A Colorization Algorithm Based on Local MAP Estimation

Total Page:16

File Type:pdf, Size:1020Kb

A Colorization Algorithm Based on Local MAP Estimation A Colorization Algorithm Based on Local MAP Estimation ∗ Hideki Noda a, , Jin Korekuni b, Michiharu Niimi a aDepartment of Systems Innovation and Informatics, Kyushu Institute of Technology, 680-4 Kawazu, Iizuka, 820-8502 Japan bDepartment of Electrical, Electronic and Computer Engineering, Kyushu Institute of Technology, 1-1 Sensui-cho, Tobata-ku, Kitakyushu, 804-8550 Japan Abstract This paper presents a colorization algorithm which adds color to monochrome im- ages. In this paper, the colorization problem is formulated as the maximum a poste- riori (MAP) estimation of a color image given a monochrome image. Markov random field (MRF) is used for modeling a color image which is utilized as a prior for the MAP estimation. The MAP estimation problem for a whole image is decomposed into local MAP estimation problems for each pixel. Using 0.6% of whole pixels as references, the proposed method produced pretty high quality color images with 25.7 dB to 32.6 dB PSNR values for eight images. Key words: Colorization, Monochrome image, MAP estimation, MRF 1 Introduction Colorization is a process, usually a computer-aided process of adding color to monochrome images or movies. There should be considerable demands for colorization of monochrome images or movies. Colorization is now generally carried out manually using some drawing software tools. A user typically car- ries out segmentation of a monochrome image by giving region boundaries by hand and then assigns a color to each region. Obviously such manual work is very expensive and time-consuming. ∗ Corresponding author. Tel.: +81-948-29-7714; fax: +81-948-29-7709 Email address: [email protected] (Hideki Noda). Preprint submitted to Elsevier Science 28 March 2006 Recently several colorization methods [1–3] have been proposed which do not require intensive manual effort. Welsh et al. proposed a semi-automatic method to colorize a monochrome image by transferring color from a reference color image [1]. The entire color ”mood” of the reference image is transferred to the target monochrome image by matching luminance and texture information between the two images. This method requires an appropriate reference color image which should be prepared by a user and works well only for images where differently colored regions have distinct luminance values or distinct textures. Levin et al. have proposed an interactive method which does not require precise manual segmentation [2]. In their method, instead of manual segmentation, a user needs to give some color scribbles, and the colors are automatically propagated to produce a fully colorized image. Horiuchi [3] has proposed a method where a user gives colors for some pixels and colors for all other pixels are determined automatically by using the probabilistic relax- ation. One of serious problems in his method is that it is computationally very expensive; it takes almost one day to colorize one image. Unlike previously proposed colorization methods, this paper formulates the colorization problem as Bayesian inference, i.e., the maximum a posteriori (MAP) estimation of a color image given a monochrome image. Markov ran- dom field (MRF) [4] is used for modeling a color image which is utilized as a prior for the MAP estimation. In this paper, the global MAP estimation problem for a whole image is approximately decomposed into local MAP es- timation problems for each pixel, and the local MAP estimation is reduced to a simple quadratic programming problem with constraints. 2 Color Image Modeling By Markov Random Field 2.1 Markov Random Field Let L = {(i, j); 1 ≤ i ≤ N1, 1 ≤ j ≤ N2} denote a finite set of sites of an N1 × X N2 rectangular lattice. Let ηij ⊂ L denote the (i, j) pixel’s neighborhood of a 1 X random field XL defined on L.LetCij denote the set of cliques C associated X X with ηij which contains the (i, j) pixel, i.e., (i, j) ∈ Cij . For example, in the X first-order neighborhood, ηij = {(i, j +1), (i, j − 1), (i +1,j), (i − 1,j)} and X Cij = {{(i, j)}, {(i, j), (i, j+1)}, {(i, j), (i, j−1)}, {(i, j), (i+1,j)}, {(i, j), (i− 1,j)}} which consists of one singleton and four doubleton cliques. Let the random field XL = {Xij;(i, j) ∈ L} be a Markov random field (MRF) defined on L with Xijs taking values from a common local state space QX .Itiswell 1 x f x {x ,...,x } In this paper, A and ( A)denotetheset a1 al and the multivariable f x ,...,x A {a ,...,a} function ( a1 al ) respectively, where = 1 l . 2 known that an MRF is completely described by a Gibbs distribution 1 p(xL)= exp{−U(xL)}, (1) ZX N1×N2 where xL is a realization of XL from the configuration space ΩX = QX and U(xL)= U(xC )(2) (i,j)∈L X C∈Cij is the global energy function whereas U(xC ) is the clique energy function and ZX = exp{−U(xL)} (3) xL∈ΩX is the partition function. For details on MRFs and related concepts such as the neighborhoods and cliques, see Ref. [4]. 2.2 A Color Image Model Using Gaussian MRF A color image can be considered as a realization xL = {xij;(i, j) ∈ L} of T a random field XL = {Xij;(i, j) ∈ L},wherexij =(rij,gij,bij) is a color vector at (i, j) pixel composed of red rij, green gij and blue bij components. Color images are modeled by a Gaussian MRF (GMRF) characterized by the following local conditional probability density function (pdf) 2 : 1 1 T −1 | X {− − − } p(xij xη )= 3/2 1/2 exp (xij mij) (ΣX ) (xij mij) ,(4) ij (2π) |ΣX | 2 1 mij = xij+τ . (5) |N| τ∈N Here mij is the mean of neighboring pixels’ color vectors x X = {xij+τ ,τ ∈ ηij N},whereN denotes the neighborhood of (0, 0)-pixel. For example, N = {(0, 1), (0, −1), (1, 0), (−1, 0)} for the first-order neighborhood, and if τ = (0, 1), xij+τ = xi,j+1. ΣX is the covariance matrix of xij − mij. 2 The used GMRF is one of the simplest GMRFs, which can model only nontextured smooth images. We here used this GMRF as a first step, though there are more complicated GMRFs applicable to textured images. 3 3 Color Image Estimation 3.1 Derivation of Estimation Algorithm We assume that a monochrome image yL = {yij;(i, j) ∈ L} is associated with a color image xL = {xij;(i, j) ∈ L} under the following relation: T yij = a xij =0.299rij +0.587gij +0.114bij, 0 ≤ yij,rij,gij,bij ≤ 255.(6) Given yL, xL can be estimated by maximizing the a posteriori probability p(xL | yL), i.e., by MAP estimation. The MAP estimate xˆL is written as xˆL =argmax p(xL | yL), (7) xL∈ΩX where the a posteriori probability p(xL | yL) is described as p(yL | xL)p(xL) | p(xL yL)= | . (8) xL∈ΩX p(yL xL)p(xL) Note that it is practically impossible to find the MAP estimate xˆL since the 3|L| search space over all possible configurations of xL is huge, i.e., |ΩX | = 256 . To overcome this problem, hereinafter we consider mean-field-based decom- position of the a posteriori probability. Considering (6), p(yL | xL) is described as T p(yL | xL)=1({yij = a xij, (i, j) ∈ L}) T = 1(yij = a xij), (9) (i,j)∈L where ⎧ ⎨⎪ T T 1ifyij = a xij 1(yij = a xij)=⎪ (10) ⎩ 0otherwise. Using the mean field approximation, p(xL) can be decomposed as [5] p(xL) p(xij |x X ), (11) ηij (i,j)∈L 4 where xηX denotes the mean fields for xηX . Substituting (9) and (11) into ij ij (8) and replacing xL∈ΩX (i,j)∈L by (i,j)∈L xij ∈QX , we obtain the following decomposition for p(xL | yL): p(xL | yL) p(xij | yij, x X ), (12) ηij (i,j)∈L where T 1(yij = a xij)p(xij |x X ) ηij p(xij | yij, x X )= . (13) ηij T ∈Q 1(yij = a xij)p(xij |x X ) xij X ηij In the following, x X is simply used for x X .Thenp(xij | yij, x X )=p(xij | ηij ηij ηij yij, x X ) is considered as a local a posteriori probability (LAP). Using these ηij LAPs, the global optimization problem shown by Eq. (7) is approximately decomposed into the local optimization problems xˆij =arg max p(xij | yij, xηX ). (14) xij ∈QX ij In order to solve (14) for all (i, j) pixels, their neighboring color vectors x X ηij should be given. Since such a problem as shown in (14) can be solved iteratively as is popular in numerical analysis, we rewrite Eq. (14) as (p+1) | (p) xij =arg max p(xij yij, xηX ), (15) xij ∈QX ij where p represents the pth iteration. Considering (4), (5), (6) and (13), the local MAP estimation (15) is rewritten as the following constrained quadratic programming problem: T −1 1 (p) minimize (xij − mij) (ΣX ) (xij − mij)withmij = xij+τ (16) |N| τ∈N T subject to a xij = yij, 0 ≤ rij,gij,bij ≤ 255 (17) 3.2 Initial Color Estimation Since the color estimation shown by Eq. (15) is carried out iteratively, an initial color image is needed to start the iterative procedure. Initial color image estimation using some reference colors is here described. Assuming that color 5 vectors for K pixels, cikjk ,k =1,...,K are given, consider how to derive an initial color image. We consider an initial color estimation procedure which consists of two steps.
Recommended publications
  • RESOURCES in NUMERICAL ANALYSIS Kendall E
    RESOURCES IN NUMERICAL ANALYSIS Kendall E. Atkinson University of Iowa Introduction I. General Numerical Analysis A. Introductory Sources B. Advanced Introductory Texts with Broad Coverage C. Books With a Sampling of Introductory Topics D. Major Journals and Serial Publications 1. General Surveys 2. Leading journals with a general coverage in numerical analysis. 3. Other journals with a general coverage in numerical analysis. E. Other Printed Resources F. Online Resources II. Numerical Linear Algebra, Nonlinear Algebra, and Optimization A. Numerical Linear Algebra 1. General references 2. Eigenvalue problems 3. Iterative methods 4. Applications on parallel and vector computers 5. Over-determined linear systems. B. Numerical Solution of Nonlinear Systems 1. Single equations 2. Multivariate problems C. Optimization III. Approximation Theory A. Approximation of Functions 1. General references 2. Algorithms and software 3. Special topics 4. Multivariate approximation theory 5. Wavelets B. Interpolation Theory 1. Multivariable interpolation 2. Spline functions C. Numerical Integration and Differentiation 1. General references 2. Multivariate numerical integration IV. Solving Differential and Integral Equations A. Ordinary Differential Equations B. Partial Differential Equations C. Integral Equations V. Miscellaneous Important References VI. History of Numerical Analysis INTRODUCTION Numerical analysis is the area of mathematics and computer science that creates, analyzes, and implements algorithms for solving numerically the problems of continuous mathematics. Such problems originate generally from real-world applications of algebra, geometry, and calculus, and they involve variables that vary continuously; these problems occur throughout the natural sciences, social sciences, engineering, medicine, and business. During the second half of the twentieth century and continuing up to the present day, digital computers have grown in power and availability.
    [Show full text]
  • THIAGO JOSÉ CÓSER Possibilidades Da Produção Artística Via
    THIAGO JOSÉ CÓSER Possibilidades da produção artística via prototipagem rápida: processos CAD/CAM na elaboração e confecção de obras de arte e o vislumbre de um percurso poético individualizado neste ensaio. Dissertação apresentada ao Instituto de Artes da Universidade Estadual de Campinas, para a obtenção do título de mestre em Artes. Área de concentração: Artes Visuais Orientador: Prof. Dr. Marco Antonio Alves do Valle Campinas 2010 3 FICHA CATALOGRÁFICA ELABORADA PELA BIBLIOTECA DO INSTITUTO DE ARTES DA UNICAMP Cóser, Thiago José. C89p Possibilidades da produção artística via Prototipagem Rápida: Processos CAD/CAM na elaboração e confecção de obras de arte e o vislumbre de um percurso poético individualizado neste ensaio. : Thiago José Cóser. – Campinas, SP: [s.n.], 2010. Orientador: Prof. Dr. Marco Antonio Alves do Valle. Dissertação(mestrado) - Universidade Estadual de Campinas, Instituto de Artes. 1. Prototipagem rápida. 2. Arte. 3. Sistema CAD/CAM. 4. Modelagem 3D. 5. escultura. I. Valle, Marco Antonio Alves do. II. Universidade Estadual de Campinas. Instituto de Artes. III. Título. (em/ia) Título em inglês: “Possibilities of Art via Rapid Prototyping: using CAD / CAM systems to create art works and a glimpse of a poetic route individualized essay.” Palavras-chave em inglês (Keywords): Rapid prototyping ; Art ; CAD/CAM systems. ; 3D modelling ; Sculpture. Titulação: Mestre em Artes. Banca examinadora: Prof. Dr. Marco Antonio Alves do Valle. Profª. Drª. Sylvia Helena Furegatti. Prof. Dr. Francisco Borges Filho. Prof. Dr. Carlos Roberto Fernandes. (suplente) Prof. Dr. José Mario De Martino. (suplente) Data da Defesa: 26-02-2010 Programa de Pós-Graduação: Artes. 4 5 Agradecimentos Ao meu orientador, profº Dr.
    [Show full text]
  • The Missing Link Between Information Visualization and Art
    Visualization Criticism – The Missing Link Between Information Visualization and Art Robert Kosara The University of North Carolina at Charlotte [email protected] Abstract of what constitutes visualization and a foundational theory are still missing. Even for the practical work that is be- Classifications of visualization are often based on tech- ing done, there is very little discussion of approaches, with nical criteria, and leave out artistic ways of visualizing in- many techniques being developed ad hoc or as incremental formation. Understanding the differences between informa- improvements of previous work. tion visualization and other forms of visual communication Since this is not a technical problem, a purely techni- provides important insights into the way the field works, cal approach cannot solve it. We therefore propose a third though, and also shows the path to new approaches. way of doing information visualization that not only takes We propose a classification of several types of informa- ideas from both artistic and pragmatic visualization, but uni- tion visualization based on aesthetic criteria. The notions fies them through the common concepts of critical thinking of artistic and pragmatic visualization are introduced, and and criticism. Visualization criticism can be applied to both their properties discussed. Finally, the idea of visualiza- artistic and pragmatic visualization, and will help to develop tion criticism is proposed, and its rules are laid out. Visu- the tools to build a bridge between them. alization criticism bridges the gap between design, art, and technical/pragmatic information visualization. It guides the view away from implementation details and single mouse 2 Related Work clicks to the meaning of a visualization.
    [Show full text]
  • Efficient Space-Time Sampling with Pixel-Wise Coded Exposure For
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 1 Efficient Space-Time Sampling with Pixel-wise Coded Exposure for High Speed Imaging Dengyu Liu, Jinwei Gu, Yasunobu Hitomi, Mohit Gupta, Tomoo Mitsunaga, Shree K. Nayar Abstract—Cameras face a fundamental tradeoff between spatial and temporal resolution. Digital still cameras can capture images with high spatial resolution, but most high-speed video cameras have relatively low spatial resolution. It is hard to overcome this tradeoff without incurring a significant increase in hardware costs. In this paper, we propose techniques for sampling, representing and reconstructing the space-time volume in order to overcome this tradeoff. Our approach has two important distinctions compared to previous works: (1) we achieve sparse representation of videos by learning an over-complete dictionary on video patches, and (2) we adhere to practical hardware constraints on sampling schemes imposed by architectures of current image sensors, which means that our sampling function can be implemented on CMOS image sensors with modified control units in the future. We evaluate components of our approach - sampling function and sparse representation by comparing them to several existing approaches. We also implement a prototype imaging system with pixel-wise coded exposure control using a Liquid Crystal on Silicon (LCoS) device. System characteristics such as field of view, Modulation Transfer Function (MTF) are evaluated for our imaging system. Both simulations and experiments on a wide range of
    [Show full text]
  • Applied Analysis & Scientific Computing Discrete Mathematics
    CENTER FOR NONLINEAR ANALYSIS The CNA provides an environment to enhance and coordinate research and training in applied analysis, including partial differential equations, calculus of Applied Analysis & variations, numerical analysis and scientific computation. It advances research and educational opportunities at the broad interface between mathematics and Scientific Computing physical sciences and engineering. The CNA fosters networks and collaborations within CMU and with US and international institutions. Discrete Mathematics & Operations Research RANKINGS DOCTOR OF PHILOSOPHY IN ALGORITHMS, COMBINATORICS, U.S. News & World Report AND OPTIMIZATION #16 | Applied Mathematics Carnegie Mellon University offers an interdisciplinary Ph.D program in Algorithms, Combinatorics, and #7 | Discrete Mathematics and Combinatorics Optimization. This program is the first of its kind in the United States. It is administered jointly #6 | Best Graduate Schools for Logic by the Tepper School of Business (Operations Research group), the Computer Science Department (Algorithms and Complexity group), and the Quantnet Department of Mathematical Sciences (Discrete Mathematics group). #4 | Best Financial Engineering Programs Carnegie Mellon University does not CONTACT discriminate in admission, employment, or Logic administration of its programs or activities on Department of Mathematical Sciences the basis of race, color, national origin, sex, handicap or disability, age, sexual orientation, 5000 Forbes Avenue gender identity, religion, creed, ancestry,
    [Show full text]
  • Colorization Algorithm Using Probabilistic Relaxation
    Colorization Algorithm Using Probabilistic Relaxation Takahiko HORIUCHI Department of Information and Image Sciences, Faculty of Engineering Chiba University 1-33, Yayoi-cho, Inage-ku, Chiba 263-8522, Japan Email: [email protected] TEL/FAX +81-43-290-3485 ABSTRACT: This paper presents a method of colorizing a black and white imagery based on the probabilistic relaxation algorithm. Since the colorization is an ill-posed problem, a user specifies a suitable color on each isolated pixel of an image as a prior information in this paper. Then other pixels in the image are colorized automatically. The colorizing process is done by assuming local Markov property on the images. By minimizing a total of RGB pixel-wise differences, the problem can be considered as a combinatorial optimization problem and it is solved by using the probabilistic relaxation. The proposed algorithm works very well when a few percent color pixels are known with confidence. Keywords – Colorization, probabilistic relaxation, Markov property, grayscale image Colorization Algorithm Using Probabilistic Relaxation 1. INTRODUCTION Colorization is a computerized process that adds color to a black and white print, movie and TV program, supposedly invented by Wilson Markle. It was initially used in 1970 to add color to footage of the moon from the Apollo mission. The demand of adding color to grayscale images such as BW movies and BW photos has been increasing. For example, in the amusement field, many movies and video clips have been colorized by human’s labor, and many grayscale images have been distributed as vivid images. In other fields such as archaeology dealing with historical grayscale data and security dealing with grayscale images by a crime prevention camera, we can imagine easily that colorization techniques are useful.
    [Show full text]
  • Metal Complexes of Penicillin and Cephalosporin Antibiotics
    I METAL COMPLEXES OF PENICILLIN AND CEPHALOSPORIN ANTIBIOTICS A thesis submitted to THE UNIVERSITY OF CAPE TOWN in fulfilment of the requirement$ forTown the degree of DOCTOR OF PHILOSOPHY Cape of by GRAHAM E. JACKSON University Department of Chernis try, University of Cape Town, Rondebosch, Cape, · South Africa. September 1975. The copyright of th:s the~is is held by the University of C::i~r:: To\vn. Reproduction of i .. c whole or any part \ . may be made for study purposes only, and \; not for publication. The copyright of this thesis vests in the author. No quotation from it or information derived from it is to be published without full acknowledgementTown of the source. The thesis is to be used for private study or non- commercial research purposes only. Cape Published by the University ofof Cape Town (UCT) in terms of the non-exclusive license granted to UCT by the author. University '· ii ACKNOWLEDGEMENTS I would like to express my sincere thanks to my supervisors: Dr. L.R. Nassimbeni, Dr. P.W. Linder and Dr. G.V. Fazakerley for their invaluable guidance and friendship throughout the course of this work. I would also like to thank my colleagues, Jill Russel, Melanie Wolf and Graham Mortimor for their many useful conrrnents. I am indebted to AE & CI for financial assistance during the course of. this study. iii ABSTRACT The interaction between metal"'.'ions and the penici l)in and cephalosporin antibiotics have been studied in an attempt to determine both the site and mechanism of this interaction. The solution conformation of the Cu(II) and Mn(II) complexes were determined using an n.m.r, line broadening, technique.
    [Show full text]
  • Quality Improvement of Compressed Color Images Using a Probabilistic Approach
    QUALITY IMPROVEMENT OF COMPRESSED COLOR IMAGES USING A PROBABILISTIC APPROACH Nobuteru Takao, Shun Haraguchi, Hideki Noda, Michiharu Niimi Kyushu Institute of Technology, Dept. of Systems Design and Informatics, 680-4 Kawazu, Iizuka, 820-8502 Japan E-mail: {takao, haraguchi, noda, niimi}@mip.ces.kyutech.ac.jp ABSTRACT method using information on luminance component which is not downsampled. In compressed color images, colors are usually represented by luminance and chrominance (YCbCr) components. Con- The interpolation aims to recover only resolution of chromi- sidering characteristics of human vision system, chromi- nance components lost by downsampling. Alternatively, we nance (CbCr) components are generally represented more aim to recover not only resolution lost by downsampling coarsely than luminance component. Aiming at possible re- but also precision lost by a coarser quantization, if possi- covery of chrominance components, we propose a model- ble. Aiming at such recovery of chrominance components, based chrominance estimation algorithm where color im- we propose a model-based method where color images are ages are modeled by a Markov random field (MRF). A sim- modeled by a Markov random field (MRF). A simple MRF ple MRF model is here used whose local conditional proba- model is here used whose local conditional probability den- bility density function (pdf) for a color vector of a pixel is a sity function (pdf) for a color vector of a pixel is a Gaussian Gaussian pdf depending on color vectors of its neighboring pdf depending on color vectors of its neighboring pixels. pixels. Chrominance components of a pixel are estimated Chrominance components of a pixel are estimated by max- by maximizing the conditional pdf given its luminance com- imizing the conditional pdf given its luminance component ponent and its neighboring color vectors.
    [Show full text]
  • Numerical Analysis to Quantum Computing
    Credit: evv/Shutterstock.com and D-Wave, Inc. and D-Wave, Credit: evv/Shutterstock.com How NASA-USRA collaborations have advanced knowledge in and with the use of new computing technologies. When USRA was created in 1969, later its first task was the management the Chief Scientist at the FROM of the Lunar Science Institute near Center. In February of 1972, Duberg NASA’s Manned Spacecraft Center wrote a memorandum to the senior NUMERICAL (now the Johnson Space Center). management of LaRC, expressing A little more than three years his view that: TO later, USRA began to manage the ANALYSIS Institute for Computer Applications The field of computers and their in Science and Engineering (ICASE) application in the scientific QUANTUM at NASA’s Langley Research Center community has had a profound (LaRC). The rationale for creating effect on the progress of ICASE was developed by Dr. John aerospace research as well as COMPUTING E. Duberg (1917-2002), who was technology in general for the the Associate Director at LaRC and past 15 years. With the advent of “super computers,” based on parallel and pipeline techniques, the potentials for research and problem solving in the future seem even more promising and challenging. The only question is how long will it take to identify the potentials, harness the power, and develop the disciplines necessary to employ such tools effectively.1 Twenty years later, Duberg reflected on the creation of ICASE: By the 1970s, Langley’s computing capabilities had kept pace with the rapidly developing John E. Duberg, Chief Scientist, LaRC; George M.
    [Show full text]
  • Mathematics (MATH) 1
    Mathematics (MATH) 1 MATHEMATICS (MATH) MATH 505: Mathematical Fluid Mechanics 3 Credits MATH 501: Real Analysis Kinematics, balance laws, constitutive equations; ideal fluids, viscous 3 Credits flows, boundary layers, lubrication; gas dynamics. Legesgue measure theory. Measurable sets and measurable functions. Prerequisite: MATH 402 or MATH 404 Legesgue integration, convergence theorems. Lp spaces. Decomposition MATH 506: Ergodic Theory and differentiation of measures. Convolutions. The Fourier transform. MATH 501 Real Analysis I (3) This course develops Lebesgue measure 3 Credits and integration theory. This is a centerpiece of modern analysis, providing a key tool in many areas of pure and applied mathematics. The course Measure-preserving transformations and flows, ergodic theorems, covers the following topics: Lebesgue measure theory, measurable sets ergodicity, mixing, weak mixing, spectral invariants, measurable and measurable functions, Lebesgue integration, convergence theorems, partitions, entropy, ornstein isomorphism theory. Lp spaces, decomposition and differentiation of measures, convolutions, the Fourier transform. Prerequisite: MATH 502 Prerequisite: MATH 404 MATH 507: Dynamical Systems I MATH 502: Complex Analysis 3 Credits 3 Credits Fundamental concepts; extensive survey of examples; equivalence and classification of dynamical systems, principal classes of asymptotic Complex numbers. Holomorphic functions. Cauchy's theorem. invariants, circle maps. Meromorphic functions. Laurent expansions, residue calculus. Conformal
    [Show full text]
  • Computational Methods for Numerical Analysis with R James P
    JSS Journal of Statistical Software November 2018, Volume 87, Book Review 2. doi: 10.18637/jss.v087.b02 Reviewer: Abdolvahab Khademi University of Massachusetts Computational Methods for Numerical Analysis with R James P. Howard, II Chapman & Hall/CRC, Boca Raton, 2017. ISBN 9781498723633. xx+257 pp. USD 99.95 (H). https://www.crcpress.com/9781498723633 Numerical approximation algorithms have traditionally been implemented in generic and specialized programming languages, such as C++, Fortran, and MATLAB. However, newer programming languages such as Python and R are becoming more popular among students and researchers. What makes the latter languages distinct from the traditional ones is their tuning towards data analysis (structurally and through libraries), their free cost and accessibility to everyone, and faster updates due to community-based development. These amenities are the main drivers behind the rise and adoption of such modern computational languages. Computational Methods for Numerical Analysis with R reflects this change and a future trend in the use of modern specialized programming languages, such as R. This book is structured in seven chapters, essentially covering the topics in an undergraduate course in numerical analysis. In each chapter, the author presents the concepts clearly, pro- vides R code for the different algorithms used for computations, presents insights, and ends the chapter with a good number of exercises for the reader. The exercises provide practice in both coding and conceptual understanding. The author’s website provides all the R code in a package and an errata sheet. Numerical analysis is defined and compared with symbolic computation in Chapter 1, In- troduction to Numerical Analysis.
    [Show full text]
  • Numerical Analysis and Computing Lecture Notes #01 — First Meeting
    The Professor The Class — Overview The Class... Introduction Application Numerical Analysis and Computing Lecture Notes #01 — First Meeting Joe Mahaffy, [email protected] Department of Mathematics Dynamical Systems Group Computational Sciences Research Center San Diego State University San Diego, CA 92182-7720 http://www-rohan.sdsu.edu/∼jmahaffy Spring 2010 Joe Mahaffy, [email protected] Lecture Notes #01 — First Meeting — (1/26) The Professor The Class — Overview The Class... Introduction Application Outline 1 The Professor Contact Information, Office Hours 2 The Class — Overview Literature & Syllabus Grading CSU Employee Furloughs Expectations and Procedures 3 The Class... Resources Formal Prerequisites 4 Introduction The What? Why? and How? 5 Application Analysis Joe Mahaffy, [email protected] Lecture Notes #01 — First Meeting — (2/26) The Professor The Class — Overview The Class... Contact Information, Office Hours Introduction Application Contact Information Office GMCS-593 Email mahaff[email protected] Web http://www-rohan.sdsu.edu/∼jmahaffy Phone (619)594-3743 Office Hours MW: 1 – 2, 3 – 4), and by appointment Joe Mahaffy, [email protected] Lecture Notes #01 — First Meeting — (3/26) The Professor Literature & Syllabus The Class — Overview Grading The Class... CSU Employee Furloughs Introduction Expectations and Procedures Application Basic Information: The Book Title: “Numerical Analysis,” 8th Edition Authors: Richard L. Burden & J. Douglas Faires Publisher: Thomson – Brooks/Cole ISBN: 0-534-39200-8 Joe Mahaffy, [email protected]
    [Show full text]