Remerciez-le!

Remerciez @Admin pour avoir partagé cet document gratuitement, de la manière la plus simple, en partageant sur les réseaux sociaux.

! ! ! Chebfun Guide 1st Edition For Chebfun version 5 ! ! ! ! ! ! ! ! ! ! ! ! !

! ! ! Chebfun Guide 1st Edition For Chebfun version 5 ! ! ! ! ! ! ! ! ! ! ! ! ! ! Edited by: Tobin A. Driscoll, Nicholas Hale, and Lloyd N. Trefethen ! ! ! Copyright 2014 by Tobin A. Driscoll, Nicholas Hale, and Lloyd N. Trefethen. All rights reserved. For information, write to help@chebfun.org.! ! ! ! MATLAB is a registered trademark of The MathWorks, Inc. For MATLAB product information, please write to info@mathworks.com. ! ! ! ! ! ! ! ! Dedicated to the Chebfun developers of the past, present, and future. ! Table of Contents! ! Preface! ! Part I: Functions of one variable! ! 1. Getting started with Chebfun Lloyd N. Trefethen! 2. Integration and differentiation Lloyd N. Trefethen! 3. Rootﬁnding and minima and maxima Lloyd N. Trefethen! 4. Chebfun and approximation theory Lloyd N. Trefethen! 5. Complex Chebfuns Lloyd N. Trefethen! 6. Quasimatrices and least squares Lloyd N. Trefethen! 7. Linear differential operators and equations Tobin A. Driscoll! 8. Chebfun preferences Lloyd N. Trefethen! 9. Inﬁnite intervals, inﬁnite function values, and singularities Lloyd N. Trefethen! 10. Nonlinear ODEs and chebgui Lloyd N. Trefethen Part II: Functions of two variables (Chebfun2) ! 11. Chebfun2: Getting started Alex Townsend! 12. Integration and differentiation Alex Townsend! 13. Rootﬁnding and optimisation Alex Townsend! 14. Vector calculus Alex Townsend! 15. 2D surfaces in 3D space Alex Townsend ! ! Preface! ! This guide is an introduction to the use of Chebfun, an open source software package that aims to provide “numerical computing with functions.” Chebfun extends MATLAB’s inherent facilities with vectors and matrices to functions and operators. For those already familiar with MATLAB, much of what Chebfun does will hopefully feel natural and intuitive. Conversely, for those new to MATLAB, much of what you learn about Chebfun can be applied within native MATLAB too. ! ! Some of the chapters give clues about how Chebfun does its work, but that is not an emphasis of this guide. For the mathematical underpinnings of Chebfun, the best source is Approximation Theory and Approximation Practice, by Lloyd N. Trefethen (Society for Industrial and Applied Mathematics, 2013). For the algorithmic backstory, refer to the list of publications maintained at www.chebfun.org.! ! We gratefully acknowledge the Engineering and Physical Sciences Research Council of the UK, The MathWorks, Inc., and the European Research Council, whose generous support has helped the project grow and thrive. We also acknowledge the support of the University of Oxford and the University of Delaware. ! ! Most especially we acknowledge and praise the many contributors to Chebfun. As of version 5, the code consists of over 100,000 lines produced over the course of the past twelve years. In addition to writing code, many individuals have contributed their time to the design, testing, review, and debugging of code written by others. For their development efforts, we thank Anthony Austin, Zachary Battles, Ásgeir Birkisson, Pedro Gonnet, Stefan Güttel, Hrothgar, Mohsin Javed, Georges Klein, Hadrien Montanelli, Ricardo Pachón, Rodrigo Platte, Mark Richardson, Alex Townsend, Grady Wright, and Kuan Xu. Others who have made key contributions include Jean-Paul Berrut, Folkmar Bornemann, Yuji Nakatsukasa, Vanni Noferini, Sheehan Olver, Joris Van Deun, and Marcus Webb.! ! ! ! ! ! ! ! ! Toby Driscoll, Nick Hale, and Nick Trefethen ! ! ! ! ! ! June 2014! ! ! ! ! ! ! ! ! ! ! ! Part I Functions of one variable 1. Getting Started with Chebfun Lloyd N. Trefethen, October 2009, latest revision June 2014 Contents • 1.1 What is a chebfun? • 1.2 Constructing simple chebfuns • 1.3 Operations on chebfuns • 1.4 Piecewise smooth chebfuns • 1.5 Inﬁnite intervals and inﬁnite function values • 1.6 Periodic functions • 1.7 Rows, columns, and quasimatrices • 1.8 Chebfun features not in this Guide • 1.9 How this Guide is produced • 1.10 References 1.1 What is a chebfun? A chebfun is a function of one variable deﬁned on an interval [a, b]. The syntax for chebfuns is almost exactly the same as the usual MATLAB syntax for vectors, with the familiar MATLAB commands for vectors overloaded in natural ways. Thus, for example, whereas sum(f) returns the sum of the entries when f is a vector, it returns a deﬁnite integral when f is a chebfun. Chebfun with a capital C is the name of the software system. The aim of Chebfun is to ”feel symbolic but run at the speed of numerics”. More precisely our vision is to achieve for functions what ﬂoating-point arithmetic achieves for numbers: rapid computation in which each successive operation is carried out exactly apart from a rounding error that is very small in relative terms [Trefethen 2007]. The implementation of Chebfun is based on the mathematical fact that smooth functions can be represented very eﬃciently by polynomial interpolation in Chebyshev points, or equivalently, thanks to the Fast Fourier Transform, by expansions in Chebyshev polynomials. For a simple function, 20 or 30 points often suﬃce, but the process is stable and eﬀective even for functions complicated enough to require 1000 or 1,000,000 points. Chebfun makes use of adaptive procedures that aim to ﬁnd the right number of points automatically so as to represent each function to roughly machine precision, that is, about 15 digits of relative accuracy. (Originally Chebfun stored function values at Chebyshev points; in Version 5 it switched to storing Chebyshev expansion coeﬃcients.) The mathematical foundations of Chebfun are for the most part well established by results scattered throughout the 20th century. A key early ﬁgure, for example, was Bernstein in the 1910s. Much of 1 2 Chebfun Guide the relevant material can be found collected in the Chebfun-based book Approximation Theory and Approximation Practice[Trefethen 2013]. Chebfun was originally created by Zachary Battles and Nick Trefethen at Oxford during 2002-2005 [Battles & Trefethen 2004]. Battles left the project in 2005, and soon four new members were added to the team: Ricardo Pachon (from 2006), Rodrigo Platte (from 2007), and Toby Driscoll and Nick Hale (from 2008). In 2009, Asgeir Birkisson and Mark Richardson also became involved, and other contributors included Pedro Gonnet, Joris Van Deun, and Georges Klein. Nick Hale served as Director of the project during 2010-2014. The Chebfun Version 5 rewrite was directed by Nick Hale during 2013-2014, and the team included Anthony Austin, Asgeir Birkisson, Toby Driscoll, Hrothgar, Mohsin Javed, Hadrien Montanelli, Alex Townsend, Nick Trefethen, Grady Wright, and Kuan Xu. Further information about Chebfun history is available at the Chebfun web site, http://www.chebfun.org . This Guide is based on Chebfun Version 5, released in June 2014. Chebfun is available at http://www.chebfun.org . 1.2 Constructing simple chebfuns The chebfun command constructs a chebfun from a speciﬁcation such as a string or an anonymous function. If you don’t specify an interval, then the default interval [−1, 1] is used. For example, the following command makes a chebfun corresponding to cos(20x) on [−1, 1] and plots it. f = chebfun(’cos(20*x)’); plot(f) −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 1.5 From this little experiment, you cannot see that f is represented by a polynomial. One way to see this is to ﬁnd the length of f: length(f) ans = 61 1. Getting Started with Chebfun 3 Another is to remove the semicolon that suppresses output: f f = chebfun column (1 smooth piece) interval length endpoint values [ -1, 1] 61 0.41 0.41 Epslevel = 2.183626e-15. Vscale = 1.000000e+00. These results tell us that f is represented by a polynomial interpolant through 61 Chebyshev points, i.e., a polynomial of degree 60. These numbers have been determined by an adaptive process. We can see the data points by plotting f|with the |’.-’ option: plot(f,’.-’) −1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1 1.5 The formula for N + 1 Chebyshev points in [−1, 1] is x(j) = −cos(jπ/N), j = 0 : N, and in the ﬁgure we can see that the points are clustered accordingly near 1 and −1. Note that in the middle of the grid, there are about 5 points per wavelength, which is evidently what it takes to represent this cosine to 15 digits of accuracy. For intervals other than [−1, 1], appropriate Chebyshev points are obtained by a linear scaling. The curve between the data points is the polynomial interpolant, which is evaluated by the barycen- tric formula introduced by Salzer [Berrut & Trefethen 2004, Salzer 1972]. This method of evaluating polynomial interpolants is stable and eﬃcient even if the degree is in the millions [Higham 2004]. What is the integral of f from −1 to 1? Here it is: sum(f) ans = 0.091294525072763 4 Chebfun Guide This number was computed by integrating the polynomial (Clenshaw-Curtis quadrature – see Section 2.1), and it is interesting to compare it to the exact answer from calculus: exact = sin(20)/10 exact = 0.091294525072763 Here is another example, now with the chebfun deﬁned by an anonymous function instead of a string. In this case the interval is speciﬁed uploads/Litterature/ chebfun-guide 1 .pdf