An Approach to Optimal Discretization of Continuous Real Random Variables with Application to Machine Learning

Total Page:16

File Type:pdf, Size:1020Kb

An Approach to Optimal Discretization of Continuous Real Random Variables with Application to Machine Learning An Approach to Optimal Discretization of Continuous Real Random Variables with Application to Machine Learning Carlos Antonio Pinzón Henao Student identification: 8945263 Director: Camilo Rocha, PhD Co-director: Jorge Finke, PhD Engineering and sciences Master in Engineering - Emphasis on Computer Science Pontificia Universidad Javeriana Cali, Cali, Colombia, October 2019. 1 Contents 1 Introduction6 2 Preliminaries 10 2.1 Random variables of interest..... 11 2.2 Entropy and redundancy....... 12 2.2.1 Shannon entropy........ 12 2.2.2 Differential entropy and LDDP 13 2.2.3 Rényi entropy......... 15 2.2.4 Redundancy.......... 16 2.3 Discretization.............. 17 2.3.1 Digitalization.......... 17 2.3.2 n-partitions........... 18 2 2.3.3 Distortion and entropy.... 21 2.3.4 Discretization functions... 21 2.3.5 Redundancy and scaled distortion 23 3 Optimal quantizers 25 3.1 Context................. 25 3.2 Formal definition............ 26 3.3 Application in sensorics........ 27 4 Uses in machine learning 30 4.1 Preprocessing.............. 31 4.2 Intra-class variance split criterion.. 33 4.2.1 Construction of decision trees 34 4.2.2 Min-variance split criterion. 37 4.2.3 Non-uniqueness........ 43 4.2.4 Non-optimality......... 45 4.3 Discussion................ 47 3 5 A real analysis approach 48 5.1 Formal derivation........... 48 5.1.1 Deriving Theorem 10..... 49 5.1.2 Deriving Theorem 17..... 57 5.1.3 Deriving Theorem 19..... 63 5.1.4 Optimality theorems..... 67 5.2 Summary of formulas and quantizers 73 5.3 Some generalizations.......... 74 5.4 An open problem............ 75 6 Conclusion and future work 77 Bibliography 78 A Optimal quantizer in information theory 81 A.1 Rate-distortion function........ 81 A.2 Optimal quantizers in the rate- distortion plane............. 83 4 Acknowledgment I would like to thank all the people that made this thesis possible, including Dr. Camilo Rocha and Dr. Jorge Finke for their trust and guidance, all the supervisors that agreed to review my work, the professors at the University whose teachings and advises helped me directly or indirectly throughout this period, including Isabel García, Carlos Ramírez and Frank Valencia, my friend Juan Pablo Berón for providing insight into some proofs, and my friends and family, specialy my future wife Adriana, for their constant support. Without their help, it would not have been possible to conclude this thesis successfully. In addition, I thank the growing Mathematics Stack Exchange community for having provided the sketch for two of the proofs presented in this document: question 3339209 for Theorem9 and question 154721 for Theorem 12. 5 Chapter 1 Introduction This document explores some discretization methods for continuous real random variables, their performance and some of their applications. Particularly, this work borrows a well-studied discretization method from the signal processing community, called the minimal distortion high resolution quantizer1, and makes use of it in machine learning with the purpose of illustrating how and when it can improve typically used techniques and methodologies. Discretization is the process of converting a continuous variable into a discrete one. It is useful because unlike the continuous input, the discrete output can be encoded and treated with discrete methods and tools, which includes computers and microcontrollers. Therefore, it occurrs in all modern sensor systems. It is used also in data analysis, either for a final categorization of the data or internally as a bridge between the continuous nature of the data and the discrete nature of some algorithms. Despite its advantages, discretization introduces forcibly an error because the discrete output is unable to mimic the continuous input with entire precision. This error can 1Herbert Gish and John Pierce. “Asymptotically efficient quantizing”. In: IEEE Transactions on Information Theory 14.5 (1968), pp. 676–683. 6 be forced to be very small by using properly designed discretization methods or at the expense of using more bits in the codification. Discretization has been deeply studied under the term quantization, especially in the signal processing community, which not only proposed the minimal distortion high resolution quantizer2, but also an iterative method, called Lloyd method, for computing optimal quantizers for any resolution level3 and several generalized versions of this algorithm that support several dimensions and are therefore called vector quantizers4. These algorithms, have a tremendous advantage over the high resolution quantizer, namely that they operate directly from samples and do not require the distribution to be known or estimated previously. Furthermore, the book Foundations of Quantization for Probability Distributions5 due to Graf and Lushgy makes an extense review of quantization methods and provides common notation and a rigorous theory, using advanced topics in mathematics, for supporting them. The book embraces several types of quantizers, including one dimensional optimal quantizers, one dimensional high resolution optimal quantizers and vector quantizers. This work explores just one of these types of quantizers using undergraduate mathematics so as to be readable and useful for a broader less experienced audience. It focuses exclusively on the discretization of one-dimensional continous real random variables X for which a Riemann-integrable probability density function with domain [a, b] exists, for reals a < b. These constraints, which might seem very strong from a mathematical point of view, are 2Herbert Gish and John Pierce. “Asymptotically efficient quantizing”. In: IEEE Transactions on Information Theory 14.5 (1968), pp. 676–683. 3Stuart Lloyd. “Least squares quantization in PCM”. in: IEEE transactions on information theory 28.2 (1982), pp. 129–137. 4Yoseph Linde, Andres Buzo, and Robert Gray. “An algorithm for vector quantizer design”. In: IEEE Transactions on communications 28.1 (1980), pp. 84–95, Robert Gray. “Vector quantization”. In: IEEE Assp Magazine 1.2 (1984), pp. 4–29. 5Siegfried Graf and Harald Luschgy. Foundations of quantization for probability distributions. Springer, 2007. 7 common in engineering applications and allow the study to be reduced to undergraduate mathematics. The analysis focuses on the following quantizers, especially on the optimal quantizer. 1. Equal width quantizer, or uniform quantizer: it is the most simple quantizer and it is used in most sensors. It consists of dividing the input interval into n bins of equal width. 2. Equal frequency quantizer: it consists of dividing the input interval into n bins of equal area under the curve f. As a consequence, it maximizes the entropy of the random output, which can be understood less precisely as its diversity. 3. Minimal distortion high resolution quantizer, called optimal quantizer throughout the document: it minimizes the error between the input and the output when n is sufficiently large. It is explained in detail in Chapter3. 4. Recursive binary quantizer: used in binary decision trees in machine learning. As explained in Chapter4, it is constructed with a recursive and greedy procedure. Namely, it divides the interval into two by means of a minimal distortion quantizer with n = 2, and repeats the procedure on each subinterval yielding 4, 8, and in general 2k parts. The main contributions of this dissertation can be summarized as follows: 1. It brings the optimal quantizer to the attention of the machine learning community as an alternative for the typical equal width and equal frequency quantizers when the input distribution is known or well estimated, especially targeting members with few knowledge in advanced mathematics. 2. It illustrates with some examples how and when can the optimal quantizer replace typical non-iterative techniques in machine learning for discretizing a dataset, and how and when can it be used as a split criterion for decision trees. 3. It presents real-analysis proofs of the optimality and uniqueness of the optimal quantizer under the constraints of being bounded and having 8 a Riemann-integrable density. In contrast to existing proofs6, these are comprehensible for the majority of undergraduate engineers because it does not use measure theory. 4. It summarizes the generalizations of these results to other distance metrics and provides insight on how to generalize as well for different entropy functions. This document is organized as follows. Chapter1 is the introduction. Chapter2 provides definitions that are used along the document and introduces the reader into discretizations with some plots and examples. Chapter3 reviews the optimal quantizer and presents its application in sensorics. Chapter4 compares the optimal quantizer against the others in machine learning. More precisely, the chapter compares the performance of some predictors that are fed with the output of the discretizators, and provides a split criterion for decision trees based on the optimal quantizer. Chapter5 uses real analysis to prove that the optimal quantizer is indeed optimal and unique. It also provides a generalization of this quantizer and an open problem that emerged naturally from the generalizations. Chapter6 concludes the document with a short summary. AppendixA explores the optimal quantizer from an information theoretical perspective. In particular, some plots of optimal quantizers on the rate-distortion plane are provided. 6Siegfried Graf and Harald Luschgy. Foundations of quantization for probability distributions. Springer, 2007, Herbert Gish
Recommended publications
  • A Modern History of Probability Theory
    A Modern History of Probability Theory Kevin H. Knuth Depts. of Physics and Informatics University at Albany (SUNY) Albany NY USA 4/29/2016 Knuth - Bayes Forum 1 A Modern History of Probability Theory Kevin H. Knuth Depts. of Physics and Informatics University at Albany (SUNY) Albany NY USA 4/29/2016 Knuth - Bayes Forum 2 A Long History The History of Probability Theory, Anthony J.M. Garrett MaxEnt 1997, pp. 223-238. Hájek, Alan, "Interpretations of Probability", The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/win2012/entries/probability-interpret/>. 4/29/2016 Knuth - Bayes Forum 3 … la théorie des probabilités n'est, au fond, que le bon sens réduit au calcul … … the theory of probabilities is basically just common sense reduced to calculation … Pierre Simon de Laplace Théorie Analytique des Probabilités 4/29/2016 Knuth - Bayes Forum 4 Taken from Harold Jeffreys “Theory of Probability” 4/29/2016 Knuth - Bayes Forum 5 The terms certain and probable describe the various degrees of rational belief about a proposition which different amounts of knowledge authorise us to entertain. All propositions are true or false, but the knowledge we have of them depends on our circumstances; and while it is often convenient to speak of propositions as certain or probable, this expresses strictly a relationship in which they stand to a corpus of knowledge, actual or hypothetical, and not a characteristic of the propositions in themselves. A proposition is capable at the same time of varying degrees of this relationship, depending upon the knowledge to which it is related, so that it is without significance to call a John Maynard Keynes proposition probable unless we specify the knowledge to which we are relating it.
    [Show full text]
  • A Brief Overview of Probability Theory in Data Science by Geert
    A brief overview of probability theory in data science Geert Verdoolaege 1Department of Applied Physics, Ghent University, Ghent, Belgium 2Laboratory for Plasma Physics, Royal Military Academy (LPP–ERM/KMS), Brussels, Belgium Tutorial 3rd IAEA Technical Meeting on Fusion Data Processing, Validation and Analysis, 27-05-2019 Overview 1 Origins of probability 2 Frequentist methods and statistics 3 Principles of Bayesian probability theory 4 Monte Carlo computational methods 5 Applications Classification Regression analysis 6 Conclusions and references 2 Overview 1 Origins of probability 2 Frequentist methods and statistics 3 Principles of Bayesian probability theory 4 Monte Carlo computational methods 5 Applications Classification Regression analysis 6 Conclusions and references 3 Early history of probability Earliest traces in Western civilization: Jewish writings, Aristotle Notion of probability in law, based on evidence Usage in finance Usage and demonstration in gambling 4 Middle Ages World is knowable but uncertainty due to human ignorance William of Ockham: Ockham’s razor Probabilis: a supposedly ‘provable’ opinion Counting of authorities Later: degree of truth, a scale Quantification: Law, faith ! Bayesian notion Gaming ! frequentist notion 5 Quantification 17th century: Pascal, Fermat, Huygens Comparative testing of hypotheses Population statistics 1713: Ars Conjectandi by Jacob Bernoulli: Weak law of large numbers Principle of indifference De Moivre (1718): The Doctrine of Chances 6 Bayes and Laplace Paper by Thomas Bayes (1763): inversion
    [Show full text]
  • Maximum Entropy: the Universal Method for Inference
    MAXIMUM ENTROPY: THE UNIVERSAL METHOD FOR INFERENCE by Adom Giffin A Dissertation Submitted to the University at Albany, State University of New York in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy College of Arts & Sciences Department of Physics 2008 Abstract In this thesis we start by providing some detail regarding how we arrived at our present understanding of probabilities and how we manipulate them – the product and addition rules by Cox. We also discuss the modern view of entropy and how it relates to known entropies such as the thermodynamic entropy and the information entropy. Next, we show that Skilling's method of induction leads us to a unique general theory of inductive inference, the ME method and precisely how it is that other entropies such as those of Renyi or Tsallis are ruled out for problems of inference. We then explore the compatibility of Bayes and ME updating. After pointing out the distinction between Bayes' theorem and the Bayes' updating rule, we show that Bayes' rule is a special case of ME updating by translating information in the form of data into constraints that can be processed using ME. This implies that ME is capable of reproducing every aspect of orthodox Bayesian inference and proves the complete compatibility of Bayesian and entropy methods. We illustrated this by showing that ME can be used to derive two results traditionally in the domain of Bayesian statistics, Laplace's Succession rule and Jeffrey's conditioning rule. The realization that the ME method incorporates Bayes' rule as a special case allows us to go beyond Bayes' rule and to process both data and expected value constraints simultaneously.
    [Show full text]
  • UC Berkeley UC Berkeley Electronic Theses and Dissertations
    UC Berkeley UC Berkeley Electronic Theses and Dissertations Title Bounds on the Entropy of a Binary System with Known Mean and Pairwise Constraints Permalink https://escholarship.org/uc/item/1sx6w3qg Author Albanna, Badr Faisal Publication Date 2013 Peer reviewed|Thesis/dissertation eScholarship.org Powered by the California Digital Library University of California Bounds on the Entropy of a Binary System with Known Mean and Pairwise Constraints by Badr Faisal Albanna A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Physics in the Graduate Division of the University of California, Berkeley Committee in charge: Professor Michael R. DeWeese, Chair Professor Ahmet Yildiz Professor David Presti Fall 2013 Bounds on the Entropy of a Binary System with Known Mean and Pairwise Constraints Copyright 2013 by Badr Faisal Albanna 1 Abstract Bounds on the Entropy of a Binary System with Known Mean and Pairwise Constraints by Badr Faisal Albanna Doctor of Philosophy in Physics University of California, Berkeley Professor Michael R. DeWeese, Chair Maximum entropy models are increasingly being used to describe the collective activity of neural populations with measured mean neural activities and pairwise correlations, but the full space of probability distributions consistent with these constraints has not been explored. In this dissertation, I provide lower and upper bounds on the entropy for both the minimum and maximum entropy distributions over binary units with any fixed set of mean values and pairwise correlations, and we construct distributions for several relevant cases. Surprisingly, the minimum entropy solution has entropy scaling logarithmically with system size, unlike the possible linear behavior of the maximum entropy solution, for any set of first- and second-order statistics consistent with arbitrarily large systems.
    [Show full text]
  • Nature, Science, Bayes' Theorem, and the Whole of Reality
    Nature, Science, Bayes' Theorem, and the Whole of Reality Moorad Alexanian Department of Physics and Physical Oceanography University of North Carolina Wilmington Wilmington, NC 28403-5606 Abstract A fundamental problem in science is how to make logical inferences from scientific data. Mere data does not suffice since additional information is necessary to select a domain of models or hypotheses and thus determine the likelihood of each model or hypothesis. Thomas Bayes' Theorem relates the data and prior information to posterior probabilities associated with differing models or hypotheses and thus is useful in identifying the roles played by the known data and the assumed prior information when making inferences. Scientists, philosophers, and theologians accumulate knowledge when analyzing different aspects of reality and search for particular hypotheses or models to fit their respective subject matters. Of course, a main goal is then to integrate all kinds of knowledge into an all-encompassing worldview that would describe the whole of reality. A generous description of the whole of reality would span, in the order of complexity, from the purely physical to the supernatural. These two extreme aspects of reality are bridged by a nonphysical realm, which would include elements of life, man, consciousness, rationality, mental and mathematical abstractions, etc. An urgent problem in the theory of knowledge is what science is and what it is not. Albert Einstein's notion of science in terms of sense perception is refined by defining operationally the data that makes up the subject matter of science. It is shown, for instance, that theological considerations included in the prior information assumed by Isaac Newton is irrelevant in relating the data logically to the model or hypothesis.
    [Show full text]
  • Distribution Transformer Semantics for Bayesian Machine Learning
    Distribution Transformer Semantics for Bayesian Machine Learning Johannes Borgstrom¨ 1, Andrew D. Gordon1, Michael Greenberg2, James Margetson1, and Jurgen Van Gael3 1 Microsoft Research 2 University of Pennsylvania 3 Microsoft FUSE Labs Abstract. The Bayesian approach to machine learning amounts to inferring pos- terior distributions of random variables from a probabilistic model of how the variables are related (that is, a prior distribution) and a set of observations of vari- ables. There is a trend in machine learning towards expressing Bayesian models as probabilistic programs. As a foundation for this kind of programming, we pro- pose a core functional calculus with primitives for sampling prior distributions and observing variables. We define novel combinators for distribution transform- ers, based on theorems in measure theory, and use these to give a rigorous se- mantics to our core calculus. The original features of our semantics include its support for discrete, continuous, and hybrid distributions, and observations of zero-probability events. We compile our core language to a small imperative lan- guage that in addition to the distribution transformer semantics also has a straight- forward semantics via factor graphs, data structures that enable many efficient inference algorithms. We then use an existing inference engine for efficient ap- proximate inference of posterior marginal distributions, treating thousands of ob- servations per second for large instances of realistic models. 1 Introduction In the past 15 years, statistical machine learning has unified many seemingly unrelated methods through the Bayesian paradigm. With a solid understanding of the theoreti- cal foundations, advances in algorithms for inference, and numerous applications, the Bayesian paradigm is now the state of the art for learning from data.
    [Show full text]
  • Report from the Chair by Robert H
    HistoryN E W S L E T T E R of Physics A F O R U M O F T H E A M E R I C A N P H Y S I C A L S O C I E T Y • V O L U M E I X N O . 5 • F A L L 2 0 0 5 Report From The Chair by Robert H. Romer, Amherst College, Forum Chair 2005, the World Year of Physics, has been a good one for the The Forum sponsored several sessions of invited lectures at History Forum. I want to take advantage of this opportunity to the March meeting (in Los Angeles) and the April meeting (in describe some of FHP’s activities during recent months and to Tampa), which are more fully described elsewhere in this Newslet- look forward to the coming year. ter. At Los Angeles we had two invited sessions under the general The single most important forum event of 2005 was the pre- rubric of “Einstein and Friends.” At Tampa, we had a third such sentation of the fi rst Pais Prize in the History of Physics to Martin Einstein session, as well as a good session on “Quantum Optics Klein of Yale University. It was only shortly before the award Through the Lens of History” and then a fi nal series of talks on ceremony, at the Tampa meeting in April, that funding reached “The Rise of Megascience.” A new feature of our invited sessions the level at which this honor could be promoted from “Award” to this year is the “named lecture.” The purpose of naming a lecture “Prize.” We are all indebted to the many generous donors and to is to pay tribute to a distinguished physicist while simultaneously the members of the Pais Award Committee and the Pais Selection encouraging donations to support the travel expenses of speak- Committee for their hard work over the last several years that ers.
    [Show full text]
  • Topics in Inference and Modeling in Physics a Dissertation Presented to the Faculty of the Graduate School In
    Respect Your Data: Topics in Inference and Modeling in Physics A Dissertation Presented to the Faculty of the Graduate School in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy by Colin Clement December 2019 c 2019 Colin Clement ALL RIGHTS RESERVED 2 ABSTRACT Respect Your Data: Topics in Inference and Modeling in Physics Colin Clement, Ph.D. Cornell University 2019 We discuss five topics related to inference and modeling in physics: image registration, magnetic image deconvolution, effective models of spin glasses, the two-dimensional Ising model, and a benchmark dataset of the arXiv pre-print service. First, we solve outstanding problems with image registration (which aims to infer the rigid shift relating two or more noisy shifted images), obtaining the information-theoretic limit in the precision of image shift estimation. Then, we use Bayesian inference and develop new physically-motivated priors in order to solve the ill-posed deconvolution problem of reconstructing electric currents from a magnetic images. After that, we apply machine learning and information geometry to study a spin glass model, finding that this model of canonical complexity is sloppy and thus allows for lower-dimensional effective descriptions. Next, we address outstanding questions regarding the corrections to scaling of the two dimensional Ising i model by applying Normal Form Theory of dynamical systems to the Renormalization Group (RG) flows and raise important questions about the RG in various statistical ensembles. Finally, we develop tools and practices to cast the entire arXiv pre-print service into a benchmark dataset for studying models on graphs with multi-modal features.
    [Show full text]
  • THE ISBA NEWSLETTER Vol
    THE ISBA NEWSLETTER Vol. 5 No. 2 September 1998 The official newsletter of the International Society for Bayesian Analysis President: Susie Bayarri President-Elect: John Geweke Past-President: Steve Feinberg Treasurer: Rob McCulloch Executive Secretary: Mike Evans Board Members: Alan Gelfand, Jay Kadane, Robert Kass, Luis Pericchi, Phil Dawid, Enrique de Alba, Ed George, Malay Ghosh, Alicia Carriquiry, David Rios Insua, Peter Mueller, William Strawderman Message from the President report on all sorts of possible publications for ISBA, including electronic (to be chaired by John Geweke). It seems that I can not escape from it any longer! I You can find information on our web page. I am always have been nicely, but insistently reminded that I surprised that people are willing to devote much-needed should address you at least once. Well, this is it! time and effort to keep ISBA rolling. I am indeed most grateful to all of you who said yes when asked to serve I am about three-quarters through my term as ISBA in some way. President, and it has been a most interesting, and rewarding experience. Our society is still very young, ISBA has been most lucky in counting on the effort, time so there are many (many, many) things to do. I have and dedication of Mike Evans. He has done a few of them, following the way so efficiently done a superb job as Executive Secretary, fixing all the lead by the former President Steve Fienberg (who has millions of details that are always pending. If this was a larger and older society to rule these days), but I am little, he also launched our new web page with plenty of leaving most of them to our next President, John yummy information (I won't tell you about it, so go and Geweke.
    [Show full text]
  • This Is IT: a Primer on Shannon's Entropy and Information
    Information Theory, 49–86 © The Author(s), under exclusive license to Springer Poincar´eSeminar 2018 Nature Switzerland AG 2021 This is IT: A Primer on Shannon’s Entropy and Information Olivier Rioul I didn’t like the term ‘information theory’. Claude [Shannon] didn’t like it either. You see, the term ‘information theory’ suggests that it is a theory about information – but it’s not. It’s the transmission of information, not information. Lots of people just didn’t understand this. – Robert Fano, 2001 Abstract. What is Shannon’s information theory (IT)? Despite its continued impact on our digital society, Claude Shannon’s life and work is still unknown to numerous people. In this tutorial, we review many aspects of the concept of entropy and information from a historical and mathematical point of view. The text is structured into small, mostly independent sections, each covering a particular topic. For simplicity we restrict our attention to one-dimensional variables and use logarithm and exponential notations log and exp without specifying the base. We culminate with a simple exposition of a recent proof (2017) of the entropy power inequality (EPI), one of the most fascinating inequalities in the theory. 1. Shannon’s Life as a Child Claude Elwood Shannon was born in 1916 in Michigan, U.S.A., and grew up in the small town of Gaylord. He was a curious, inventive, and playful child, and probably remained that way throughout his life. He built remote-controlled models and set up his own barbed-wire telegraph system to a friend’s house [48].
    [Show full text]
  • Intelligent Machines in the 21St Century: Automating the Processes of Inference and Inquiry by Kevin H
    Intelligent machines in the 21st century: automating the processes of inference and inquiry By Kevin H. Knuth Computational Sciences Div., NASA Ames Research Center, Moffett Field, CA The last century saw the application of Boolean algebra toward the construction of computing machines, which work by applying logical transformations to infor- mation contained in their memory. The development of information theory and the generalization of Boolean algebra to Bayesian inference have enabled these comput- ing machines, in the last quarter of the twentieth century, to be endowed with the ability to learn by making inferences from data. This revolution is just beginning as new computational techniques continue to make difficult problems more accessible. However, modern intelligent machines work by inferring knowledge using only their pre-programmed prior knowledge and the data provided. They lack the ability to ask questions, or request data that would aid their inferences. Recent advances in understanding the foundations of probability theory have re- vealed implications for areas other than logic. Of relevance to intelligent machines, we identified the algebra of questions as the free distributive algebra, which now allows us to work with questions in a way analogous to that which Boolean algebra enables us to work with logical statements. In this paper we describe this logic of inference and inquiry using the mathematics of partially ordered sets and the scaffolding of lattice theory, discuss the far-reaching implications of the methodol- ogy, and demonstrate its application with current examples in machine learning. Automation of both inference and inquiry promises to allow robots to perform sci- ence in the far reaches of our solar system and in other star systems by enabling them to not only make inferences from data, but also decide which question to ask, experiment to perform, or measurement to take given what they have learned and what they are designed to understand.
    [Show full text]
  • Samenvatting
    UvA-DARE (Digital Academic Repository) Maximum caliber approach to reweight dynamics of non-equilibrium steady states Bause, M. Publication date 2021 Document Version Other version License Other Link to publication Citation for published version (APA): Bause, M. (2021). Maximum caliber approach to reweight dynamics of non-equilibrium steady states. General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl) Download date:26 Sep 2021 7 Summary The present thesis adds a new option for the sparse field of non-equilibrium steady states (NESS) reweighting of dynamics. It is built on the basis of Jaynes Maxi- mum Caliber, an ensemble description of NESS by global balance an local entropy productions, dynamical information drawn from stochastic thermodynamics and coarse-grained dynamics by a Markovian assumption.
    [Show full text]