of Lagrange multipliers. mental tool in such costructions. Somewhat surprisingly, this operation does not always preseve the decidability of Eric Schost reachability problem i.e., even if we prove the decidabil- Computer Science Department ity of reachability over component automata, we cannot Univ. of Western Ontario guarantee the decidability over their parallel composition. [email protected] Despite these limitations,it is still possible to provide a re- duction for the reachability problem over parallel compo- Safey El Din Mohab sition of first-order constant reset automata (FOCoRe) to University Pierre and Marie Curie the satisfiability of a particular linear Diophantine system. [email protected] Moreover, by proving that such satisfiability problem is decidable for systems with semi-algebraic coefficients, this paper presents an interesting class of hybrid automata for MS11 which the reachability problem of parallel composition is Extensions and Applications of Parametric Align- decidable. The resulting hybrid automata appear in sys- ment tems biological modeling, and hence could be applied when one is interested in understanding a complex biological sys- We review extensions and applications of parametric se- tem composed of smaller self-organizing systems. Joint quence alignment, including recently published and sub- work with Alberto Casagrande, Pietro Corvaja, and Carla mitted works. Parametric alignment (or RNA folding, Piazza, expanding on some earlier preliminary results. etc.) gives rise to sum-product algorithms over semirings of polytopes. Generalizing parametric alignment to the Bud Mishra k-best alignment setting leads to k-set polytopes and re- New York University lated objects. General results about lattice polytopes give [email protected] best-known output complexity and running time bounds for several extensions of parametric alignment and related problems. MS10 The Critical Point Method into Practice : State of Peter Huggins the Art and Perspectives University of Kentucky [email protected] The Critical Point Method has been designed initially to provide an algorithmic alternative to Cylindrical Alge- braic Decomposition for computing sample points in semi- MS11 algebraic sets or for performing quantifier elimination over Algebraic Models in Systems Biology the reals. One feature is that it provides a framework to obtain algorithms running in asymptotically optimal de- Progress in systems biology relies on the use of mathemat- gree bounds. In this talk, I will present a new algorithm ical and statistical models for system level studies of bi- for computing sample points in semi-algebraic sets defined ological processes. Several different modeling frameworks by equations and inequalities when the set of equations sat- have been used successfully, including traditional differen- isfies some regularity assumptions. Practical experiments tial equations based models, a variety of stochastic models, show that it outperforms previously known algorithms on agent-based models, and Boolean networks, to name some some examples. Some preliminary results on its complexity common ones. This talk will focus on several types of dis- crete models, and will describe a common mathematical approach to their comparison and analysis, which relies 46 AG11 Abstracts

on computational algebraic geometry. A software package, ities which allows to use classical algebraic tools without Analysis of Dynamic Algebraic Models (ADAM) will be referring to the real algebraic geometry. This led for ex- presented. ample to the method of phylogenetic invariants. Only re- cently some effort has been put to understand the full semi- Reinhard Laubenbacher algebraic structure of graphical models with hidden data. Virginia Bioinformatics In my talk I will make a small step back trying to explain Institute why do we care about inequality constraints. I will show [email protected] how they may improve our understanding of statistical in- ference. My analysis is very simplistic and focuses on the simplest naive Bayes model. However, the main ideas will MS11 generalize to more complicated models. A Semi-algebraic Description of the of the General Markov Model on Phylogentic Trees Piotr Zwiernik Mittag-Leffler Institute The phylogenetic variety for the general Markov model Stockholm, Sweden (GM) on a phylogenetic tree is the closure of the image of [email protected] a parameterization map over the complex numbers of this k-state substitution model. Since the variety arises from a closure operation over the complex numbers, there are MS12 points in the variety that satisfy the defining polynomials, Algebraic Geometry for Analysis of Microscope but do not correspond to real stochastic parameters for the Images model. Therefore a semi-algebraic description of the image is needed. We begin with the case of 3-dimensional tensors This talk will describe potential applications for algebraic and 2 states and describe several approaches for giving the geometry in the analysis of microscope images. semi-algebraic conditions that a three dimensional tensor is in the image of the real parameterization map for the GM Mary Comer model. Then we discuss generalizations of this description School of ECE to more leaves and more states. Purdue University [email protected] Amelia Taylor Colorado College [email protected] MS12 Signal Registration via Polynomial System Solu- Elizabeth S. Allman tion Method and Image Recognition via Moment University of Alaska Fairbanks Method [email protected] We will introduce a new method for solving over- constrained polynomial systems of equations that can be John Rhodes applied to register a query object in 1D or 2D onto a cor- University of Alaska, Fairbanks responding template. Also we will introduce the ”shape [email protected] dictionary”, which is setted up by extracting a set of mo- ment features from the new moments we define and can be MS11 applied in image recognition. Reverse Engineering of Regulatory Networks Using Shanshan Huang Algebraic Geometry School of ECE Purdue University Discrete models have been used successfully in modeling bi- [email protected] ological processes such as gene regulatory networks. When certain regulation mechanisms are unknown it is important to be able to identify the best model with the available MS12 data. In this context, reverse engineering of finite dynami- Finding the Non-reconstructible Locus cal systems from partial information is an important prob- lem. In this talk we will present a framework and algorithm Given the unordered collection of the pairwise distances to reverse engineer the possible wiring diagrams of a finite of a finite point configuration in affine space, Boutin and dynamical system from data. The algorithm consists on Kemper proved that one can recover the configuration up using algebraic sets to encode all possible wiring diagrams, to Euclidean motion from these distances if the configu- and choose those that are minimal using the irreducible ration is generic. In special cases, non-equivalent config- components. urations may give the same distance set, but these non- reconstructible cases are contained in an algebraic hyper- Alan Veliz-Cuba surface as Boutin and Kemper show also. University of Nebraska Lincoln [email protected] David Knott, Uli Walther Department of Mathematics Purdue University MS11 [email protected], [email protected] Why Do We Care about Inequalities in (Algebraic) Statistics? Michael Burkhart Graphical models with hidden nodes usually have a compli- Purdue University cated description involving inequality constraints. A stan- [email protected] dard approach in algebraic statistics is to ignore inequal- AG11 Abstracts 47

MS12 MS13 Object-image Correspondence Under Projections Chamber Cones and Fast Solving over Local Fields We provide criteria for deciding whether a given planar Abstract not available at time of publication. curve is an image of a given spatial curve, obtained by a parallel or central projection with unknown parameters. J. Maurice Rojas These criteria reduce the projection problem to a certain Texas A&M University modification of the equivalence problem of planar curves [email protected] under affine and projective transformations. The latter problem can be addressed using Cartan’s moving frame method. The computational advantage of the method MS13 presented here, in comparison to algorithms based on a Symbolic-numeric Methods for Near-singular Sys- straightforward solution, lies in a significant reduction of a tems number of real parameters that has to be eliminated in or- der to establish existence or non-existence of a projection In this talk we will discuss recent developments on handling that maps a given spatial curve to a given planar curve. multivariate polynomial systems which are near singular The same technique can be applied to solve object image ones, using hybrid symbolic-numeric methods. First we correspondence for finite lists of points. will consider structure preserving iterative methods which find the distance to- and the nearest element on certain Joseph Burdis, Irina Kogan discriminant varieties. Second, we will consider methods North Carolina State University that turn ill-conditioned systems (with clusters of roots) [email protected], [email protected] into well-conditioned ones such that the roots of the new system are at the center of gravity of the clusters.

MS12 Agnes Szanto Invariant Histograms and Signatures for Object North Carolina State University Recognition and Symmetry Detection in Images [email protected]

I will survey recent developments in the use of group- invariant histograms, using distances, areas, etc., and sig- MS14 natures, using differential invariants, joint invariants, in- Morse Theory in Topological Data Analysis variant numerical approximations, etc., for object recogni- tion and symmetry detection in images. We introduce a method for analyzing high-dimensional data. Our approach is inspired by Morse theory and uses Peter Olver the nudged elastic band method from computational chem- Mathematics Department istry. As output, we produce an increasing sequence of cell University of Minnesota complexes modeling the dense regions of the data. We test [email protected] the method on data sets arising in social networks, in image processing, and in microarray analysis, and we obtain small cell complexes revealing informative topological structure. MS13 Title Not Available at Time of Publication Henry Adams Abstract not available at time of publication. Stanford University [email protected] Martin Avendano Texas A&M University Atanas Atanasov [email protected] [email protected] MS13 Gunnar E. Carlsson Bounding the Sum of Square Roots via Lattice Re- Stanford University duction [email protected] Let k and n be positive integers. Define R(n, k)tobethe minimum positive value of MS14 Expansion of Random Simplicial Complexes √ √ √ |e s e s ··· e s − t| i 1 + 2 2 + + k k Expander graphs are widely studied in theoretical com- puter science and discrete mathematics and have now found a host of applications. There has been recent interest s ,s , ···,s n where 1 2 k are positive integers no larger than , in defining and exhibiting higher-dimensional analogues of t is an integer and ei ∈{1, 0, −1} for all 1 ≤ i ≤ k.Itis expander graphs, by Gromov and others. I will discuss re- important in computational geometry to determine a good cent joint work with Dotterrer, where we show that many lower and upper bound of R(n, k). In this talk we will kinds of random cell complexes satisfy a certain ”cobound- present an algorithm to find lower bounds based on lat- ary expansion” property which is a generalization of edge tice reduction algorithms. It produces lower bounds much expansion of graphs to higher dimensions. Time permit- better than the root separation technique does. ting, I will also discuss recent work with Babson, Hoffman, and Paquette on the threshold for Kazhdan’s Property T Qi Cheng in the fundamental groups of random complexes, and how University of Oklahama [email protected] 48 AG11 Abstracts

this ties in with the expansion picture. MS15 Theta Functions, Curves and Monopoles Matthew Kahle Institute for Advanced Study The modern approach to integrability proceeds via a Rie- [email protected] mann surface, the spectral curve, and solutions of the in- tegrable system may be built from theta (and allied) func- Dominic Dotterrer tions on the curve. In many applications this curve is University of Toronto specified by transcendental constraints in terms of peri- ”dominic dotterrer” ¡[email protected]¿ ods and implementing these requires a good understanding of the curve, for example its homology and period matrix. Computational algebraic geometry facilitates this. In some MS14 cases, including the construction of magnetic monopoles, Alexander Duality for Functions physical symmetries are inherited by the spectral curve and there may be consequent simplifications of both the This work contributes to the point calculus of persistent function theory and transcendental constraints. We shall homology by extending Alexander duality to real-valued look at this interplay of ideas. functions. Given a perfect Morse function f : Sn+1 → [0, 1] and a decomposition Sn+1 = U ∪V such that M = U ∩V is Harry Braden an n-manifold, we prove elementary relationships between School of Mathematics the persistence diagrams of f restricted to U,toV ,andto University of Edinburgh M. hwb@staffmail.ed.ac.uk

Herbert Edelsbrunner Department of Computer Science MS15 Duke University Singularities of Theta Divisors [email protected] Abstract: We will survey the current state of knowledge Michael Kerber in complex algebraic geometry of the structure of the sin- Institute of Science and Technology (Austria) gularities of the theta divisor of a principally polarized [email protected] abelian variety. We will discuss the dimension of the sin- gular locus of the theta divisor, the multiplicity of points on it, the rank of the tangent cone, and other recent results MS14 and the remaining classical open questions. The Optimality of the Interleaving Distance on Multidimensional Persistence Modules Samuel Grushevsky Department of Mathematics Building on an idea of Chazal et al. [1], we define and Suny Stony Brook study the interleaving distance, a pseudometric on multidi- [email protected] mensional persistence modules generalizing the bottleneck distance. We present several results about the interleav- ing distance. Our main result is that when the underlying MS15 field is Q or Z/pZ for a prime p, the interleaving distance Some Computational Problems Using Riemann is (in a certain sense) optimal. This result is new even for Theta Functions in Sage 1-D persistence. As a byproduct of our results, we obtain a converse to the algebraic stability theorem of [1]. This Computational tools in algebraic geometry are useful for answers a question posed in that paper. generating new conjectures and providing a means for solv- ing problems in applications such as optimization and in- tegrable systems. Recent features in Sage for performing computations with Riemann theta functions and algebraic curves provide steps towards solving a large class of these problems. In this talk we will discuss current and future Reference developments in Sage for computational algebraic geom- etry and examine two applications in particular: gener- [1] Chazal, Cohen-Steiner, Glisse, Guibas, and Oudot. ating genus two and three solutions to the Kadomtsev– Proximity of persistence modules and their diagrams. Petviashvili equation and computing determinantal repre- sentations of homogenous plane curves. Michael Lesnick Stanford University Christopher Swierczewski [email protected] Applied Mathematics University of Washington [email protected] MS14 Algebraic Well Groups MS16 Given a mapping f : X → M from a topological space to Asymptotics of D-finite Sequences and the Cheetah an orientable manifold and U a subset of M,weidentify Algorithm a subgroup of the homology of f −1(U) that is stable to homotopic perturbations of the mapping. This talk tackles the following problem: input : recurrence of a D-finite sequence n α Amit Patel output: asymptotics an ∼ CA n with closed forms for INRIA C, A, α. Several experimental approaches are existing (e.g. [email protected] packages by Doron Zeilberger, Manuel Kauers, Marc Mez- AG11 Abstracts 49

zarobba & Bruno Salvy, see also recent work on G-functions Short Random Walks by Tanguy Rivoal & Stphane Fischler), but several prob- lems remain difficult (choosing the right branch, singular- We consider random walks in the plane which consist of n ity, etc) and to decide if C =0isconjecturedbysome steps of fixed length each taken into a uniformly random people to be undecidable. We use here a mixture of sym- direction. Our interest lies in the probability density func- bolic approach (local behaviour of solution of linear differ- tion of the distance travelled by such a walk. While Lord ential equations, as well described by works of Frobenius, Rayleigh’s limiting density is an excellent approximation Fuchs...) and numerical analysis (the so-called cheetah al- for moderately large n, we seek closed forms for the den- gorithm) to get C for a lot of problems coming from number sities in the case of small n. One of the goals of the talk theory, physics, probability theory, combinatorics. will be to show that in the cases n=3 and n=4 hypergeo- metric evaluations can be given. The basic ingredients are Cyril Banderier combinatorial properties of the associated even moments, Universit´e Paris-Nord computer algebra as well as a surprising modularity of the [email protected] densities. This is joint work with Jonathan M. Borwein, James Wan, and Wadim Zudilin.

MS16 Armin Straub Lattice Green’s Functions of the Tulane University Higher-Dimensional Face-Centered Cubic Lattices [email protected]

We study the lattice Green’s functions of the face-centered cubic lattice (fcc) in up to six dimensions. We give MS17 computer algebra proofs of results that were conjectured Spectrahedra and Determinantal Representations by Guttmann and Broadhurst for the four- and five- dimensional fcc lattices. Additionally we derive a differ- Spectrahedra are the feasible sets of semidefinite program- ential equation for the six-dimensional fcc lattice, a result ming. It is an important task to classify spectrahedra. The that was not believed to be achievable with current com- algebraic problem behind this question is to write polyno- puter hardware. We also present some conjectures concern- mials as determinants of linear matrix polynomials. I will ing the nature of the lattice Green’s function in arbitrary discuss recent results concerning this problem. In particu- dimensions. lar I will show how such determinantal representations are closely linked to sums of squares decompositions of an as- Christoph Koutschan sociated Hermite matrix. The results are joint work with RISC Linz Daniel Plaumann and Andreas Thom. [email protected] Tim Netzer University of Leipzig MS16 [email protected] The Method of Brackets: A Heuristic Method for Integration MS17 The method of brackets consists of a small number of rules Jacobian SDP Relaxation for Polynomial Opti- that provide an effective procedure for the evaluation of mization many integrals appearing in Feynman diagrams. Examples and heuristic rules will be presented. This is joint work Consider the optimization problem of minimizing a poly- with Ivan Gonzalez, Karen Kohl and Armin Straub. nomial function subject to polynomial equalities and/or in- equalities. Jacobian SDP Relaxation is new type semidef- VIctor Moll inite programming method that would solve the problem Tulane University exactly. Its basic idea is to add some new polynomial equal- [email protected] ities from the Jacobian of the polynomials, and then apply the hierarchy of Lasserre type sum of squares relaxations. The main result is that if we apply a sufficiently high re- MS16 laxation order, then the relaxation will be exact and global Recent Progress in Automation of Asymptotics optimal solutions for the original polynomial optimization could be obtained. It is well known that coefficient asymptotics depend mainly on the geometry of the algebraic surface Q=0. One of the Jiawang Nie challenges in moving from theorems that handle most cases University of California, San Diego in practice to automated asymptotics is to combinatorialize [email protected] the geometric data. This talk concerns some of the infras- tructure necessary to carry out the combinatorialization. A completely automated and rigorous algorithm for the MS17 smooth bivariate case is described in a forthcoming paper. Positivity of Piecewise Polynomials

Real algebraic geometry provides certificates for the pos- Robin Pemantle itivity of real polynomials under polynomial constraints University of Pennsylvannia by expressing them as a suitable combination of sums of [email protected] squares and the defining inequalitites. We show how Puti- nar’s theorem for strictly positive polynomials on compact sets can be applied in the case of strictly positive piecewise MS16 polynomials on a simplicial complex. In the 1-dimensional Hypergeometric Evaluations of the Densities of case, we improve this result to cover all non-negative piece- 50 AG11 Abstracts

wise polynomials and give explicit degree bounds. number of connected components of all realizable sign con- ditions of a family of real polynomials in Rk, restricted to Daniel Plaumann a real variety of dimension smaller than k. Unlike previous Universit¨at Konstanz results in this direction, our bound distinguishes the roles [email protected] of the degrees of the polynomials defining the variety from that of the other polynomials. This distinction appears to be crucial in certain applications – particularly, in certain MS17 problems of discrete geometry on bounding the number of Semidefinite Programs with Rank Constraints incidences. I will also point out connections to previous algorithmic work that led to the new proof. In 1953 Grothendieck worked on the theory of Banach spaces where he proved the fundamental theorem in Saugata Basu, Barone Sal the metric theory of tensor product, nowadays called Purdue University Grothendieck inequality. This inequality is a fundamen- [email protected], [email protected] tal and unifying tool in many areas of mathematics and computer science (functional analysis, combinatorics, ma- chine learning, system theory, quantum information the- MS18 ory, numerical linear algebra). With hindsight one can Fast Estimates of Hankel Matrix Condition Num- view Grothendiecks inequality and its proof (which is algo- bers and Numeric Sparse Interpolation rithmic) as the first randomized approximation algorithm based on semidefinite programming. In the talk I want to We investigate our early termination criterion for sparse extend Grothendiecks inequality so that it can be used to polynomial interpolation when substantial noise is present give approximation algorithms for finding ground states of in the values of the polynomial. Our criterion in the exact the n-vector model in statistical mechanics. Grothendiecks case uses Monte Carlo randomization which introduces a inequality itself together with the best known constant second source of error. We harness the Gohberg-Semencul (due to Krivine) gives a 0.56 approximation algorithm for formula for the inverse of a Hankel matrix to compute in the Ising model on the integer lattice. For the three- quadratic arithmetic time estimates for the structured con- dimensional Heisenberg model the algorithm achieves a ra- dition numbers of all the arising Hankel matrices, and ex- tio of 0.78. (based on joint work with Jop Briet, Fernando plain how false ill-conditionedness can arise from our ran- de Oliveira Filho) domizations. Finally, we demonstrate by experiments that our condition number estimates lead to a viable termina- Frank Vallentin tion criterion for polynomials with about 20 non-zero terms Delft University of Technology and of degree about 100, even in the presence of noise of [email protected] relative magnitude 0.00001. Joint work with Wen-shin Lee (Univ. Antwerp) and Zhengfeng Yang (ECNU, Shanghai) MS17 Erich Kaltofen Reformulating Polynomial Programs as Copositive North Carolina State University Programs Mathematics Department [email protected] We present a canonical convexification procedure which yields an equivalent formulation of polynomial program- ming problems as linear conic programs over the dual of MS18 the cone of copositive forms. This formulation is inspired On the Computing Time of the Continued Frac- by Burer’s dual copositive formulation of binary quadratic tions Method programming, which can be recovered as a special case of our procedure. The convexification procedure is based on The maximum computing time of the continued fractions new certificates of non-negativity for polynomials over the method for polynomial real root isolation is shown to be intersection of an unbounded closed domain and the zero at least quintic in the degree of the input polynomial. set of a given polynomial. This computing time is realized for an infinite sequence of polynomials of increasing degrees, each having the same Javier Pena coefficients. This is the first non-trivial lower bound for Carnegie Mellon University the maximum computing time of the continued fractions [email protected] method. Until recently such large computing times were not even observed. For each input polynomial under con- Juan C. Vera sideration, the continued fractions method recursively calls Tilburg School of Economics and Management itself on polynomials that become ever harder to process. Tilburg University In the analysis, the coefficients of those polynomials are [email protected] traced indirectly, by tracing images of the real and nonreal roots of the input polynomial under certain linear frac- Luis F. Zuluaga tional transformations. The recursion tree is completely Faculty of Business Administration described; its height is more than half the degree of the in- University of New Brunswick put polynomial. The quintic lower bound is proven using [email protected] a series of about eighty theorems and lemmas. The proof also uses well-known theorems that have not been used in computing time analyses before. Key words: Polynomial MS18 real root isolation, computing time lower bounds, symmet- Refined Bounds on Connected Components of Sign ric functions, subadditivity, Fibonacci numbers, Mignotte Conditions on a Variety polynomials, loxodromic transformations. This is a report In this talk I will talk about a recent result bounding the AG11 Abstracts 51

on recent joint work with G. E. Collins, Madison, WI. tion Networks

Werner Krandick For a reaction network, persistence is the property that no Drexel University species tend to extinction if all species are initially present. [email protected] We call vacuous persistence a stronger property: the same asymptotic feature when all species are implicitly present. We will present a necessary and sufficient condition for MS18 vacuous persistence in terms of reachability, describe two Connectivity in Semialgebraic Sets classes of vacuously persistent networks relevant to bio- chemistry, and relate our condition to known sufficient con- A semialgebraic set is a subset of real space defined by poly- ditions for persistence. nomial equations and inequalities. A semialgebraic set is a union of finitely many maximally connected components. Gilles Gnacadja In this talk, we consider the problem of deciding whether Amgen two given points in a semialgebraic set are connected, that [email protected] is, whether the two points lie in a same connected com- ponent. In particular, we consider the semialgebraic set defined by f not equal 0 where f is a given bivariate poly- MS19 nomial. The motivation comes from the observation that Identifiability of Species Phylogenies Under the many important/non-trivial problems in science and engi- Coalescent Model neering can be often reduced to that of connectivity. Due to it importance, there has been intense research effort on A phylogenetic tree is a graph that displays evolution- the problem. We will describe a method based on gradient ary relationships among a collection of organisms. The fields and provide a sketch of the proof of correctness based sequence data available for phylogenetic inference often in- Morse complex. clude samples taken from multiple genes within each organ- ism. This necessitates modeling of the evolutionary process James Rohal,HoonHong at two distinct scales. First, given an overall phylogeny rep- North Carolina State University resenting the actual evolutionary history of the species, in- [email protected], [email protected] dividual genes evolve their own histories, called gene trees. Then, along each gene tree, sequence data evolve, leading Mohab Safey El Din to the observed data that is used for inference. The coales- UPMC, Univ Paris 6 cent model provides the link between the evolution of the [email protected] gene trees given the species tree, and the evolution of the sequence data given the gene trees. Phylogenetic invariants have been proposed as a tool for inferring phylogenies using MS18 data from a single gene, and their mathematical properties A Divide-and-conquer Method for Computing have been widely studied. In this talk, we consider the Cylindrical Algebraic Decomposition development of methods based on phylogenetic invariants developed specifically for species trees, as opposed to gene Cylindrical algebraic formulas (CAF) provide an explicit trees. In particular, we use methods from algebraic statis- representation of semialgebraic sets as finite unions of tics to establish identifiability of the species phylogeny. cylindrically arranged disjoint cells bounded by graphs of algebraic functions. For a quantified system of polynomial Laura Kubatko equations and inequalities, cylindrical algebraic decompo- The Ohio State University sition computes a CAF representation of its solution set. If [email protected] the system is given as a Boolean combination of formulas, its cylindrical algebraic decomposition can be computed ei- Julia Chifman ther directly, using the cylindrical algebraic decomposition Mathematical Biosciences Institute (CAD) algorithm, or in two steps, by first finding a CAF The Ohio State University representation of the solution set of each formula and then [email protected] computing the required Boolean combination of the CAF representations using the CAFCombine algorithm of [1]. In my talk I will discuss the problem of deciding automati- MS19 cally which of the two methods to use and how to subdivide Algebraic Results on the Multispecies Coalescent the input system into a Boolean combination of subformu- Model las for the two-step method. I will describe how graph theory methods can be used for this purpose. [1] Compu- Phylogenetic models relate a gene tree to sequence data tation with Semialgebraic Sets Represented by Cylindrical arising on it. To study evolutionary relationships between Algebraic Formulas, Proceedings of the International Sym- species, one also needs the coalescent to relate a species posium on Symbolic and Algebraic Computation, ISSAC tree to the gene trees arising on it. Without strong as- 2010, 61-68, Munich, Germany, July 25-28, 2010. ACM, sumptions, the ’time’ units in these models are unrelated, Stephen M. Watt, ed. so the distribution of topological gene trees is of interest. In this setting, we describe several recent results on iden- Adam Strzebonski tifiability of species trees which have used a viewpoint of Wolfram Research Inc. algebraic geometry. [email protected] Elizabeth S. Allman University of Alaska Fairbanks MS19 [email protected] Reachability Approach to the Persistence of Reac- James Degnan 52 AG11 Abstracts

University of Canterbury MS20 [email protected] Applications of Numerical Algebraic Geometry to Pure Mathematics John A. Rhodes University of Alaska Fairbanks This talk will survey known applications of the homotopy [email protected] continuation methods and numerical algebraic geometry to the problems arising in several areas of pure mathematics. I will showcase our joint work with Frank Sottile on comput- MS19 ing Galois groups of Schubert problems via homotopy con- New Sufficient Conditions for Ruling out Multiple tinuation. Rigorous certification of results obtained with Steady States in Chemical Reaction Systems approximate numerical algorithms will be discussed as well.

In a chemical reaction system, the concentrations of chem- ical species evolve in time, governed by the polynomial Anton Leykin differential equations of mass-action kinetics. This talk School of Mathematics concerns the problem of determining whether a chemical Georgia Institute of Technology reaction network admits multiple steady states. In gen- [email protected] eral, establishing the existence of (multiple) steady states is challenging, as it requires the solution of a large system of polynomials with unknown coefficients. However, for MS20 networks that have special structure, various easy criteria Symbolic-Numeric Algorithms for the Computa- can be applied. This talk will highlight a new criterion tion of Discrete Invariants for ruling out multiple steady states based on the Jacobian Criterion of Craciun and Feinberg, and will present work This talk will focus on the use of symbolic-numeric methods on a classification of small multistationary chemical reac- to compute discrete invariants of varieties, schemes, and tion networks. Relevant examples from biochemistry will sheaves. be given. Chris Peterson Anne Shiu, Badal Joshi Colorado State University Duke University [email protected] [email protected], [email protected] MS20 MS20 Applying Littlewood-Richardson Homotopies to Certification and Complexity in Solving Nonlinear Schubert Problems Systems of Equations. Based on Ravi Vakil’s geometric proof of the Littlewood- In this talk I will present some of the recent progress in the Richardson rule, we developed a numerical homotopy problem of solving systems of equations with two require- continuation algorithm for finding all solutions to Schu- ments: the solutions must be certified, and the complexity bert problems on Grassmannians. For generic Schu- of the algorithms must be understood. bert problems the number of paths tracked is optimal. The Littlewood-Richardson homotopy algorithm is imple- Carlos Beltran mented using the path trackers of the software package Departamento de Matematicas PHCpack. Universidad de Cantabria [email protected] Jan Verschelde Department of Mathematics, Statistics and Computer Science MS20 University of Illinois at Chicago Numerical Algorithms for Dual Bases of Positive- [email protected] Dimensional Ideals Frank Sottile An ideal of a local polynomial ring can be described by Texas A&M University calculating a standard basis with respect to a local mono- [email protected] mial ordering. However if we are only allowed approxi- mate numerical computations, this process is not numeri- Ravi Vakil cally stable. On the other hand we can describe the ideal Department of Mathematics numerically by finding the space of dual functionals that Stanford University annihilate it. There are several known algorithms for find- [email protected] ing the truncated dual up to any specified degree, which is useful for zero-dimensional ideals. I present a stopping criterion for positive-dimensional cases based on homoge- MS21 nization that guarantees all generators of the initial mono- Search Bounds with Respect to Height in Linear mial ideal are found. This has applications for calculating and Quadratic Spaces Hilbert functions. In this talk, I will discuss a variety of results on existence Robert Krone of points and subspaces of bounded height, possibly sat- School of Mathematics isfying some additional algebraic conditions, in linear and Georgia Institute of Technology quadratic spaces over global fields and rings. These results [email protected] represent some of the recent developments on extensions and generalizations of such classical Diophantine theorems AG11 Abstracts 53

as Siegel’s lemma and Cassels’ theorem on small zeros of MS22 quadratic forms. Hyperdeterminantal Varieties from Tensor Com- plexes Lenny Fukshansky Claremont McKenna College I will discuss a family of hyperdeterminantal varieties [email protected] which include varieties cut out by maximal minors of a ma- trix as well as hypersurfaces given by hyperdeterminants of the boundary format. These varieties arise from an explicit MS21 construction of certain free resolutions, called tensor com- A Deterministic Polynomial Time Algorithm for plexes. Finding Roots of Polynomials Over Complex Num- bers Christine Berkesch Purdue University We present an iterative algorithm that has global conver- [email protected] gence, that is, for any univariate polynomial of degree n over complex numbers and any initial point, the algorithm Daniel Erman will find a root of the polynomial, and the number of com- University of Michigan plex operations used is bounded by a polynomial in n and [email protected] the number of digits for desired accuracy. Yong Feng Manoj Kummini Chengdu Institute of Computer Applications Chennai Mathematical Institute Chinese Academy of Sciences [email protected] [email protected] Steven V Sam Shuhong Gao MIT Clemson University [email protected] [email protected] MS22 MS21 Markov Bases and Beyond Higher Mahler Measure and Lehmer’s Question Describing Markov bases of toric statistical models has The k-higher Mahler measure of a nonzero polynomial P been recognized as a rich source of interesting combina- is the integral of logk |P | on the unit circle. I will discuss torial problems. We will give an overview over past and Lehmer’s question for k>1 and show some interesting ongoing research in this area. formulae for 2- and 3-higher Mahler measure of cyclotomic polynomials. Thomas Kahle EPDI / Institut Mittag-Leffler Matilde Lalin [email protected] University of Montreal [email protected] MS22 Kaneenika Sinha Testing Chromosome Proximity Hypothesis Using Indian Institute of Science Education and Research, Log-linear Models Kolkata Studying chromosome proximity and its effect on chromo- [email protected] some exchange plays important role in understanding of development of certain types of cancer. The proximity- MS21 effect hypothesis states that number of exchanges is larger between chromosomes that are located close to each other. Title Not Available at Time of Publication To evaluate previously published data from numerous ex- Abstract not available at time of publication. periments involving 22 human autosomes of respectively three data sets of lymphocyte cells that have been subject Korben Rusek to ionizing radiation we used tools of algebraic statistics Texas A&M University together with asymptotic properties of log linear models. [email protected] Specifically we used Markov basis of a log-linear model which does not have proximity-effect parameters, to sample from a large space of tables that have the same minimal suf- MS21 ficient statistics as the experiment data table. The Markov The Subset Sum Problem Chain Monte Carlo approach did not provide sufficient ev- idence to the reject null hypothesis of no proximity-effect. This will be an introduction to some new results and prob- We considered a modified model, were proximity effect of lems on the subset sum problem, particularly those with individual pairs was included as an additional parameter. additional algebraic structure. After using asymptotic tests we could not reject certain modified models. Daqing Wan University of California, Irvine Tatsiana Maskalevich [email protected] San Francisco State University [email protected] 54 AG11 Abstracts

MS22 alised through the computation of the ranks and of the in- Primary Decomposition of a Class of Conditional teger Smith normal forms of the boundary matrices of the Independence Ideals groups. We present novel methods for the efficient compu- tation of these routines, specialized for the large and sparse We describe (in a combinatorial way) the primary decom- matrices arising in these applications. position of a class of ideals arising in the context of condi- tional independence models. The ideals we consider gener- Jean-Guillaume L. Dumas alize the ideals considered by Fink (2010) in a way distinct MNC-IMAG, UJF Grenoble from that of Herzog, Hibi, Hreinsdottir, Kahle, and Rauh [email protected] (2010). We give a combinatorial description of the the minimal components, along with the corresponding prime ideals (they turn out to be the same, although there are MS23 embedded primes) of these conditional independence ide- Integral Cohomology of Some Arithmetic Groups als. Along the way we introduce an equivalence relation G and recover some other interesting algebra and geometry The cohomology of a discrete group is the cohomology of the quotient space X/G where X is any contractible space results as a consequence of the development of the proof of G our main result. admitting a free action of . This talk will describe an im- plemented algorithm for computing the integral cohomol- Amelia Taylor ogy of groups such as PSL4(Z), Sp4(Z)andPSL2(O)for Colorado College various rings O of quadratic integers. The algorithm can [email protected] also compute the cohomology of finite index subgroups of these three groups. The talk will discuss bottlenecks that Irena Swanson arise when the homology degree or the subgroup index is Reed College large. [email protected] Graham Ellis National University of Ireland, Galway MS22 [email protected] Pairwise Ranking: Choice of Method Can Produce Arbitrarily Different Order MS23 We showed that for any two of the three popular methods Symbolic-numeric Linear Algebra and Lattice Ba- for ranking by pairwise comparison (HodgeRank, Tropical sis Reduction and Principal Eigenvector), and for any pair of rankings of LLL basis reduction is an important algorithm in computer at least four items, there exists a comparison matrix for the science and mathematics making worthwile efficiency im- items such that the rankings found by the two methods are provements. Currently fastest reduction algorithms are us- the prescribed ones. We discuss the implications, study the ing both symbolic and numeric linear algebra techniques. geometry and combinatorics, and state some open prob- Revisiting and improving classical tools from the field of lems. numerical analysis, we propose new reduction algorithms Ngoc M. Tran and certificate ([X.-W. Chang, D. Stehl´e, G. Villard][A. UC Berkeley Novocin, D. Stehl´e, G. Villard]). We describe these devel- [email protected] opments, we show in particular how approximate compu- tations may be introduced at various levels in the overall reduction process. MS23 Gilles Villard The Berlekamp/Massey Algorithm and Counting ´ Singular Hankel Matrices over a Finite Field CNRS, Ecole Normale Sup´erieuredeLyon,U.Lyon [email protected] We derive a formula for the number of singular n × n Han- kel matrices over a finite field of q elements by observing MS23 the Berlekamp/Massey algorithm run on the entries, al- lowing some entries above or on the anti-diagonal to be Berlekamp/Massey: Implementations in LinBox fixed.Wealsoderiveaformulaforthenumberofn × n Hankel matrices whose first r leading principal submatri- The Berlekamp/Massey algorithm and its variants com- ces are non-singular and the rest are singular. This result prise a key component to many exact linear algebra meth- generalizes to block-Hankel matrices as well. ods. Often the computation of the minimal generator of a linearly generated sequence is a vital sub-procedure in Matthew Comer,ErichKaltofen methods like linear solving, smith forms and rank com- North Carolina State University putations. A discussion of recent and current work on the Mathematics Department implementation of this algorithm within the LinBox library [email protected], [email protected] will be given. The algorithm presents interesting choices for the implementation and data structures required during its computations. The different design choices, applications MS23 and computational results will be presented. Exact Linear Algebra and Algebraic Topology George Yuhasz We present exact linear algebra methods for the compu- Morehouse College tation of homology groups of simplical complexes and in Division of Science & Mathematics algebraic K-theory for computation of the cohomology of [email protected] the linear group GL7(Z). These compputations are re- AG11 Abstracts 55

MS24 University of Oslo Analysis Aware Representations, Parameteriza- Department of Informatics and CMA tions, and Models tom@ifi.uio.no Isogeometric Analysis (IA) has been proposed as a method- ology for bridging the gap between Computer Aided De- MS24 sign (CAD) and Finite Element Analysis (FEA). In order Splines on Rectangular Subdivisions: Dimension to support design and full 3D IA, new ab initio design and Basis methods must create suitable representations and approxi- mation techniques must include parameterization method- Standard parametrisations of surfaces in CAGD are based ologies for volumes. This presentation discusses some of on tensor product bspline functions, defined from a grid the challenges in moving from current representations and of nodes over a rectangular domain. This representation datafitting techniques towards this goal and demonstrates is not well-adapted to local refinement, which is particu- initial results and analyses. larly important in applications such as surface reconstruc- tion or isogeometric analysis. To extend the representa- Elaine Cohen tion of tensor-product splines while providing local refine- University of Utah ment capabilities, some recent works have considered T- [email protected] subdivisions. These are partitions of an axis-aligned box into smaller axis-aligned boxes. In this talk, we will an- alyze the space of bivariate functions that are piecewise MS24 polynomial of bidegree (m, m)andclassC(r, r)oversuch Splines and Isogeometric Analysis (IGA) a planar T-subdivision. We shall show how algebraic tech- niques yield a good control of its resolution, give a new Finite Element Analysis (FEA) was already established in formula for its dimension and explicit bases for small de- 1970 when Computer Aided Geometric Design (CAGD) gree and regularity. community was established. Until the introduction of IGA by T. J. R. Hughes in 2005 little cooperation took place be- Bernard Mourrain tween the FEA and CAGD communities. IGA has the po- INRIA Sophia Antipolis tential to drastically improve the interoperability of CAGD [email protected] and FEA and the quality FEA. To achieve this we need im- proved technology and increased research into spline tech- nology, FEA and visualization. MS25 Reeb Graph Reconstruction and Zigzag Persistent Tor Dokken Homology SINTEF ICT, Department of Applied Mathematics Oslo, Norway Reeb graph reconstruction is an important problem in [email protected] topological data analysis. We study the error in recon- structing the Reeb graph of a point cloud sampled from a manifold, using the MAPPER algorithm. Using the lan- MS24 guage of zigzag persistence, we are able to succinctly cap- Characterization of Hierarchical Spline Spaces ture the difference between the reconstructed and actual Reeb graphs. A worst case bound on the error is derived The only requirement which characterizes the hierarchical in terms of the sampling density. We demonstrate practical model is a refinable nature of the underlying basis func- applications such as dataset classification. tions defined on nested approximation spaces. The talk will present recents results on the construction of normal- Aravindakshan Babu ized hierarchical spline bases and their properties. ICME, Stanford University PhD Student Carlotta Giannelli [email protected] Johannes Kepler University Linz Austria Daniel Muellner [email protected] Stanford Math Department [email protected] Bert J¨uttler Johannes Kepler University Linz Gunnar E. Carlsson [email protected] Stanford University [email protected] MS24 Locally Refined B-splines MS25 We will address local refinement of tensor product B- PHIsoMap: Estimating Intrinsic Distance Using splines specified as a sequence of inserted line segments Persistent Homology parallel to the knot lines. We obtain a quadrilateral grid Many dimension reduction techniques (including IsoMap) with T-junctions in the parameter domain, and a collec- start by constructing a graph on the data in order to es- tion of tensor product B-splines on this mesh here named timate the geodesic distance from the underlying mani- an LR-mesh. The approach applies equally well in dimen- fold. We present a algorithm that uses persistent homology sions higher than two. Moreover, in the two dimensional (in dimensions one and zero) to construct an -proximity case this collection of B-splines spans the full spline space graph; our initial experiments show that this graph pro- on the LR-mesh. duces strikingly good estimates of the underlying geodesic Tom Lyche 56 AG11 Abstracts

distance. Universidad de Buenos Aires [email protected] Paul Bendich Duke University Luis Tabera [email protected] Universidad de Cantabria [email protected] Jacob Harer North Carolina State University [email protected] MS26 Tropical Reparameterisations

MS25 It is well known that the tropicalisation of a polynomial Covering and Packing by Diagonal Distortion map φ : Am → An maps (R ∪{∞})m into the tropical- isation of the image X of φ. Typically this tropical map n We address the problem of covering R with congruent is not surjective. To remedy this, one may precompose φ balls, while minimizing the number of balls that contain with other polynomial maps and then tropicalise the re- an average point. Considering the 1-parameter family of sult. Ideally, one would like to prove that finitely many lattices defined by stretching or compressing the integer such reparameterisations suffice to cover the entire trop- grid in diagonal direction, we give a closed formula for the icalisation of X. I will give an overview of when this is covering density that depends on the distortion parameter. known to be the case. We observe that our family contains the thinnest lattice coverings in dimensions 2 to 5. We also consider the prob- Jan Draisma lem of packing congruent balls in Rn, for which we give a T.U.Eindhoven, The Netherlands closed formula for the packing density as well. Again we Department of Mathematics observe that our family contains optimal configurations, [email protected] this time densest packings in dimensions 2 and 3.

Herbert Edelsbrunner, Michael Kerber MS26 Institute of Science and Technology (Austria) Computing Tropical Resultants [email protected], [email protected] Tropical resultant varieties arise when fixing a set of New- ton polytopes and asking for which set of coefficients their MS26 tropical hypersurfaces have a common intersection. In this Implicitization of Surfaces via Geometric Tropical- talk we compare various methods for computing tropical ization resultants. Some of the algorithms involve traversing sub- fans of secondary fans while other works by reconstruction In this talk we discuss recent developments in tropical tropical hypersurfaces from projections. Using these tech- methods for implicitization of surfaces. This study was pi- niques we also get a new method for computing mixed fiber oneered in the generic case by work of Sturmfels, Tevelev polytopes. and Yu, and is based on the theory of geometric tropi- calization, developed by Hacking, Keel and Tevelev, The Anders Jensen latter hinges on computing the tropicalization of subvari- Universit¨at des Saarlandes eties of tori by analyzing the combinatorics of their bound- [email protected] ary in a suitable (tropical) compactification. We enhance this theory by providing a formula for computing weights Josephine Yu on tropical varieties, a key tool for tropical implicitization. Georgia Institute of Technology Finally, we address the question of tropical implicitization [email protected] for non-generic surfaces and illustrate our techniques with several numerical examples in 3-space. MS26 MariaAngelicaCueto An Effective Computation of Tropical Linear University of California, Berkeley Spaces, with an Application to A-discriminants [email protected] We develop and implement a very fast algorithm for com- puting tropical linear spaces, making some computational MS26 applications of tropical geometry now viable. For this pur- Singular Tropical Hypersurfaces pose, we study a fan structure on the Bergman fan of a matroid which slightly refines its nested set structure, and The concept of a singular point of a tropical variety is not which is amenable to our computational purposes. Some well established yet. A natural definition is to declare a software implementations will be shown, including an ap- point q in a tropical variety V singular if q is the image plication to the computation of A-discriminants. under the valuation of a singular point of a classical alge- braic variety, defined over the field of Puiseux series and Felipe Rincon with tropicalization V. We present a purely tropical char- UC Berkeley acterization of a singularity of a tropical hypersurface of [email protected] arbitrary dimension (with fixed support) in terms of tropi- cal Euler derivatives. We also describe in this setting non- transversal intersections of planar tropical curves. We show MS27 an algorithm to compute all singularities and we relate our The Moment Problem for Continuous Linear Func- approach to the known tropicalization of the discriminant. Alicia Dickenstein AG11 Abstracts 57

tionals variate rational functions.

Let V be the countable dimensional real polynomial alge- Nikolay Gravin bra. Let τø be a locally convex topology on V .LetK be a NTU Singapore n M M closed subset of R ,andlet := {g1,···,gs} be a finitely [email protected] generated quadratic module in V . We investigate the fol- lowing question: WhenistheconePsd(K) (of polynomials Jean B. Lasserre nonnegative on K) included in the τ closure of M?.We LAAS CNRS and Institute of Mathematics give an interpretation of this inclusion with respect to rep- [email protected] resenting continuous linear functionals by measures. We discuss several examples; we compute the closure of the Dmitrii Pasechnik, Sinai Robins cone of sums of squares with respect to weighted norm- p NTU Singapore topologies. We show that this closure coincides with [email protected], [email protected] the cone Psd(K) for certain convex compact K.Weuse these results to generalize Berg et al. work on exponentially bounded moment sequences. Joint work with M. Ghasemi MS27 and E. Samei Using Symmetries in Polynomial Optimization

Salma Kuhlmann Solving polynomial optimization problems is known to be University of Konstanz , Germany a hard task in general. The semidefinite methods which [email protected] envolved within recent years give possibilities to calculate these hard problems. However, as the sizes of the resulting MS27 SDPs grow fast, it is helpful to exploit some structure of the problem. We present approaches for exploiting symmetries A New Look at Nonnegativity on Closed Sets and within this framework. Our special focus is on problems Polynomial Optimization which are invariant by the symmetric group. C We provide a new characterization of the convex cone K Cordian B. Riener of degree-d polynomials that are nonnegative on a closed K ⊆ Rn University of Konstanz, Germany set , provided that one knows all moments of a [email protected] finite Borel measure whose support is exactly K.From this we also provide a hierarchy of outer approximations Thorsten Theobald (Pj )ofCK where each each convex cone Pj is a spectra- hedron described only in terms of the coefficients of the J.W. Goethe-Universit¨at polynomials, i.e., with no lifting. As a by-product, with Frankfurt am Main, Germany n [email protected] K = R+, we obtain converging outer approximations of the cone of copositive matrices. Finally, and also as a by- product, when K is a simple set like e.g., a box, an ellip- Jean B. Lasserre soid, the hypercube {−1, 1}n, the positive orthant, or the LAAS CNRS and Institute of Mathematics whole space Rn, etc., we provide a monotone nonincreasing [email protected] sequence of upper bounds on the global minimum of any polynomial f on K. These upper bounds complement the sequence of lower bounds obtained from the moment-sos MS27 relaxations. Successive Convex Relaxation Methods for Opti- mization over Semi-Algebraic Sets Jean B. Lasserre LAAS CNRS and Institute of Mathematics I will discuss various techniques (new and old) used in con- [email protected] vergence proofs for successive convex relaxation methods applied to optimization problems expressed as maximiza- tion of a linear function over a finite set of polynomial in- MS27 equality constraints. I will focus on convergence rates and On the Inverse Moment Problem for Polytopes specially structured non-convex optimization problems.

Reconstructing a measure from its moments is known as an Levent Tuncel inverse moment problem, and has important applications University of Waterloo for data processing, e.g. in geophysics, image recognition, Dept. of Combinatorics and Optimization etc. The problem is well-studied in dimension 2, where [email protected] tools of complex analysis are applicable. The basic in the approach there are procedures for expressing coordinates of the vertices V (P ) of a polygon P , given moments of MS28 a measure supported on P . Higher-dimensional problems Toward a Salmon Conjecture are traditionally treated in an approximate way, by working with 2-D slices, and only lead to approximate solutions. We Methods from numerical algebraic geometry are applied in present efficient procedures for reconstructing polynomial combination with techniques from classical representation measures dμ,supportedonconvexd-dimensional polytopes theory to show that the variety of 3-by-3-by-4 tensors of a d border rank 4 is cut out by polynomials of degree 6 and P , given finitely many moments X dμ, a ∈ Z+.Our P 9. Combined with results of Landsberg and Manivel, this procedures recover the vertices of P and μ,andcanbeused furnishes a computational solution of an open problem in in exact (rational inputs, exact rational outputs) as well algebraic statistics, namely, the set-theoretic version of All- as in non-exact setting (inputs with errors, approximate man’s salmon conjecture for 4-by-4-by-4 tensors of border outputs). We use tools from toric geometry, and symbolic- rank 4. A proof without numerical computation was given numeric computations with representations of V (P ) as uni- 58 AG11 Abstracts

recently by Friedland and Gross. Jeremy Sumner University of Tasmania Luke Oeding [email protected] University of California Berkeley Universit`a degli Studi di Firenze [email protected] MS28 Ranks and Generalized Ranks

MS28 The Waring rank of a polynomial of degree d is the least Geometry of Tensors and Numerical Decomposi- number of terms in an expression for the polynomial as a tion sum of dth powers. The problem of finding the rank of a given polynomial and studying rank in general is related This talk will consider tensor parameter spaces from a ge- to secant varieties, and there are applications throughout ometric point of view. The study of tensors from the van- engineering and the sciences, such as in signal processing tage point of algebraic geometry leads to symbolic and and computational complexity; and of course, it has been symbolic-numeric algorithms related to tensor decompo- a central problem of classical algebraic geometry. For ex- sitions, tensor approximations, and generic rank. Both a ample, J.J. Sylvester gave a lower bound for rank in terms geometric description and basic implementations of several of catalecticant matrices in the mid-19th century. While of these algorithms will be presented. catalecticant matrices and varieties have become objects of study in their own right, there has been relatively little Hirotachi Abo progress in the last 160 years on the problem of bound- University of Idaho ing or determining the rank of a given (not general) poly- [email protected] nomial. I will describe joint work with J.M. Landsberg which gives a new, elementary improvement to the catalec- Giorgio Ottaviani ticant lower bound for the rank of a polynomial, in terms University of Firenze of the geometry of the polynomial. I will also describe new [email protected]fi.it work which generalizes this improved bound to multiho- mogeneous polynomials, corresponding to secant varieties Chris Peterson of Segre-Veronese varieties. A further generalization to the Colorado State University recently described Young flattenings (treating secant vari- [email protected] eties of arbitrary varieties) is work in progress.

J.M. Landsberg MS28 Texas A&M Some non-minimal Canonical Representations of [email protected] Forms as a Sum of Powers of Linear Forms Zach Teitler We will give some illustrations of the title phenomenon, Boise State University especially Reichstein’s beautiful “completion of the cube” [email protected] from the late 1980s.

Bruce Reznick MS29 University of Illinois, Urbana-Champaign A Practical, Principled Approach to Fast Simplifi- [email protected] cation of Tarski Formulas This talk focuses on the problem of false simplification of MS28 Tarski formulas, which means finding meaningful simplifi- Stochastic Models, Tensor Rank, and Inequalities cations that occur in practice within a period of time that × × is small relative to most symbolic computations involving The hyperdeterminant Δ on 2 2 2 tensors appears in polynomial systems over the reals. An effective algorithm a semialgebraic characterization of the probability distri- for fast simplification can be used to improve other algo- butions arising from binary models of evolution on phylo- rithms, by pre-processing input, post-processing output, genetic trees, through its connection to tensor rank. To and, perhaps most importantly, by reducing the size of in- understand inequality constraints for non-binary models, termediate results. The talk will describe extensions of we construct an explicit representation space of functions k×k×k k work I have previously published on the subject that, ul- which, for tensors of border rank , is analogous to timately, provide a fast simplification procedure that is ef- Δ. In particular, their non-vanishing on a tensor of border k k fective at improving the performance and quality of results rank indicates it has rank . for several other algorithms. The approach makes some in- Elizabeth S. Allman teresting connections between a sub-class of simplifications University of Alaska Fairbanks on Tarski formulas and coding theory; a connection which [email protected] is exploited in complexity analyses. Christopher Brown Peter D. Jarvis United States Naval Academy University of Tasmania [email protected] [email protected]

John A. Rhodes MS29 University of Alaska Fairbanks Certified Global Optimization with Exact Sum-Of- [email protected] AG11 Abstracts 59

Squares [email protected]

Given a multivariate polynomial f with real or rational co- efficients, we focus on computing and certifying its lower MS29 bounds or its global infimum. The global optimization Computing Rational Points in Convex Semi- problem is NP-hard. A well known technique is to use the Algebraic Sets sum-of-squares(SOS) and semidefinite programming(SDP) to relax it to get lower bounds. We implement the SDP Let P = {h1,...,hs}⊂Z[Y1,...,Yk], D ≥ deg(hi)for interior-point method in Maple and get high precision SDP 1 ≤ i ≤ s, σ bounding the bit length of the coefficients results which can not be obtained by the fixed precision of the hi’s, and Φ be a quantifier-free P-formula defining SDP solvers in matlab. Then we certify the lower bounds a convex semi-algebraic set. We design an algorithm re- by making projections to get the exact SOS representa- turning a rational point in S if and only if S∩Qk = ∅. 3 tions. To compute the global infimum, we use the re- It requires σO(1)DO(k ) bit operations. If a rational point sults from the computation of sample points in the smooth is outputted its coordinates have bit length dominated by real algebraic set, then relax the problem by computing ( 3) σDO k . Using this result, we obtain a procedure decid- its infima over some finite semialgebraic sets which can ing if a polynomial f ∈ Z[X1,...,X ]isasumofsquares be reduced to SDPs. From Artin’s affirmative solution of n of polynomials in Q[X1,...,X ]. Denote by d the degree Hilbert’s 17th problem, we know that a nonnegative poly- n of f, τ the maximum bit length of the coefficients in f, nomial can be written as a sums of squares of rational func- D n+d k ≤ D D − n+2d = n and ( +1) n . This procedure tions. Given a fixed degree of the denominators, we give 3 the certification if the polynomial can not be represented as requires τ O(1)DO(k ) bit operations and the coefficients of sums of squares of rational functions with denominators of the outputted polynomials have bit length dominated by 3 the fixed degree. This is a joint work with Erich Kaltofen, τDO(k ). This is joint work with Mohab Safey El Din Mohab Safey El Din, Zhengfeng Yang and Lihong Zhi. Lihong Zhi Feng Guo Academia Sinica North Carolina State University [email protected] [email protected]

MS30 MS29 Using Fiber Products and Numerical Algebraic Ge- Multivariate Root Bound ometry to Find Exceptional Mechanisms Given a system F of n polynomials in n variables, a bound An exceptional mechanism is a mechanism having special B is obtained such that if F(x) = 0, then x1,...,xn ¡ B, for all movement as compared to other mechanisms in the same real x. A root bound gives a useful starting point for var- class. These correspond to points in the parameter space of ious domain decomposition methods used to approximate a given class, which have fiber dimension greater than the the solution of systems of polynomials. Such a bound can generic fiber dimension. Such points are usually obscured also be used to estimate the complexity of root finding al- by their containment in an irreducible component. We use gorithms. We present a bound which is significantly tighter numerical algebraic geometry, starting with the fiber prod- than those previously studied. The bound is obtained by uct method of Sommese and Wampler, to identify these integrating results from Elimination theory and Eigenvalue points, thus finding exceptional mechanisms. inclosure theory. Daniel J. Bates Aaron Herman,HoonHong Colorado State University North Carolina State University Department of Mathematics [email protected], [email protected] [email protected]

MS29 Eric Hanson Department of Mathematics On Hilbert’s 16th Problem Colorado State University At the start of the 20th century David Hilbert proposed [email protected] 23 problems to be solved over the coming century. The first portion of his 16th problem asks about the arrange- Jonathan Hauenstein ment of the connected components of nonsingular algebraic Texas A&M University curves in the real projective plane. Despite progress over [email protected] the last century, the problem has remained unsolved in a general sense. Efforts to resolve this problem have previ- Charles Wampler ously focused on constructions of maximal curves by hand General Motors Research Laboratories, Enterprise through slight permutations of singular curves. Approach- Systems Lab ing the problem by hand unfortunately tends to produce 30500 Mound Road, Warren, MI 48090-9055, USA polynomials with large cofficients and curves which are vi- [email protected] sually poor when plotted. Instead this talk will approach the problem from a computerized perspective after using grid processing to compute topologies for many generated MS30 polynomials. The statistical distribution of curve topolo- Beyond Parametrization of Algebraic Varieties by gies by degree and coefficient range will be discussed as well Means of Global Optimization as ’nice’ witness polynomials for known topological cases. This research work presents a new global optimization al- Jason Yellick gorithm, Evo-Runge-Kutta, to tackle the open problem of North Carolina State University 60 AG11 Abstracts

minimizing a positive function on an algebraic variety. We Notre Dame, IN 46556-4618, USA illustrate the application of Evo-Runge-Kutta, a two-phase [email protected] evolutionary optimization algorithm using a multicriteria approach, for computing explicit and implicit Runge-Kutta Wenrui Hao methods. In particular, we present theoretical motivations Dept. of Applied and Comp. Mathematics and Statistics for the application of the class of evolutionary algorithms: University of Notre Dame the hardness to find 1) the dimension of the variety and [email protected] 2) feasible solutions for the parametrization. The mapping between algebraic geometry and global optimization is di- Jonathan Hauenstein rect. In order to design methods having specific properties, Texas A&M University we expect that many open problems in algebraic geometry [email protected] will be modeled as constrained multi-objective optimiza- tion problems. Bei Hu Ivan Martino Dept. of Applied & Computational Mathematics & Stockholms Universitet Statistics [email protected] University of Notre Dame [email protected] Giuseppe Nicosia University of Catania MS31 Department of Mathematics and Computer Science [email protected] Ideals of Graph Homomorphisms and Graph Col- orings

MS30 We show how ideals of graph homomorphisms in algbraic statistics can be employed to study graph colorings. Solving Principal Agent Problems by Polynomial Programming Alex Engstrom UC Berkeley We present a new way to solve principal agent problems by [email protected] polynomial programming techniques. We study the case where the agent’s actions are unobservable by the prin- cipal but the outcomes are. We assume that the agent’s MS31 actions lie in an interval and the space of outcomes is a The Binomial Ideal of the Intersection Axiom for finite set. Furthermore the agent’s expected utility is a Conditional Probabilities rational function in his actions. The resulting problem is a bilevel optimization problem with the principal’s prob- The binomial ideal associated with the intersection axiom lem as the upper and the agent’s problem as the lower level. of conditional probability is shown to be radical and is The key idea is to find an exact reformulation of the agent’s expressed as intersection of toric prime ideals. This resolves problem as a semidefinite optimization problem. Since this a conjecture in algebraic statistics due to Cartwright and is a convex optimization problem, we then have necessary Engstrom. and sufficient global optimality conditions for the agent’s problem. The reformulation can be done by using classical Alex Fink results from real algebraic geometry linking positive poly- University of California at Berkeley nomials and semidefinite matrices. We obtain a nonlinear arfi[email protected] program. If all functions are rational functions, we then can solve it to global optimality. MS31 Philipp Renner Commutative Algebra of Statistical Ranking Mod- Department of Business Administration els Universitaet Zuerich [email protected] A model for statistical ranking is a family of probability distributions whose states are orderings of a fixed finite set Karl Schmedders of items. We represent the orderings as maximal chains University of Zuerich in a graded poset. The most widely used ranking mod- [email protected] els are parameterized by rational function in the model parameters, so they define algebraic varieties. We study these varieties from the perspective of combinatorial com- MS30 mutative algebra. One of our models, the Plackett-Luce Tumor Growth Models and Numerical Algebraic model, is non-toric. Five others are toric.For these models Geometry we examine the toric algebra, its lattice polytope, and its Markov basis. Numerical Algebraic Geometry is being used to explore free boundary problems arising from tumor growth models. Volkmar Welker These methods are used to compute solutions, check their Philipps-Universit¨at Marburg stability, and track the solutions through bifurcations as Fachbereich Mathematik und Informatik critical parmaters change. The polynomial systems, which [email protected] involve thousands of variables, pose interesting challenges. Bernd Sturmfels UC Berkeley Andrew Sommese [email protected] Department of Mathematics, University of Notre Dame AG11 Abstracts 61

Lukas Katth¨an MS32 University of Marburg Pivoting Strategies for Sparse Matrices over Finite [email protected] Fields Solving a polynomial system of equations over a finite field MS31 is at the heart of algebraic cryptanalysis and numerous Polytopes and Imsets other applications, including the Quadratic Sieve. Several algorithms for solving such systems of equations over those We study polytopes and imsets of generalized tree graphs. fields involve reducing a matrix over a finite field. In this paper, we analyze a plethora of pivoting strategies, ranging from the naive to the complex, for performing structured Patrik Nor´en Gaussian Elimination to minimize fill-in and field operation KTH Stockholm counts. We consider random matrices, as well as matrices [email protected] of the forms produced by the Quadratic Sieve and also the XL/F4 algorithms. The ability of the various techniques to MS31 maintain sparsity is compared for various sizes and input matrices, as well as the running times. Conditional Independence Ideals and Robustness Gregory Bard The talk discusses a class of conditional independence (CI) Mathematics Department ideals motivated by conceptual studies of robustness [Ay, N., and Krakauer, D. C. Fordham University, The Bronx, NY. , “Geometric robustness theory [email protected] and biological networks,’ Theory in Biosciences, vol. 125, no. 2, pp. 93 – 121, 2007.]. Consider a collection of n +1 X ,X ,...,X finite random variables 0 1 n with the interpre- MS32 X tation that 0 is the output of the system, computed from Groebner Bases and Linear Algebra the inputs X1,...,Xn. Such a system is robust if the com- plete knowledge of all input variables is not necessary in There is a strong interplay between computing efficiently order to compute the output. This notion of robustness can Gr¨obner bases and linear algebra. In this talk, we focus on be modelled as a collection of CI statements, generalizing several aspects of this convergence: the set of CI statements studied in [Fink, A.,“Thebino- mial ideal of the intersection axiom for conditional proba- • Algorithmic point of view: algorithms for comput- bilities,’ Journal of Algebraic Combinatorics, vol. 33, no. 3, ing efficiently Gr¨obner bases (F4,F5,...) not only rely pp. 455–463, 2011.]. CI statements give polynomial restric- heavily on linear algebra but were based on linear al- tions on the joint probability distribution of the system. If gebra techniques. Recent algorithms to change the the output variable is binary, then the resulting CI ideal ordering of a Gr¨obner basis is also related to multivari- is a binomial edge ideal, as defined in [Herzog, J., Hibi, ate generalization of the Wiedemann algorithm. Mix- T., Hreinsdottir,´ F., Kahle, T., and Rauh, J.,“Bino- ing Gr¨obner bases methods and linear algebra tech- mial edge ideals and conditional independence statements,’ nique for solving sparse linear systems leads to an ef- Advances in Applied Mathematics, vol. 45, no. 3, pp. 317 ficient algorithm to solve boolean quadratic equations; – 333, 2010.]. The results generalize to the case that the this algorithm is faster than exhaustive search by an output is non-binary. In particular, the CI ideal is radical, exponential factor and the primary decomposition has a nice interpretation. • Application point of view: for instance, a generaliza- Johannes Rauh tion of the eigenvalue problem to several matrices – Max Planck Institute for Mathematics in the Sciences the MinRank problem – is at the heart of the security [email protected] of many multivariate public key cryptosystems. • Design of library: when implementing Gr¨obner bases Nihat Ay algorithms the crucial task is the linear algebra part; Max Planck Institute for Mathematics in the Sciences moreover the specific structure of the matrices has [email protected] to be taken into account to speedup even more the computations. MS32 Linear Algebra over Dense Matrices over GF(2) Jean-Charles Faugere and Small Extensions - Albrecht Universite Pierre and Marie Curie [email protected] Linear algebra over F2 and F2e where e is a small integer has many applications such as cryptography and coding theory. Since matrices over these base fields can be effi- MS32 ciently bit-packed, dense techniques are often feasible for Verified Numerical Linear Algebra: Linear System considerable matrix sizes beyond 100,000 x 100,000. In Solving this talk we will present our work on matrices over these fields and discuss algorithms, implementation techniques Solving numerically a linear system can be performed very and issues. efficiently, using optimized routines, but it yields an ap- proximate solution without any indication about its accu- Martin Albrecht racy. Getting an enclosure of the error between the ap- Universit´e Pierre et Marie Curie proximate and the exact solutions is called ”verification”. Paris, France We present a verified algorithm which gets an accurate en- [email protected] closure, with a moderate overhead both in complexity and in practical performance. Its key ingredients are interval arithmetic, iterative refinement and well-chosen computing 62 AG11 Abstracts

precision. simulations as well as exhaustive algebraic computations for graphs with up to five nodes. Nathalie Revol Arenaire project, LIP, Ecole Normale Superieure de Lyon Rina Foygel and Lab. ANO, Univ. Lille, France Department of Statistics [email protected] University of Chicago [email protected]

MS32 Jan Draisma Exact Solutions to Mixed-integer Linear Program- Eindhoven University of Technology ming Problems [email protected] We present a hybrid symbolic-numeric approach for solving mixed-integer-programming problems exactly over the ra- Mathias Drton tional numbers. By performing many operations using fast Department of Statistics floating-point arithmetic and then verifying and correcting University of Chicago results using symbolic computation, exact solutions can [email protected] be found without relying entirely on rational arithmetic. Computational results will be presented based on an ex- MS33 act branch-and-bound algorithm implemented within the constraint integer programming framework SCIP. Parameter Identification of Structural Equation Models Dan Steffy Zuse Institute Berlin Structural equation models (SEMs) are used to formalize steff[email protected] a variety of causal queries as certain types of probability distributions. A central problem in SEMs is the analysis of identification. A model is identified if it only admits William Cook a unique parametrization to be compatible with a given Georgia Institute of Technology data set. Here, I will introduce a computer algebra soft- [email protected] ware to test identifiability, and discuss recent advances on global identification via the analysis of the Jacobian of the Thorsten Koch, Kati Wolter parametrization. Zuse Institute Berlin [email protected], [email protected] Luis D. Garcia-Puente Department of Mathematics and Statistics Sam Houston State University MS33 [email protected] Algebraic Problems in Graphical Modeling A graphical model is a statistical model associated with a MS33 graph whose nodes correspond to random variables. I will Discrete Graphical Models with One Hidden Vari- give an introduction to these models and provide statistical able motivation for algebraic problems that will be addressed by the speakers in the session. Kruskal’s theorem applies to a simple latent class model, in which three observed variables are independent when con- Mathias Drton ditioned on a single hidden one. Our aim is to give general Department of Statistics conditions for identifiability using Krukal’s theorem and, University of Chicago when the conditions fail to hold, to characterize the space [email protected] where identifiability breaks down. We present a technique that, given an arbitrary directed graphical model with a single hidden variables, modifies the model in such a way MS33 that we can apply Kruskal’s theorem. The technique is Parameter Identifiability via the Half-trek Crite- based on the following three operations: rion in Mixed Graphs • Clump several variables (all hidden or all observed) In the setting of Gaussian graphical models, mixed graphs into a single one, with larger state space. allow for both linear effects between the nodes (directed • Condition on the state of an observed variable, edges) and covariance among the Gaussian errors on each • Marginalize over an observed variable (making it hid- node (bidirected edges). We study parameter identifiabil- den). ity in these models, that is, we ask for conditions that Each of these operations can be done multiple times, and ensure that the edge coefficients and correlations appear- in combination with one another. ing in a linear structural equation model can be uniquely recovered from the covariance matrix of the associated nor- Elizabeth S. Allman, John A. Rhodes mal distribution. In particular, we are interested in generic University of Alaska Fairbanks parameter identifiability, where recovery is possible for al- [email protected], [email protected] most every choice of parameters. We give a new graphical criterion that is sufficient for generic identifiability. Unlike Elena Stanghellini prior work on this problem, our criterion does not require Dipartimento ti Economia Finanza e Statistica the directed part of the graph to be acyclic, and addition- Universita’diPerugia ally it improves on existing criteria for the acyclic case. We [email protected] also develop a related necessary condition, and examine the “gap” between sufficient and necessary conditions through Marco Valtorta AG11 Abstracts 63

Department of Computer Science and Engineering [email protected] University of South Carolina [email protected] MS34 Homotopy Theory of Dynamical Systems MS33 Identifiability of Phylogenetic Tree Models for Bi- A dynamical system is a space X with a pairing from X x S nary Data to X for some parameter space S, and a map of such dynam- ical systems is an S-equivariant map. There is an injective Our focus in this talk is on undirected discrete graphical tree models when all the variables in the system are binary, and a projective Quillen model structure for the result- where leaves represent the observable variables and where ing category of spaces with S-action, and both are easily all the inner nodes are unobserved. A novel approach, derived. Simultaneously varying the parameter space S based on the theory of partially ordered sets, allows us to along with the space X up to weak equivalence is more in- obtain a convenient parametrization of this model class. A teresting, and requires a new model structure having weak simple product-like form of the resulting parameterization equivalences defined by homotopy colimits, as well as a gen- gives insight into identifiability issues associated with this eralization of Thomason’s model structure for small cate- model class. In particular, we provide necessary and suffi- gories. These techniques give methods of detecting when cient conditions for such a model to be identified up to the one dynamical system is close to another, either spatially switching of labels of the inner nodes. When these con- or combinatorially. ditions hold we give explicit formulas for the parameters of the model. Whenever the model fails to be identified Rick Jardine we use the new parameterization to describe the geometry University of Western Ontario of the unidentified parameter space. We illustrate these [email protected] results using a simple example.

Piotr Zwiernik MS34 Mittag-Leffler Institute Detecting Persistent Knotting Stockholm, Sweden 3 [email protected] GivenafiltrationofspacesX0 ⊂ X1 ⊂···⊂Xn ⊂ R , we discuss what it means for any knotting of the space Xi be persist in the larger space Xi+p. This generalizes MS34 techniques from classical knot theory. In particular, this Evasion Paths in Mobile Sensor Networks knotting can be characterized by the persistent homotopy groups of its complement. We will discuss techniques to In “Coordinate-free Coverage in Sensor Networks with determine if knotting of a complex persists in a larger space Controlled Boundaries via Homology,” Vin de Silva and or if the knotting is transient. Robert Ghrist use the local connectivity data of a mobile sensor network to determine, in some cases, whether an David Letscher evasion path exists in the network. We consider exam- Saint Louis University ples that show the existence of an evasion path depends [email protected] not only on the network’s connectivity data but also on its embedding, and we search for invariants of the embedding that provide sharper criteria for the existence of an evasion MS35 path. Rank 4 Premodular Categories Henry Adams By relaxing the non-degeneracy condition of modular cat- Stanford University egories one obtains the more general class of premodular [email protected] categories. These are of interest for several reasons: firstly, as spherical categories they lead to modular categories via the Drinfeld center construction. Secondly, they give rise MS34 to link invariants. Finally, recent constructions suggest Mapping Spaces in Computational Algebraic that to get non-trivial 4-D TQFTs the modularity condi- Topology tion should be removed. We will discuss recent work on the classification of rank 4 premodular categories. A basic aspect of the modern perspective on homotopy theory is that the set of maps between spaces can itself Paul Bruillard be regarded as a topological space, where paths in this Math. Dept. mapping space represent homotopies. All of the homotopy Texas A&M University theory of spaces can be encoded in the data of the mapping [email protected] space. This talk will present recent work on combinatorial simplicial mapping spaces in computational topology, em- phasizing connections between simplicial subdivision and MS35 sampling. Principle of Maximum Entropy and Ground Spaces of Local Hamiltonians Andrew Blumberg University of Texas Correlations are of key importance in understanding many [email protected] fundamental phenomena of many-body quantum physics. Motivated by the concept of irreducible correlations stud- Michael Mandell ied by [Linden et al., PRL 89, 277906] and [Zhou, PRL University of Indiana 101, 180505], which is in turn based on the principle of maximum entropy, we present that, some quantum states 64 AG11 Abstracts

are with certain k-particle correlation if and only if they liptic Curves span a ground space of some k-local Hamiltonian. It es- tablishes a better understanding of k-body correlations and It is not known whether each complex dimension d contains a set of equiangular lines with the maximum possible car- builds an intimate link of correlations of quantum states to 2 ground spaces of local Hamiltonians. dinality of d (a SIC-POVM). Zauner conjectured that one always exists as an orbit of a finite Heisenberg group. In di- Bei Zeng mension 3, Hughston observed that the inflection points of Dept. of Math. and Stats. any Hesse cubic give a particular Heisenberg SIC-POVM. I University of Guelph give a geometric interpretation for the entire infinite family [email protected] as the 3-torsion points of singular elliptic curves.

Zhengfeng Ji Jon Yard Perimiter Institute Los Alamos National Laboratory [email protected] [email protected]

MS35 MS36 Localizing Topological Quantum Computers Convexity and SOS-Convexity

A multivariate polynomial p(x)=p(x1, ..., xn)issos- The quantum circuit and topological models for quantum H x H x computation are computationally equivalent in the sense convex if its Hessian ( ) can be factored as ( )= M T x M x that suitably chosen models in each setting can efficiently ( ) ( ) with a possibly nonsquare polynomial matrix M x simulate a universal model in the other setting. I will dis- ( ). The notion of sos-convexity is a sufficient condition cuss several recent attempts to refine this theoretical result for convexity of polynomials that can be checked efficiently to a more practical, hybrid model. This is based upon joint with semidefinite programming. We will present three re- work with Zhenghan Wang. sults on the subject: (i) We will show that two other nat- ural sos relaxations for convexity based on the definition Eric Rowell of convexity and its first order characterization turn out to Texas A&M University be equivalent to sos-convexity. (ii) We will present an ex- [email protected] plicit example of a convex but not sos-convex polynomial in six variables and of degree four, whose construction comes Zhenghan Wang directly from our recent proof of NP-hardness of deciding Microsoft Research convexity of quartic polynomials. (iii) We will show that n d Station Q for polynomials in variables and of degree , the notions [email protected] of convexity and sos-convexity are equivalent if and only if n =1ord =2or(n, d)=(2, 4). Remarkably, these are exactly the cases where nonnegative polynomials are MS35 guaranteed to be sums of squares, as proven by Hilbert. Hidden Symmetry Subgroup Problems in Quan- Amir Ali Ahmadi tum Computing MIT We advocate a new way of addressing hidden structure a a [email protected] problems and of finding efficient quantum algorithms. We define and investigate the Hidden Symmetry Subgroup MS36 Problem (HSSP), a generalization of the well studied Hid- den Subgroup Problem (HSP). Given a group acting on Positive Semidefinite Gorenstein Ideals a set and an oracle whose level sets define a partition of I will explain how positive semidefinite Gorenstein ideals the set, the task is to recover the subgroup of symmetries naturally arise as objects that separate nonnegative poly- of this partition inside the group. The HSSP provides a nomials from sums of squares. It is possible to deduce all unifying framework that, beside the HSP, encompasses a of the cases where nonnegative polynomials are equal to wide range of oracle problems, including hidden polyno- sums of squares as the cases where psd Gorenstein ideals mial problems. While we show that the HSSP can have do not exist. I will present applications to decompositions instances with exponential quantum query complexity, for of ternary forms as sums of powers of rational functions various instances we obtain efficient quantum algorithms. and symmetric tensor decompositions To achieve that we design a general framework for reducing the HSSP to the HSP which works efficiently in many in- Greg Blekherman teresting cases related to symmetries of polynomials. The University of California, San Diego HSSP therefore connects in a rather surprising way cer- [email protected] tain hidden polynomial problems with the HSP. Using this connection, we obtain the first ever efficient quantum al- gorithm for the hidden polynomial problem, for quadratic MS36 polynomials over fields of constant characteristic. Strong Nonnegativity and Sums of Squares on Real Varieties Pawel Wocjan Computer Science Dept. Motivated by scheme theory, we introduce strong nonneg- U. Central Florida ativity on real varieties, which has the property that a sum [email protected] of squares is strongly nonnegative. We show that this al- gebraic property is equivalent to nonnegativity for nonsin- gular real varieties. Moreover, for singular varieties, we MS35 reprove and generalize obstructions of Gouveia and Netzer SIC-POVMs in Dimension Three via Singular El- to the convergence of the theta body hierarchy of convex AG11 Abstracts 65

bodies approximating the convex hull of a real variety. in rings of algebraic integers.

Mohamed Omar Nadia Heninger UC Davis and Caltech Princeton University [email protected] [email protected]

Brian Osserman UC Davis MS37 [email protected] List Decoding for AG Codes Using Gr¨obner Bases Finding efficient algorithms for the interpolation step in MS36 Guruswami-Sudan list decoding remains a problem of in- The Algebraic Boundary of SO(2)-orbitopes terest. We will discuss how the recent algorithms of [Lee- O’Sullivan, List Decoding of Hermitian Codes, JSC 44 An SO(2)-orbitope is the convex hull of an orbit under (2009), 1662-1675] and [Trifonov, Efficient Interpolation some linear action of SO(2) on a finite dimensional real in the Guruswami-Sudan Algorithm, preprint 2010], which vector space. Such a set is always a compact, convex, semi- find the desired interpolating polynomial in a certain mod- algebraic set. We will characterize the Zariski closure of its ule Gr¨obner basis, can be extended to codes from other boundary in the 4-dimensional case. One component is the curves. We will also compare their methods with an FGLM secant variety to the Zariski closure of the orbit, which is approach. a real algebraic curve, and for the other components, we give explicit equations. Our answer will have implications John B. Little on the semi-algebraic geometry of these orbitopes and on College of the Holy Cross the question, whether or not they are spectrahedra. This [email protected] is work in progress.

Rainer Sinn MS37 University of Konstanz General Bounds for Generalized Toric Codes [email protected] Toric codes are algebraic evaluation codes corresponding to linear systems on toric varieties. Their parameters are MS36 strongly related to the geometry of the lattice polytope The Central Curve of a Semidefinite Program defining a liner system. Recently Diego Ruano introduced, so called, generalized toric codes (GTC) when one takes The central curve of a semidefinite program is an algebraic arbitrary subspaces of linear systems on a toric variety. It curve specified by an affine linear space of symmetric ma- is harder to capture their properties as they are defined by trices and a cost vector. Here we examine the algebraic arbitrary finite lattice point configurations S which lack properties and beautiful geometry of this curve. We will the geometry of lattice polytopes. We introduce a gen- focus on examples and special cases coming from appli- eral lower bound for the minimum distance of a GTC. The cations. In the special case of diagonal matrices (linear bound involves the minimum distance of GTCs defined by programming) we will see connections to hyperplane ar- projections of the configuration S and the fibers of the pro- rangements and matroid theory. jection.

Cynthia Vinzant Ivan Soprunov University of California, Berkeley Cleveland State University [email protected] [email protected]

MS37 MS37 Ideal Forms of Coppersmith’s Theorem and Toric codes and Minkowski Length of 3D Polytopes Guruswami-Sudan List Decoding, Part 2 The Minkowski sum of two polytopes is the set of all pair- Abstract not available at time of publication. wise sums of their points. The central object of my talk is the Minkowski length L(P ) of a lattice polytope P which Henry Cohn is defined to be the largest number of primitive lattice Microsoft Research New England segments whose Minkowski sum is in P . The Minkowski [email protected] length represents the largest possible number of factors in a factorization of polynomials with exponent vectors in P MS37 and comes up in lower bounds for the minimum distance Ideal Forms of Coppersmith’s Theorem and of toric codes. I will explain some combinatorial results Guruswami-Sudan List Decoding about L(P )whereP is a 3D lattice polytope in connection with 3D toric codes. Coppersmith’s algorithm is a celebrated technique for find- ing small solutions to polynomial equations modulo inte- Jenya Soprunova gers. It has many important applications in cryptogra- Kent State University phy, particularly in the cryptanalysis of RSA. In this talk, [email protected] we show how the ideas of Coppersmith’s theorem can be extended to a more general framework encompassing the Olivia Beckwith original number-theoretic problem, list decoding of Reed- Harvey Mudd College Solomon and algebraic-geometric codes, and the problem [email protected] of finding solutions to polynomial equations modulo ideals 66 AG11 Abstracts

Matthew Grimm MS38 Kent State University Using Homotopy Continuation to Model Rate- [email protected] dependent Magnetic Phenomena

Bradley Weaver Rate-dependent magnetic phenomena occur because ther- Grove City College mal fluctuations can drive a system of magnetic moments [email protected] over energy barriers between stable states. A new method for modeling these transitions uses the Bertini homotopy continuation package. Systems of polynomial equations are MS38 solved for the stable states and energy barriers. The so- A Domain Decomposition Algorithm for Comput- lutions are incorporated in a master equation for the time ing Multiple Steady States of Differential Equa- evolution of the probability of each stable state. tions Andrew Newell Problems in engineering often lead to systems of partial North Carolina State University deferential equations, for which the only hope of solution andrew [email protected] is to compute numerical solutions. This talk will describe domain decomposition method for solving polynomial sys- Matt Niemerg tems derived from the discretization of differential equa- Colorado State University tions based on homotopy continuation. This method di- Fort Collins, CO vides the domain into subdomains and solves the polyno- [email protected] mial system arising from each subdomain. Homotopy con- tinuation then uses these solutions to build solutions for Daniel J. Bates the original domain. Colorado State University Department of Mathematics Wenrui Hao [email protected] Dept. of Applied and Comp. Mathematics and Statistics University of Notre Dame [email protected] MS38 Point Clouds of Varities and Persistent Homology Jonathan Hauenstein Texas A&M University Computing topological invariants of algebraic objects is si- [email protected] multaneously challenging, from an algebraic standpoint, and of large interest to application fields. We approach Bei Hu this challenge with a two-pronged method: first, using Dept. of Applied & Computational Mathematics & techniques from homotopy continuation and numerical al- Statistics gebraic geometry, we generate point clouds – finite sample University of Notre Dame sets – on the variety, and subsequently, we approach the [email protected] point cloud with techniques from applied algebraic topol- ogy. Numerical algebraic geometry provides techniques to Andrew Sommese solve systems of polynomial equations and retrieve solution Department of Mathematics, University of Notre Dame points that are guaranteed to contain the isolated points Notre Dame, IN 46556-4618, USA in the variety defined by the equational system. By inter- [email protected] secting the variety of interest with generic hyperplanes, we can produce equation systems with isolated points located on all components of the original variety. Furthermore, MS38 the techniques of persistent homology were originally de- Numerical Algebraic Geometry Applied to Various veloped in order to acquire topological information from Theoretical Physics Problems finite point samples, from point clouds. We apply these techniques to gain computational indications of the topo- Nonlinear equations arise in theoretical physics naturally logical structure of the point cloud generated from homo- and frequently. In general, it is always difficult to solve topy continuation, and describe some ways to decompose (i.e., get ALL solutions, either exactly or numerically) and re-assemble topological information for better compu- them. However, if the non-linearity of the equations is tational performance. polynomial-like, then one can use the computational and numerical algebraic geometry methods to solve the equa- Mikael Vejdemo-Johansson tions exactly or numerically, respectively. In the talk, this Stanford University will be explained in more detail by taking a few examples [email protected] from theoretical physics, namely, in statistical mechanics, string theory and lattice field theories, polynomial equa- Jonathan Hauenstein tions arise naturally and it will be shown how the alge- Texas A&M University braic geometry methods can be used to get very interesting [email protected] physics out. David Eklund Dhagash Mehta Institut Mittag-Leffler Department of Physics [email protected] Syracuse University [email protected] Martina Scolamiero Politecnico di Torino [email protected] AG11 Abstracts 67

Chris Peterson may compute approximate implicit equations. Colorado State University [email protected] Tatjana Kalinka,IoannisZ.Emiris National and Kapodistrian University of Athens [email protected], [email protected] MS39 Approximate Implicitization using Chebyshev Christos Konaxis Polynomials National Kapodistrian University of Athens [email protected] Whereas traditional approaches to implicitization of ratio- nal parametric curves have focussed on exact methods, the past two decades have seen increased interest in the appli- MS39 cation of approximate methods for implicitization. In this Approximate Implicitization of Envelope Surfaces talk we will discuss how the properties of the Chebyshev polynomial basis can be used to improve the speed, sta- Given a rational family of parametric surfaces in a cer- bility and approximation quality of existing algorithms for tain region of interest, we are interested in computing an approximate implicitization. We will also look at how the implicit representation of the envelope. Although several algorithm is well suited to parallelization. methods for exact implicitization are known, they tend to be computationally expensive and may even introduce un- Oliver Barrowclough wanted branches and singularities. We adapt the concept SINTEF of approximate implicitization to envelopes and formulate Norway an algorithm for computing a piecewise algebraic approxi- [email protected] mation of low degree.

Tino Schulz MS39 Johannes Kepler University Blending Natural Quadrics and Associated Vol- Linz, Austria umes [email protected]

Natural quadrics are simplest primitive shapes used in Bert J¨uttler CAD: spheres, right circular cones or right circular cylin- Johannes Kepler Univ. Linz ders. We construct exact rational fixed and variable ra- Department of Applied Geometry dius blends of minimal possible parametrization degrees [email protected] between two natural quadrics using canal surfaces in all cases where it is possible. Bounds on parameters that en- sure non-singular blends and associated volumes will be MS40 presented. Dynamics of Weakly Reversible Reaction Networks HeidiE.I.Dahl and the Global Attractor Conjecture Vilnius University I will discuss some recent work pertaining to weakly re- [email protected] versible chemical reaction networks and, in particular, the Global Attractor Conjecture. I will present some details of MS39 the proof of the conjecture in the setting where the under- lying reaction diagram consists of a single linkage class, or Why Approximate Algebraic Methods? connected component.

The talk will address the potential of the use of approx- David Anderson imate algebraic methods in CAGD systems, with a focus Department of Mathematics on intersection and self-intersection algorithms. An intro- University of Wisconsin Madison duction to the original approach to approximate impliciti- [email protected] zation will be given.

Tor Dokken MS40 SINTEF ICT, Department of Applied Mathematics Oslo, Norway QSSA and Algebraic Elimination in Chemical Ki- [email protected] netics Chemical mechanisms for reaction network dynamics in- MS39 volve many highly reactive and short-lived species (inter- mediates), present in small concentration, in addition to Application of Numerical Methods in Implicitiza- the main reactants and products, present in larger concen- tion tration. A classic model reduction method known as the Relying on methods for implicit support prediction, we re- quasi-steady-state approximation (QSSA) is often used to duce implicitization of rational curves and (hyper)surfaces eliminate the highly reactive intermediate species and re- to interpolating the implicit coefficients. We apply nu- move the large rate constants that cannot be determined merical methods, including SVD, in order to compute a from concentration measurements of the reactants and matrix kernel. Computational experiments in Maple and products. We show that the program usually taught to Matlab are performed to compare different approaches for students for applying the 100 year-old approach of clas- constructing a real or complex matrix, either square or sic QSSA model reduction cannot be carried out in many, rectangular, and for computing its kernel, with respect to perhaps most, relevant kinetics problems. In particular, by robustness and speed. Our methods exploit sparsity and using Galois theory, we prove that the required algebraic equations cannot be solved for as few as five bimolecular 68 AG11 Abstracts

reactions between five species. We also describe algorithms can only approach the boundary of the non-negative or- that can test any mechanism for solvability, and propose thant at points belonging to a semi-locking set. We general- a new strategy for dealing with unsolvable systems, based ize results of Angeli, De Leenheer and Sontag (2007) which on rescaling the reactive intermediate species. assume that every semi-locking set satisfies a conservation condition or is dynamically non-emptiable by introducing a Gheorghe Craciun notion of a weakly dynamically non-emtiable semi-locking Department of Mathematics, University of set. Systems whose semi-locking sets are weakly dynami- Wisconsin-Madison cally non-emptiable are persistent. The facet result of An- [email protected] derson and Shiu (2010) also fits into this framework.

Casian Pantea, James Rawlings David Siegel, Matthew Johnston University of Wisconsin, Madison University of Waterloo [email protected], [email protected] [email protected], [email protected]

MS40 Catalysis in Reaction Networks MS41 Pentads and Minors Define the Gaussian Two- We define catalytic networks as chemical reaction networks factor Model with an essentially catalytic reaction pathway: one which is “on’ in the presence of certain catalysts and “off’ in their In 1928, educational psychologist Truman Lee Kelley pub- absence. We show that examples of catalytic networks lished Crossroads in the Mind of Man, which contained a include synthetic DNA molecular circuits that have been new pentad test for Gaussian factor analysis with two fac- shown to perform signal amplification and molecular logic. tors. Pentads are degree-five polynomials that vanishes on Recall that a critical siphon is a subset of the species in a all symmetric n × n-matrices that can be expressed as a chemical reaction network whose absence is forward invari- rank-two matrix plus a diagonal matrix. In their recent al- ant and stoichiometrically compatible with a positive point. gebraic approach to factor analysis, Drton, Sturmfels, and Our main theorem is that all weakly-reversible networks Sullivant raise the problem of determining all such polyno- with critical siphons are catalytic. Consequently, we obtain mials. I will report on a finite computation that determines new proofs for the persistence of atomic event-systems of a generating set of polynomials for all n. Thisisjointwork Adleman et al., and normal networks of Gnacadja. We de- with Andries Brouwer. fine autocatalytic networks, and conjecture that a weakly- reversible reaction network has critical siphons if and only Jan Draisma if it is autocatalytic. T.U.Eindhoven, The Netherlands Department of Mathematics Manoj Gopalkrishnan [email protected] Tata Institute of Fundamental Research Mumbai, India Andries Brouwer [email protected] T.U. Eindhoven, The Netherlands Department of Mathematics and Computer Science [email protected] MS40 Linear Conjugacy of Chemical Reaction Networks MS41 Under suitable assumptions, the dynamic behaviour of a Geometry of Deep Belief Networks chemical reaction network is governed by an autonomous set of polynomial ordinary differential equations over con- We report on recent progress on the algebraic statistics tinuous variables representing the concentrations of the re- of deep belief networks. These are graphical models built actant species. It is known that two networks may possess from layered restricted Boltzmann machines, which in turn the same governing mass-action dynamics despite disparate correspond to Hadamard products of secant varieties of network structure. To date, however, there has only been Segre varieties. limited work exploiting this phenomenon even for the cases where one network possesses known dynamics while the Jason Morton other does not. In this presentation, I will present results Pennsylvania State University which bring these known results together into a broader [email protected] unified theory which we have called conjugate chemical re- action network theory. In particular, I will present a theo- rem which gives conditions under which two networks with MS41 different governing mass-action dynamics may exhibit the Parametrization of Causal Models Associated with same qualitative dynamics and use it to extend the scope Acyclic Directed Mixed Graphs of the well-known theory of weakly reversible systems. Acyclic directed mixed graphs may be used to repre- Matthew D. Johnston, David Siegel sent causal directed acyclic graph (DAG) models in which University of Waterloo some variables are unmeasured (latent). In this talk we [email protected], [email protected] present a parametrization of the observed distribution in the multivariate binary case. The resulting model satisfies Markov constraints, and so-called Verma constraints. The MS40 parametrization leads to efficient algorithms for computing Persistence of Chemical Kinetics Systems intervention distributions, as well as allowing scoring-based model search. [Joint work with Robin Evans (University It is known that a solution to a chemical kinetics system of Washington), James Robins and Ilya Shpitser (Harvard AG11 Abstracts 69

School of Public Health).] Applications of Numerical Algebraic Geometry minisym- posium, though, I will keep the introductory parts short! Thomas Richardson Statistics University of Washington Daniel J. Bates [email protected] Colorado State University Department of Mathematics [email protected] MS41 Treks and Determinants of Path Matrices in Gaus- sian Graphical Models MS42 Numerical Algebraic Geometry via Numerical This talk will explore a connection between combinatorics Polynomial Algebra and algebraic statistics. In particular, we will look at Gaus- sian graphical models, whose covariance matrices can be Techniques of numerical polynomial algebra are applied to given in terms of certain path families called treks. Inspired numerical algebraic geometry. These techniques include by classical results in algebraic combinatorics, we develop Sylvester and Macaulay matrices, and local and global du- a graph-theoretic criterion for determining the rank of a ality. Macaulay’s H-bases are connect the homogeneous submatrix of the covariance matrix. and affine cases. The computations are done by use of SVD based numerical linear algebra using Mathematica. Kelli Talaska The main goal of these methods is to provide explicit poly- Department of Mathematics nomials defining the ideals defining algebraic sets which are University of California, Berkeley components of, or unions of, other algebraic sets. [email protected] Barry H. Dayton Seth Sullivant Department of Mathematics North Carolina State University [email protected] [email protected] MS42 Jan Draisma T.U.Eindhoven, The Netherlands Certification of Approximate Multiple Roots Department of Mathematics Approximation and certification of multiple roots is a real [email protected] challenge in numerical computation. In this talk, we will describe a symbolic-numeric algorithm for the certification MS41 of singular isolated points. The approach is based on the computation of the inverse systems of the isolated singular Geometry of Maximum Likelihood Estimation in point. We derive a one-step deflation technique, from the Gaussian Graphical Models description of the multiplicity structure in terms of differ- We study maximum likelihood estimation in Gaussian entials. The deflated system can be used in Newton-based graphical models from a geometric point of view. In current iterative schemes with quadratic convergence. Starting applications of statistics, we are often faced with problems from a polynomial system and a sufficiently small neighbor- involving a large number of random variables, but only a hood, we obtain a criterion for the existence and unique- small number of observations. An algebraic elimination ness of a singular root of a given multiplicity structure, criterion allows us to find exact lower bounds on the num- applying a well-chosen symbolic perturbation. Standard ber of observations needed to ensure that the maximum verification methods, based e.g. on interval arithmetic and likelihood estimator exists with probability one. This is a fixed point theorem, can be employed to certify that there applied to bipartite graphs and grids, leading to the first exists a unique perturbed system with a singular root in instance of a graph for which the maximum likelihood es- the domain. timator exists with probability one even when the number Bernard Mourrain of observations equals the treewidth of the graph. INRIA Sophia Antipolis Caroline Uhler [email protected] Department of Statistics UC Berkeley Angelos Mantzaflaris [email protected] GALAAD, INRIA Sophia Antipolis angelos.mantzafl[email protected]

MS42 Basic Notions and Current Trends in Numerical MS42 Algebraic Geometry Isosingular Sets and Deflation This talk will overview some of the recent advances within This talk will discuss the new concept of isosingular sets, the field of numerical algebraic geometry without dis- which are pure-dimensional irreducible algebraic subsets cussing applications (covered in an earlier minisympo- of the set of solutions to a system of polynomial equations sium). The goal is to describe some of the latest trends such that generic points of an isosingular set share a com- and current open problems in advancing this field. Basic mon singularity structure. The definition of these sets de- notions of numerical algebraic geometry (homotopy contin- pends on deflation, a procedure that uses differentiation to uation, numerical irreducible decomposition, etc.) will be regularize solutions. A weak form of deflation has proven introduced very briefly as this is an introductory talk for useful in regularizing algebraic sets, making them amenable this minisymposium. Coming after the 12-15 talks in the to treatment by the algorithms of numerical algebraic ge- 70 AG11 Abstracts

ometry. We introduce a strong form of deflation and use We discuss methods of discretization that use matrices that it to define deflation sequences, which are similar to the are parametrized by finite Frobenius rings, and sketch how sequences arising in Thom-Boardman singularity theory. quantum codes can be constructed in this case. Isosingular sets consist of points that have the same defla- tion sequence. From this, one may define the isosingular Andreas Klappenecker set of a given point and also the isosingular local dimen- Computer Science Dept. sion. While isosingular sets are of theoretical interest as Texas A&M University constructs for describing singularity structures of algebraic [email protected] sets, they also expand the kinds of algebraic sets that can be investigated with methods from numerical algebraic ge- ometry. MS43 Classifying Small Index Subfactors Charles Wampler General Motors Research Laboratories, Enterprise I’ll describe recent progress on the classification of subfac- Systems Lab tors with index at most 5, and corollaries for fusion cate- 30500 Mound Road, Warren, MI 48090-9055, USA gories. [email protected] Scott Morrison Math. Dept. Jonathan Hauenstein U.C. Berkeley Texas A&M University [email protected] [email protected]

MS43 MS43 | | | |2 Recent Progress in Exactly Soluble Discrete Mod- ZKup = ZHenn for Lens Spaces els for Topological Phases in Two Dimensions

M. A. Hennings and G. Kuperberg defined quantum in- The study of two-dimensional topological phases in con- variants ZHenn and ZKup of 3-manifolds based on Hopf Z L p, q ,H densed matter systems is a frontier in the field of condensed algebras, respectively. We prove that Kup( ( ) )= matte theory as well as topological quantum computation. Z L p, q L p, q ,H Henn( ( )# ( ) ) for lens spaces when both in- Discrete or lattice models, which are exactly soluble have variants are based on a factorizable finite dimensional rib- been proposed by Kitaev and by Levin and Wen, respec- bon Hopf algebra H. tively. Here we present a summary of recent progress, made by us and other groups, in studying these models and their Liang Chang generalizations. The topics to be reviewed include 1) Dual- Math. Dept. ity between the Kitaev and Levin-Wen modeles in certain U.C. Santa Barbara special cases; 2) General procedure for computing ground [email protected] state degeneracy when the models are put on a topolog- ically non-trivial surface; 3) More detailed study of the Zhenghan Wang sectors of topolgical excitations; 4) General framework for Microsoft Research topology-preseving renormalization group flow that char- Station Q acterizes these models as fixed points; 5) Generalization of [email protected] these models to more general graphs or to incorporating more general degrees of freedom. Our approach, though closely related to topological field theory and tensor cate- MS43 gory theory, could be understood by physicists. Quantum Stabilizer Codes from Toric Varieties Yong-Shi Wu The technology to produce classical codes from higher di- Physics Dept. mensional algebraic varieties has been known for approxi- University of Utah mately 10 years. The combinatorial structure of toric va- [email protected] rieties made them particularly suitable for applying these techniques. Nevertheless, the only algebraic varieties from Spencer Stirling which one could construct quantum codes seemed to be Dept. of Math. curves. In this talk, we outline how to construct quantum University of Utah stabilizer codes from higher dimensional toric varieties. Ex- [email protected] amples of such codes constructed from toric surfaces will be discussed briefly. Yuting Hu Roy Joshua Dept. of Physics and Astronomy Math. Dept. University of Utah Ohio State U. [email protected] [email protected] MS44 MS43 Minimum Fuel Multi-impulse Orbit Transfer Quantum Error-Correcting Codes over Rings Most of the equations in orbital mechanics are expressed The discretization of errors in quantum error-correcting in terms of trigonometric functions. However, except for codes typically relies on a generalization of the Pauli matri- the case of Kepler’s equation, all the remaining instances ces. As a consequence, most quantum codes have been con- correspond to simple planar or spatial rotations. By rep- structed with the help of classical codes over finite fields. resenting these rotations by points in the unit circle (for AG11 Abstracts 71

2D rotations) or unit quaternions (for 3D rotations), all the Earth and other planets. The repetition of the ground the equations become polynomials, hence tractable with tracks allows for multiple scientific observations to be taken algebraic geometry techniques. In this talk, we will focus of the same column of space all the way to a little below mainly in the problem of orbit transfer using a finite se- the planet surface at various times of day. The symmetry quence of thrusts, and minimum fuel. and geometry of such orbits are governed by finite groups which classify them. Martin Avendano Texas A&M University Martin Lo [email protected] JPL-CalTech [email protected] Maurice Rojas Department of Mathematics, Texas A&M University [email protected] MS44 Evolution of Flower Constellations Theory Daniele Mortari The theory of Flower Constellations consists of a general Texas A&M University methodology to design constellations of Ns satellites. The [email protected] original theory evolved over time and a rich literature was produced with subsequent insights and mathematical re- MS44 formulations. This talk describes these mathematical pro- gresses making now the theory connected with the 2D and Reinterpreting Regularization of Collisions 3D Lattice theory and the Necklaces theory. through Real Algebraic Geometry Daniele Mortari Regularization of binary and/or simultaneous binary col- N Texas A&M University lisions in the collinear -body problem is a common tool [email protected] used to analyze the geometry of orbits near these kinds of singularities. The Levi-Civita type coordinate and time transformations typically used in regularization are rein- MS45 terpreted in terms of real algebraic geometry. This reinter- On the Decodability of Primitive Reed-Solomon pretation applies to Hamiltonians whose potential part is Codes a finite sum of reciprocals of homogeneous polynomials in the position coordinates. Reed-solomon codes are (list) decodable up to the Johnson- Guruswami-Sudan bound. No polynomial time decoding Lennard F. Bakker algorithm is known when number of errors is larger than Brigham Young University the JGS bound. It appears hard to establish complexity [email protected] results for the primitive Reed-Solomon codes. In this talk, I will present several results on this problem. I will also MS44 discuss whether the results can be generalized to algebraic geometric codes. Resonances in Celestial Mechanics Qi Cheng A dynamical system is in resonance p : q if there are two ω ω ω /ω p/q p University of Oklahama frequencies 1 and 2 such that 1 2 = ,with and [email protected] q two coprime numbers. Thus, resonances are related with periodic motions, and periodic motions are inherent to Ce- lestial Mechanics problems. Nature presents many exam- MS45 ples of resonances, for instance the orbital-rotational mo- Interpolation Decoding of a Class of Quasi-cyclic tion of the Moon, but there are many other cases of reso- Codes nances, which appear in the equations of motion and that creates many difficulties in the integration, mainly by an- For a quasi-cyclic code of index , we transform the de- alytic methods. For these cases, we present some ways coding problem into an interpolation problem for a Reed- in tackling this problem by means of appropriate sets of Solomon code. For each location,  values are assigned variables. Some examples are presented as well as some (counting multiplicities) and the usual interpolation decod- applications to the stability of the solutions. ing algorithms may be applied. We compute upper bounds on the decoding capability of the algorithm, showing that Antonio Elipe it can correct well beyond the minimum distance bound University of Zaragoza with high probability. Experimental results corroborate [email protected] our computations. Victor Lanchares Fernando Hernando, Michael E. O’Sullivan Universidad de la Rioja San Diego State University Spain [email protected], [email protected] [email protected] Diego Ruano Aalborg University MS44 [email protected] The Algebra and Geometry of Periodic Satellites Ground Tracks MS45 We survey the algebra and geometry of periodic satellites Small Bias Sets from General Algebraic Geometry ground tracks for remote sensing observations from space of 72 AG11 Abstracts

Codes [email protected]

Small bias sets are important in a number of applica- Laureano Gonzalez-Vega tions, from derandomization of algorithms to hash func- Universidad de Cantabria tions. There is a connection between small bias sets and Dpto. Matematicas linear codes. In this talk we consider the construction of [email protected] small bias sets from general algebraic geometry codes. Gretchen L. Matthews Mario Alfredo Fioravanti Villa Clemson University University of Cantabria [email protected] mario.fi[email protected]

MS45 MS46 Goppa-like Codes Beating the Best Known Codes Toric Degenerations of Toric Varieties and Appli- cations to Modeling The paper presents a construction of subfield subcodes from generalized Reed-Solomon codes that is similar to Toric varieties have interesting combinatorial properties classical Goppa codes, but simpler. Following an idea of that make them attractive for applications to geometric Roseiro et al, we use Delsarte’s theorem to create subfield modeling. They were employed by R. Krasauskas to define subcodes of larger than expected dimension by ensuring toric patches that are a generalization of B´ezier patches. their duals, which are trace codes, have small dimension. Toric degenerations were used by L. D. Garc´ıa-Puente, F. This method produced several codes over F5, F3 and F2 Sottile and C. Zhu to explain some ways that control points that beat the best known codes, and numerous others that govern the shape of toric patches when the weights vary. In match the best know parameters. this talk we will present our efforts to extend these proper- ties to irrational toric patches. This is a joint project with Michael E. O’Sullivan, Fernando Hernando, Kyle F. Sottile and N. Villamizar. Marshall San Diego State University Elisa Postinghel [email protected], [email protected], mar- University of Oslo [email protected] Norway [email protected]

MS45 Frank Sottile Extended Norm-Trace Codes Texas A&M University [email protected] Consider the function field Fqr (x, y)/Fqr defined by

u Nelly Villamizar x = L(y), CMA University of Oslo Norway r d i u| q −1 L y a yq [email protected] where q−1 and ( )= i is a linearized poly- i=0 d nomial with a0,ad =0and q distinct roots in qr .Note that the Hermitian and norm-trace function fields are spe- MS46 cial cases of this function field. Recent work has yielded On the Dimension of Triangular Splines explicit bases for certain Riemann-Roch spaces of this func- tion field. In this talk, we apply this progress to construc- We consider the space of spline functions defined on a trian- tion of codes arising from these spaces. gular subdivision of a polygonal domain. Using homologi- cal techniques we will present a lower and an upper bound Justin Peachey to the dimension of this spline space which are, in many or Clemson University perhaps most of the cases, more general and give better ap- [email protected] proximations to the exact value of the dimension than the already existing ones. These results can also be extended to spline spaces on 3-dimensional (simplicial) complexes. MS46 On the Algebraic Representation of Bisectors for Nelly Villamizar Low Degree Surfaces CMA University of Oslo Norway Bisectors are geometric constructions appearing very often [email protected] in CAD. For two given low degree parametric surfaces, it will be shown a new approach to determine an algebraic Bernard Mourrain representation of their bisector by using the so-called gener- INRIA Sophia Antipolis alized Cramer rules. The new introduced approach allows [email protected] to obtain easily a parametrization of the quadric-plane or quadric-cylinder bisectors, which is rational in most cases (in the remaining cases the parametrization involves one MS47 square root which is well-suited for approximation pur- Determining Multiple Steady States in Mass Ac- poses). tion Networks by Linear Inequality Systems

Ibrahim Adamou Ordinary Differential Equations (ODEs) are an important University of Cantabria tool in many areas of Quantitative Biology. For many ODE Spain AG11 Abstracts 73

systems multistationarity (the existence of ≥ 2 positive Wisconsin-Madison steady states) is a desired feature. For mass action net- [email protected] works establishing multistationarity is equivalent to estab- lishing the existence of at least two positive solutions of Fedor Nazarov a polynomial system with unknown coefficients. For net- University of Wisconsin-Madison works with certain structural properties, necessary and suf- [email protected] ficient conditions for multistationarity that take the form linear inequality systems are presented. MS47 Carsten Conradi Sequential and Distributive Multisite Phosphory- Max-Planck-Institut Dynamik komplexer technischer lations Have Toric Steady States Systeme [email protected] We show that the steady states of the sequential and distributive multisite phosphorylations system, are de- scribed by binomial equations, and can thus be explic- MS47 itly parametrized by monomials. This result is implicit Tools From Computational Algebraic Geometry for in [Wang and Sontag, 2008] and it is a particular case of the Study of Biochemical Reaction Networks [Thomson and Gunawardena, 2009]. We moreover give suf- ficient conditions for any chemical reaction system to have We will illustrate the use of some tools from computational toric steady states. This is joint work with Alicia Dicken- algebraic geometry for the study of (bio)chemical reaction stein, Anne Shiu and Carsten Conradi. networks, extracted from our joint work with Carsten Con- radi, Mercedes P´erez Mill´an, and Anne Shiu. Mercedes P´erez Mill´an Dto. de Matem´atica - FCEN - Universidad de Buenos Alicia Dickenstein Aires Universidad de Buenos Aires [email protected] [email protected] Alicia Dickenstein MS47 Universidad de Buenos Aires Complexity of Steady States in Molecular Net- [email protected] works and Numerical Algebraic Geometry Anne Shiu One of the most pressing questions in modern biology is Duke University how molecular interactions at the cellular level lead to [email protected] physiology, particularly disease. The most accessible math- ematical approach uses the law of mass-action to describe Carsten Conradi the underlying molecular network in terms of a polynomial Max-Planck-Institut Dynamik komplexer technischer dynamical system. Using tools from projective algebraic Systeme geometry coupled with high performance numerical analy- [email protected] sis, we investigate the complexity of steady states in net- works of established biological relevance, such as the Wnt pathway, and translate the parameter problem of biology MS48 into a question about complex structure moduli of the un- Finite Groebner Bases in Infinite Dimensional derlying algebraic variety. Polynomial Rings and Applications Robert Karp We discuss the theory of monoidal Groebner bases, a con- Harvard University cept which generalizes the familiar notion in a polynomial robert [email protected] ring and allows for a description of Groebner bases of ide- als that are stable under the action of a monoid. The main motivation for developing this theory is to prove finiteness MS47 theorems in commutative algebra and its applications. A Persistence and the Global Attractor Conjecture: major result of this type is that ideals in infinitely many Recent Approaches indeterminates stable under the action of the symmetric group are finitely generated up to symmetry. We dis- We describe recent approaches to proving the Persistence cuss how this machinery gives new proofs of some classical Conjecture (which describes a class of mass-action systems finiteness theorems in algebraic statistics as well as a proof for which variables do not approach zero) and the Global of the independent set conjecture of Hosten and the second Attractor Conjecture (which describes a class of mass- author. action systems for which trajectories converge to a single positive equilibrium). We show that an extended version Christopher Hillar of the Persistence Conjecture holds for systems with two- Mathematical Sciences Research Institute dimensional stoichiometric subspace and bounded trajec- [email protected] tories and we prove the Global Attractor Conjecture for systems with three-dimensional stoichiometric subspace. MS48 Casian Pantea Finiteness Theorems for Chains of Lattice Ideals University of Wisconsin, Madison [email protected] We study chains of lattice ideals that are invariant under a symmetric group action. In our setting, the ambient rings Gheorghe Craciun for these ideals are polynomial rings which are increasing in Department of Mathematics, University of (Krull) dimension. Thus, these chains will fail to stabilize 74 AG11 Abstracts

in the traditional commutative algebra sense. However, MS48 we prove a general theorem which says that ”up to the Finiteness on Homogeneous Markov Chain Models action of the group”, these chains stabilize up to monomial localization. This gives a partial resolution to a conjecture In 2010, Takemura and Hara showed a complete description of Aschenbrenner and Hillar. of a Markov basis for conditional tests of the toric homoge- neous Markov chain model, the envelope exponential fam- Christopher Hillar ily for the homogeneous discrete time Markov chain model, Mathematical Sciences Research Institute with the following cases; (1) two-state, arbitrary length; [email protected] (2) arbitrary finite state space and length of three; and (3) two-state, arbitrary length without initial parameters. Abraham Martin del Campo Motivated by these results, we study finiteness of Markov Texas A&M University bases for the toric homogeneous Markov chain model. In [email protected] this talk we consider the toric homogeneous Markov chain model without initial parameters and without any loops for the state space S with |S| =3, 4. A key tool is a state MS48 graph, that is, the directed multigraph G(x) such that ver- Defining Equations of Secant Varieties to Segre- tices are given by the states in S and the directed edges are Veronese Varieties given by the transitions from state i to j in x, a summary of moves in Markov chains. We describe the defining ideal of the rth secant variety of P2 × Pn embedded by O(∞, ∈), for arbitrary n and r ≤ 5. Ruriko Yoshida We also present the Schur module decomposition of the University of Kentucky space of generators of each such ideal. Our main results [email protected] are based on a more general construction for producing explicit matrix equations that vanish on secant varieties of Abraham Martin del Campo products of projective spaces. This extends previous work Texas A&M University of Strassen and Ottaviani. [email protected] Dustin Cartwright Yale University David Haws [email protected] University of Kentucky [email protected] Daniel Erman Stanford University MS49 [email protected] Mahonian Partition Identities Via Polyhedral Ge- ometry Luke Oeding University of California Berkeley In a series of papers, George Andrews and various coau- Universit`a degli Studi di Firenze thors successfully revitalized seemingly forgotten, powerful [email protected] machinery based on MacMahon’s Omega operator to sys- tematically compute generating functions of integer parti- tions. Our goal is to geometrically prove and extend many MS48 of the Andrews et al theorems, by realizing a given family Algebraic Identification of Binary-Valued Hidden of partitions as the set of integer lattice points in a certain Markov Processes polyhedron. This is joint work with Matthias Beck and Nguyen Le. The complete identification problem is to decide whether a stochastic process (Xt) is a hidden Markov process and if Benjamin Braun yes to infer a corresponding parametrization. So far only University of Kentucky partial answers to either the decision or the inference part [email protected] have been given all of which depend on further assumptions on the processes. Here we present a full, general solution for binary-valued hidden Markov processes. Our approach MS49 is rooted in algebraic statistics hence geometric in nature. Lattice Points in Polyhedra and the Summation We demonstrate that the algebraic varieties which describe Method for Mixed-Integer Optimization the probability distributions associated with binary-valued hidden Markov processes are zero sets of determinantal We present a new fully polynomial-time approximation equations which draws a connection to well-studied objects scheme for the problem of optimizing non-convex polyno- from algebra. As a consequence, our solution provides im- mial functions over the mixed-integer points of a polytope mediate algorithmic access where tests come in form of of fixed dimension [J.A. De Loera, R. Hemmecke, M. Kppe, elementary (linear) algebraic routines. R. Weismantel, FPTAS for optimizing polynomials over the mixed-integer points of polytopes in fixed dimension, Alexander Schoenhuth Math. Prog. Ser. A 118 (2008), 273–290]. The algo- IPAM and rithm also extends to a class of problems in varying di- CWI Amsterdam mension. This is the culmination of an effort to extend the [email protected] efficient theory of discrete generating functions of lattice points in polyhedra to the mixed-integer case in a practi- cal way, without using discretization, and to bring it to use in optimization. Matthias Koeppe AG11 Abstracts 75

Dept. of Mathematics Velleda Baldoni Univ. of California, Davis U. Roma Tor Vergata [email protected] [email protected]

Nicole Berline Nicole Berline Centre de Math´ematiques Laurent Schwartz Ecole Polytechnique Ecole´ Polytechnique [email protected] [email protected] Jesus De Loera Mich`ele Vergne University of California, Davis Institut Math´ematique de Jussieu [email protected] Universit´e Paris 7 Diderot [email protected] Matthias Koeppe Dept. of Mathematics Univ. of California, Davis MS49 [email protected] Vector Partition Functions and Applications to Lie Representation Theory MS50 I Given a finite set of non-zero integral vectors with non- Expected Complexity of Bisection Methods for γ negative coordinates, and a vector with coordinates Real Solving (γ1,...,γn), the vector partition function PI (γ)isthenum- ber of ways we can split γ as an integral sum with non- We examine degree-d polynomials with iid coefficients un- negative coefficients of the vectors in I. We formulate an der two zero-mean normal  distributions, such that the i-th algorithm for computing the vector partition function as coefficient has variance d ,or1/i!. The expected bit com- γ ,...,γ i quasipolynomials in ( 1 n) over a finite set of combi- plexity of the Sturm solver is O∗(rd2τ), where r bounds the natorial chambers and demonstrate an on-line implemen- number of real roots and τ the coefficient bitsize. For Bern- tation. We explain some problems from Lie representation stein polynomials with iid coefficients of moderate standard theory motivating this study. √ deviation, we show E[r]= 2d ± O(1). Todor Milev Ioannis Z. Emiris Eduard Czech Center National and Kapodistrian University of Athens [email protected] [email protected]

MS49 Andre’ Galligo Ehrhart Theory and Lecture Hall Polytopes University of Nice France For a sequence s =(s1,...,sn) of positive integers, define [email protected] s-lecture hall polytope {x ∈ Rn| ≤ x1 ≤ the to be the set 0 s1 x2 ≤ ...≤ xn } Elias Tsigaridas s2 sn . We prove a theorem relating the Ehrhart series of the s-lecture hall polytope to ascent statistics on University of Aarhus s-inversion sequences. We show how the theorem can be [email protected] refined and specialized to yield known results and derive new ones. This includes joint work with Michael Schuster, Gopal Viswanathan, and Thomas Pensyl. MS50 Hybrid Method for Solving Bivariate Polynomial Carla D. Savage System North Carolina State Univ Department of Computer Science Many problems in science and engineering can be reduced [email protected] to that of solving bivariate polynomial systems of equa- tions. Intensive research efforts have yielded reliable sym- bolic methods and efficient numeric methods. We com- MS49 bine the advantages of symbolic methods (in particular An Efficient Calculation of the Top Ehrhart Coef- resultant) and numeric methods (in particular Hansen- ficients for the Knapsack. Sengupta operator with slope form). The resulting method can reliably approximate all the solutions more efficiently Let A be a sequence of N + 1 positive integers. For t than the other (symbolic, numeric) methods in most cases. an integer, let E(A)(t) be the number of solutions in non  +1 negative integers x of the equation N A x = t. Then Hoon Hong  i i=1 i i North Carolina State University E t N E t ti E t (A)( )= i=0 i( ) where i( ) is a periodic function [email protected] of t. Fixing k, we show that the residue theorem in one variable, and the signed unimodular decomposition of a cone, leads to an algorithm of polynomial complexity to MS50 compute the top k-coefficients of E(A)(t). Accurate Path Tracking Michele Vergne Tracking solution paths defined by a family of polynomial Inst. Math. Jussieu systems is a basic operation for many algorithms in numer- [email protected] ical algebraic geometry. Version 2.3.55 of PHCpack incor- porated the QD-2.3.9, a software library of Y. Hida, X.S. 76 AG11 Abstracts

Li, and D.H. Bailey for quad double arithmetic. In this MS51 talk we will explain the adaptive use of double double and Hom4ps in Parallel quad double arithmetic in the path trackers of PHCpack. When solving systems of polynomial equations by the poly- Jan Verschelde hedral homotopy methods, while the path tracing part is Department of Mathematics, Statistics and Computer naturally parallel, the mixed cell computation which pro- Science vides the starting points for the homotopy paths is highly University of Illinois at Chicago serial. Recently, we have successfully reformulated the [email protected] mixed cell computation and achieved very high speed-ups in parallel. In this talk, we will report our most up- Genady Yoffe dated parallel implementation of both path tracing and Dept. of Mathematics, Statistics, and CS mixed cell computation algorithms. The parallel imple- University of Illinois, Chicago mentations are designed for a variety of parallel architec- gyoff[email protected] tures such as multi-core processors, symmetric multipro- cessors, nonuniform memory access architectures, clusters, distributed computers, and the exciting new technology of MS50 general purpose GPUs. An Efficient Exact Numerical Algorithm for Iso- topic Approximation of NonSingular Surfaces Tien-Yien Li Department of Mathematics We describe a subdivision algorithm for computing an Michigan State University isotopic -approximation of a nonsingular surface f −1(0) [email protected] where f : R3 → R is a C1 function. Our algorithm (called Cxyz) is based on numerical predicates which are easy to Tianran Chen implement and does not suffer from implementation gaps. Michigan State University Cxyz combines the technique of non-local isotopy from [email protected] Plantinga-Vegter (2004) and parametrizability from Sny- der (1996). Our implementation shows that this is more efficient than either approach separately. The correctness MS51 of Cxyz is non-trivial. Real Solutions and Numerical Algebraic Geometry Chee K. Yap In many applications, the real solutions to polynomial sys- New York University tems are the ones of interest. This talk will consider using Courant Institute classical approaches with methods in numerical algebraic [email protected] geometry to compute real solutions of a given polynomial system. Examples arising from applications will be pre- Long Lin sented to demonstrate the practicality of the methods. Courant Institute, NYU [email protected] Jonathan Hauenstein Texas A&M University [email protected] MS51 Polyhedral Methods For Positive Dimensional So- Charles Wampler lution Sets General Motors Research Laboratories, Enterprise Systems Lab We present a polyhedral algorithm to manipulate positive 30500 Mound Road, Warren, MI 48090-9055, USA dimensional solution sets. Using facet normals to Newton [email protected] polytopes as pretropisms, we zero in at the first two terms of a Puiseux series expansion. We illustrate how this poly- hedral algorithm gives insight into the structure of a tropi- MS51 cal prevariety. This polyhedral algorithm is well suited for Multithreaded Newton’s Method and Path Track- exploitation of symmetry, when it arises in systems of poly- ing nomials. Initial form systems with pretropisms in the same group orbit are solved only once, allowing for a systematic We have developed a multithreaded implementation of filtration of redundant data. The computational results Newtons method on a multicore workstation in a static will be illustrated on cyclic n-roots polynomial systems. memory allocation multiprecision environment using the quad double arithmetic in the QD-2.3.9. library. Further- Danko Adrovic more, based on this implementation, we obtained a com- Department of Mathematics, Statistics, and Computer plete parallel version of a path tracker. In my talk I will Science report on the details of work load distribution among the University of Illinois, Chicago threads and of synchronization in our parallel implemen- [email protected] tation. I will also discuss to what extent our use of multi- threading compensates the overhead caused by employing Jan Verschelde higher precision arithmetic, when tracking a solution path. Department of Mathematics, Statistics and Computer Science University of Illinois at Chicago Genady Yoffe [email protected] Dept. of Mathematics, Statistics, and CS University of Illinois, Chicago gyoff[email protected] AG11 Abstracts 77

Jan Verschelde able attention lately. A subclass of the latter is the Sys- Department of Mathematics, Statistics and Computer tematic Authentication Codes (SACs). Several types of Science SACs have appeared constructed by using highly nonlin- University of Illinois at Chicago ear functions over finite fields or non-degenerated and ra- [email protected] tional functions on a Galois ring. In this talk a class of bent functions over a Galois ring of characteristic p2 is in- troduced and based on these functions a class of SACs is MS52 given. These SAC’s generalize some appearing in the liter- AG Codes for Secure Multiparty Computation ature for finite fields. Chen and Cramer (Crypto 2006) use algebraic curves to Horacio Tapia-Recillas construct linear secret sharing schemes that are suitable UAM Mexico for secure multiparty computation. We describe the con- [email protected] struction and we analyse its main properties in terms of the geometry of the curve. Juan Carlos Ku-Cauich UAM-Mexico Iwan Duursma [email protected] University of Illinois at Urbana-Champaign Department of Mathematics [email protected] Claude Carlet University of Paris 8 and INRIA INRIA Projet CODES MS52 [email protected] Groebner Bases and Linear Codes Many classes of good codes have been constructed from MS52 algebraic curves. These constructions, however, have re- Complexity in Coding Theory strictions on the alphabet size (say q has to be a square) and on the block length (which depends on the number of We will give an introduction to the complexity issues in rational points). We are interested in a more general ap- coding theory, both for general codes and for specific codes proach that avoids these restrictions. More precisely, given such as algebraic geometric codes of low genus. n any prime power q and any block length ,FarrandGao Daqing Wan (2003) shows how Groebner basis theory can be used to University of California, Irvine construct linear codes of length n over GF(q) from any set [email protected] of n points from an affine space. In this talk, we shall dis- cuss various computational problems related to this class of codes, including decoding via Groebner bases. MS53 Shuhong Gao Some Examples of Non-finitely Generated Sym(N)- Clemson University invariant Ideals [email protected] We are concerned with Sym(N)-invariant (or similar) ide- als of subalgebras of polynomial algebras in countably in- MS52 finitely many variables. These ideals arose recently from applications to statistics and chemistry. Aschenbrenner, Bounding the Number of Points on a Curve using Hillar, Sullivant and Draisma have proved that in certain a Generalization of Weierstrass Semigroups and an subalgebras all Sym(N)-invariant ideals are finitely gener- Application to Toric Codes ated (as such). In this talk, we will present some examples Sym N In this talk we use techniques from coding theory to derive of non-finitely generated ( )-invariant ideals in some upper bounds for the number of rational places of an alge- subalgebras closed to ones mentioned above. braic curve defined over a finite field. The used techniques Alexei Krasilnikov yield upper bounds if the (generalized) Weierstrass semi- n Departamento de Matem´atica group for an -tuple of places is known. This sometimes Universidade de Braslia, Braslia - DF, 70910-900, Brazil enables one to get an upper bound for the number of ra- [email protected] tional places for families of function fields. We consider an application to toric codes. MS53 Diego Ruano Aalborg University Tensors of Bounded Rank are Defined in Bounded [email protected] Degree An element of a p-fold tensor product of vector spaces has Peter Beelen border rank k if it may be approximated by a sum of k Technical University of Denmark (pure) tensors. The variety of tensors of a given border [email protected] rank is a classical object of algebraic geometry (a higher secant variety) with important modern applications in var- ious diverse areas (algebraic geometry, statistics, complex- MS52 ity theory). Except for small k no system of defining equa- Bent Functions on a Galois Ring and Systematic tions for this variety is known. We show that for each k Authentication Codes there exists a universal degree bound d such that for each p there are defining equations of degree at most d for the Authentication codes with secrecy and without secrecy, variety of p-dimensional tensors of border rank k. The cru- produced by different techniques, have received consider- 78 AG11 Abstracts

cial ingredient is to remove the dependence on p by passing and Sturmfels, and the result of Kanev. to the limit p →∞and using the action of some natural group of symmetries. In fact, we show that up to these Claudiu Raicu symmetries, there are finitely many equations defining the UC Berkeley / variety for all p-dimensional tensors as long as p is large Princeton enough. [email protected]

Jan Draisma T.U.Eindhoven, The Netherlands MS53 Department of Mathematics Twisted Commutative Algebras and Delta-modules [email protected] Twisted commutative algebras and Δ-modules are two Jochen Kuttler kinds of algebraic structures that can be interpreted as Department of Mathematical and Statistical Sciences having infinitely many variables. I will explain what they University of Alberta are and state two finiteness theorems about them: the first [email protected] is a noetherianity result, the second a rationality result for Hilbert series. If time permits, I will explain how these re- sults can be applied to study the syzygies of certain families MS53 of varieties. Associative Algebras and Letterplace Embedding Andrew Snowden We introduce an algebra embedding ι : KX→S from MIT the free associative algebra KX generated by a finite or [email protected] countable set X into the skew monoid ring S = P ∗ Σ, P K X × N σ where = [ ]andΣ= is the monoid, gener- MS54 ated by a suitable endomorphism σ : P → P .Wepresent a general Gr¨obner bases theory for graded two-sided ideals The Convex Geometry of Linear Inverse Problems of the graded algebra S = Pσi,whereP = K[Y ]is i σ P → P Building on the success of generalizing compressed sens- any commutative polynomial ring and : is an ing to low-rank matrix recovery, we extend the catalog of abstract endomorphism satisfying two explicit conditions. structures that can be recovered from partial information. We obtain a bijective correspondence, preserving Gr¨obner We describe algorithms to decompose signals into sums bases, between graded Σ-invariant ideals of P and a class S of atomic signals from a simple, not necessarily discrete, of graded two-sided ideals of . As an application we show, set. These algorithms are derived in a convex optimization that up to a given degree, Gr¨obner bases of finitely gener- framework that generalizes previous methods based on l1- ated graded ordinary difference ideals can be computed by S norm and nuclear norm minimization. We discuss general the proposed methods in in a finite number of steps. recovery guarantees for our approach and several example Viktor Levandovskyy applications. Lehstuhl D fuer Mathematik, RWTH Aachen Venkat Chandrasekaran [email protected] Massachusetts Institute of Technology [email protected] RobertoLaScala Dipartimento di Mathematica Benjamin Recht Universita degli studi di Bari University of Wisconsin – Madison [email protected] [email protected]

MS53 Pablo A. Parrilo, Alan Willsky Secant Varieties of Segre-Veronese Varieties Massachusetts Institute of Technology [email protected], [email protected] Secant varieties of Segre and Veronese varieties are classi- cal objects that go back to the Italian school in the nine- teen century. Surprisingly, very little is known about their MS54 equations. Inspired by experiments related to algebraic Extended BIC and Bayesian Marginal Likelihood statistics, Garcia, Stillman and Sturmfels gave a conjec- in Sparse Graphical Models tural description of the generators of the ideal of the secant line variety Sec(X) of a Segre variety X. This generalizes We consider the problem of recovering sparse graphical the familiar result which states that matrices of rank two structure in the setting of Gaussian or discrete graphical are defined by the vanishing of their 3 × 3minors.For modeling. While the total number of possible sparse mod- a Veronese variety X, it was known by work of Kanev els is often too large for exhaustive search, information crit- that the ideal of Sec(X) is generated in degree three by era may be used to select a sparse model from a reduced minors of catalecticant matrices. I will introduce a new set of candidate models (for instance, the “path” of models technique for studying the equations of the secant varieties selected by a parameterized procedure such as the graphi- of Segre-Veronese varieties, based on the usual representa- cal lasso). Modern applications of graphical models often tion theoretic approach to this problem. I will explain how have a number of nodes that is comparable to or greater this technique applies to show that for X a Segre-Veronese than the number of data points observed. In our treat- variety, the ideal of Sec(X) is generated in degree three ment of an extended Bayesian information criterion (BIC) we thus consider asymptotic scenarios in which the num- by minors of matrices of flattenings, and to give a descrip- p n tion of the decomposition into irreducible representations ber of nodes togrowsasthesamplesize increases. Our of the homogeneous coordinate ring of Sec(X). This will work examines the consistency properties of the extended recover as special cases the conjecture of Garcia, Stillman BIC as well as its relationship to the marginal likelihood AG11 Abstracts 79

arising in fully Bayesian approaches. University of Waterloo Dept of Combinatorics & Optimization Rina Foygel, Mathias Drton [email protected] Department of Statistics University of Chicago [email protected], [email protected] MS55 Continuous Amortization for the Complexity of Adaptive Subdivision Methods MS54 Diagonal/Low-Rank Decomposition and Correla- Adaptive subdivision algorithms subdivide more times near tion Matrices challenging features and fewer times elsewhere. Computa- tions of the size of the subdivision tree should consider Suppose an n × n matrix X is the sum of a diagonal and a adaptivity: e.g., bounds on the maximum depth do not psd low-rank matrix. Decomposing X into these unknown reflect situations where the tree has a few deep paths, but constituents has applications in statistics, signal process- remains modest in overall size. I use the new analysis tech- ing, and elsewhere. We give a simple condition on the col- nique of continuous amortization to show that the tree size umn space of the low rank matrix that ensures a tractible of a simple, evaluation-based root isolation algorithm is convex program can correctly decompose X. Our anal- nearly optimal. ysis highlights connections between this problem and the structure of the set of correlation matrices. Michael A. Burr Fordham University James Saunderson [email protected] Massachusetts Institute of Technology [email protected] Felix Krahmer Hausdorff Center for Mathematics MS54 Universitat Bonn felix.krahmer@hausdorff-center.uni-bonn.de Chromosome Packing in Cell Nucleus

During most of the cell cycle each chromosome occupies Chee K. Yap a roughly spherical domain called a chromosome territory. New York University Chromosome territories can overlap and their radial and Courant Institute relative positions are non-random and similar among simi- [email protected] lar cell types. A chromosome arrangement can be viewed as a packing of overlapping spheres of various sizes inside an ellipsoid, the cell nucleus. We present a non-convex model MS55 for chromosome arrangements together with an alternat- Patterns in Roots of the Derivatives of Random ing minimization algorithm. Using this model, we simulate Univariate Polynomials chromosome arrangements and study the resulting number and volume of internal ’holes’, which make chromosomes I will first recall old and recent results on distributions of deep inside accessible to regulatory factors. roots of random polynomials and eigenvalues of random matrices. Then I will present original observations of pat- Caroline Uhler terns of roots of the derivatives of random polynomials. I Department of Statistics will set some conjectures enforced by experiments and out- UC Berkeley line proofs of some claims. This can be applied to design [email protected] root isolation algorithms. Andre Galligo MS54 Mathematics Department Finding Approximately Rank-One Submatrices lUniversit´e Nice Sophia Antipolis [email protected] with the Nuclear Norm and l1-Norm

We propose a tractable method to find large approximately rank-one submatrices which can be used in nonnegative MS55 matrix factorization. We apply a proximal point algo- Subadditivity in Polynomial Real Root Isolation rithm to solve the resulting optimization problem which Isac Schoenberg’s theory of variation-diminishing linear is formulated with the nuclear norm and l1-norm. We re- port numerical results on realistic datasets of decomposable transformations, published in 1930, is used to prove the fol- bitmap images and nearly separable corpus. lowing subadditivity property for all univariate real poly- nomials A, Xuan Vinh Doan A ◦ A ≤ A . C&O Department var(T( )) + var((T R)( )) var( ) University of Waterloo Here, T, R and var are, respectively, the translation by 1, [email protected] the reciprocal transformation, and the number of sign vari- ations. The subadditivity property is useful in the imple- Kim-Chuan Toh mentation and analysis of the continued fractions method National University of Singapore for polynomial real root isolation. This research is joint Department of Mathematics and Singapore-MIT Alliance work with G. E. Collins. [email protected] Werner Krandick Stephen A. Vavasis Drexel University [email protected] 80 AG11 Abstracts

MS55 ture Subdivision Methods for the Topology and Ar- rangement of Semi-algebraic Curves and Surfaces The Monotone Secant Conjecture poses a rich class of poly- nomial systems, all of whose solutions are real. These Computing the topology and arrangement of algebraic systems come from the Schubert calculus on ?ag mani- curves and surfaces appears in many geometric mod- folds, and the Monotone Secant Conjecture is a compelling elling problems, such as surface-surface intersection, self- generalization of the Shapiro Conjecture for Grassmanni- intersection, ... In the presentation, we will describe subdi- ans (Theorem of Mukhin, Tarasov, and Varchenko). We vision methods which exploit information on the boundary present the Monotone Secant Conjecture as a generaliza- of regions and which only require the isolation of character- tion of the Shapiro Conjecture and explain the massive istic points. We will show how topological degree compu- computational evidence in its favor. tation can help analysing the number of branches at singu- lar points. Combining regularity criterion with subdivision Nickolas J. Hein strategies, this yields a complete algorithm for comput- Texas A&M University ing the topology of (singular) algebraic curves. Extension [email protected] of this approach to arrangement computation of curves or surfaces will be described. Experimentation with the PP1 algebraic-geometric modeller AXEL will shortly be demon- strated. Dual Space Computations for Homotopy Continu- ation Using H-Bases Bernard Mourrain INRIA Sophia Antipolis Using homotopy continuation methods, certain systems [email protected] end up with a number of paths tending toward infinity. Us- ing dual space computations through H-bases, our goal is to pre-condition these systems. By moving to the dual space PP1 and using H-bases, we can cut out many of these extraneous The Geometry of Tree Reconstruction paths. When we come back from the dual space, we have a system with fewer paths tracking to infinity thereby saving Tree agglomeration methods such as the Neighbor-Joining time and computation. We examine how many times to and UPGMA algorithms continue to play a significant role efficiently implement this for optimal time and results. in phylogenetic research. We use polyhedral geometry to analyze the natural subdivision of Euclidean space induced Steven L. Ihde by classifying the input vectors according to tree topologies Colorado State University returned by an algorithm. We also use lattice theory to [email protected] analyze disparities between observed and predicted shape statistics on reconstructed trees. PP1 Ruth E. Davidson, Seth Sullivant Markov Bases for the Analysis of Partially Ranked North Carolina State University Data [email protected], [email protected] Suppose survey respondents have been asked to give a com- plete ranking of a collection of items. Diaconis and Sturm- PP1 fels, and Diaconis and Eriksson, have shown how Markov The Maximum Likelihood Degree for the Random bases can be used to better understand relationships be- Effects Model tween natural summary statistics associated with the re- sulting fully ranked data. In this talk, we show how their Given a statistical model, the maximum likelihood degree ideas can be fruitfully extended to situations in which re- (ML degree) is the degree of the variety defined by the spondents are asked to provide rankings of only a subset likelihood equations. It measures the algebraic complexity of the items. of the computation to find model parameters that best ex- plain a given data point. In this poster, we give an explicit Michael E. Orrison, Ann Johnston formula for the ML degree of the analysis of variance model Harvey Mudd College with random effects. This is joint work with Mathias Drton [email protected], ann k [email protected] and Sonja Petrovic. Michael Hansen Elizabeth Gross N/A University of Illinois at Chicago [email protected] [email protected]

Mathias Drton PP1 Department of Statistics Algebraic Ecological Inference University of Chicago [email protected] Ecological inference is the process of inferring individual level behavior from aggregate data. The problem of ecolog- ical inference occurs frequently in many applied sciences. Sonja Petrovic In the present work, we present an algebraic solution to the University of Illinois, Chicago problem of ecological inference using tools from computa- [email protected] tional algebra. This method can handle multi-dimensional contingency tables, deterministically respects bounds and PP1 can incorporate information from many different sources. We illustrate the method by a few examples and provide Unexpected Reality: the Monotone-Secant Conjec- AG11 Abstracts 81

Rcode. Systems Biology

Vishesh Karwa We present a new modeling framework for gene regulatory Penn State University networks that incorporates state dependent delays and that [email protected] is able to capture the cell-to-cell variability. We present this framework in the context of polynomial dynamical sys- tems where we can use computational algebra to analyze PP1 our model. The state dependent delays represent the time The Convex Hull of Circles delays of activation and degradation. One of the new fea- tures of this framework is that it allows a finer analysis of We describe the boundaries of the convex hulls of perhaps discrete models and the possibility to simulate cell popu- the simplest compact curves in three-dimensional space, lations. We applied our methods to one of the best known namely pairs of circles. We show that the edge surface stochastic regulatory networks, that is involved in control- of such a convex hull is in general an irrational ruled sur- ling the outcome of lambda phage infection of bacteria. face whose ruling forms a (2,2)-curve, and conjecture that any real (2,2)-curve can arise from the edge surface of the David Murrugarra convex hull of two circles. Virginia Bioinformatics Institute Virginia Tech Tina Mai [email protected] Department of Mathematics, Texas A&M University College Station, TX 77843-3368 Reinhard Laubenbacher [email protected] Virginia Bioinformatics institute 1880 Pratt Drive Frank Sottile [email protected] Texas A&M University [email protected] PP1 An Implementation of Parameter Homotopies and PP1 their Applications Developing Water Loss Prevention Robot System Geometrical Design There are several implementations of homotopy continuation-based polynomial system solvers, This research describes the geometrical design of an in-pipe some of which allow for so-called parameter homotopies robot system able to redevelop fresh water pipes of different between polynomial systems that differ only in coefficients. diameters. The robots geometry is calculated in conjunc- Some applications from science and engineering require the tion of its working environment, the size of the pipes, and repeated solution of polynomial systems with slightly dif- its physical properties, such as weight and required force ferent coefficients. The new software package described in to move and redevelop the pipe. this poster is a parallelized implementation of parameter homotopies, making use of Bertini (D. Bates, J. Hauen- Luis A. Mateos, Markus Vincze stein, A. Sommese, C. Wampler). An example considering Automation and Control Institute (ACIN) bistability in biochemical kinetic reactions is provided to Vienna University of Technology show the usefulness of these methods. This is joint work [email protected], [email protected] with Dan Bates and Dan Brake.

Matt Niemerg PP1 Colorado State University Perturbation Analysis to Design a Robust Decou- Fort Collins, CO pling Geometric Technique in Linear Multi-Input [email protected] Multi-Output Systems

The linear state-feedback control is one of the basic and PP1 fundamental part of the linear control theory. Eigenvectors Computing the Variety for K-Algebra Homomor- are used through the singular value decomposition (svd) phisms in robust analysis and synthesis of robust controllers in multi-input multi-output systems. In this paper the prob- Let k be an algebraically closed field. Let A and B be ar- lem of robustness of a decoupling controller is analyzed. bitrary commutative (unitary) k-algebras. Assume V ⊂ A In particular, it is shown that the robustness of a decou- and W ⊂ B are finite dimensional k-linear subspaces. De- pling controller is achieved using subspaces closer to the note the subalgebras of A and B generated by V and W eigenvectors associated to the system. The analysis of the as A(V )andB(W ). Then the set RHom(A, B, V, W )of robustness is based on a perturbation method and a syn- k-algebra homomorphisms f : A(V ) → B(W ) such that thesis of a controller is shown. An application concerning f(V ) ⊂ W is an affine k-variety in a natural way. The an electrical motor is shown. structure of the proof of this claim suggests an algorithm could be developed to allow software to calculate the affine Paolo Mercorelli variety Hom(A, B, V, W ). The goal of my research is to de- Ostfalia University of Applied Sciences velop software capable of doing this calculation and use this [email protected] software to compute some classical algebras (e.g. group al- gebras, monomial algebras etc).

PP1 Jon Yaggie A Stochastic Framework for Discrete Models in University of Illinois Chicago [email protected]