ebook img

Scientific Computing: Eigenvalues and Optimization PDF

621 Pages·2018·6.781 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Scientific Computing: Eigenvalues and Optimization

var sdd = ( maxev - minev ) * sqrtrnd - minev; sdd = Math.max( sdd, 0 ); var mu2 = Math.min( maxadd, sdd ); A[0] += mu2; A[3] += mu2; mu += mu2; 19 John A. Trangenstein Scientifi c Computing Vol. II – Eigenvalues and Optimization Editorial Board T. J.Barth M.Griebel D.E.Keyes R.M.Nieminen D.Roose T.Schlick Texts in Computational 19 Science and Engineering Editors TimothyJ. Barth MichaelGriebel DavidE. Keyes RistoM. Nieminen DirkRoose TamarSchlick Moreinformationaboutthisseriesathttp://www.springer.com/series/5151 John A. Trangenstein Scientific Computing Vol. II – Eigenvalues and Optimization 123 JohnA.Trangenstein ProfessorEmeritusofMathematics DepartmentofMathematics DukeUniversity Durham NorthCarolina,USA Additionalmaterialtothisbookcanbedownloadedfromhttp://extras.springer.com. ISSN1611-0994 ISSN2197-179X (electronic) TextsinComputationalScienceandEngineering ISBN978-3-319-69106-0 ISBN978-3-319-69107-7 (eBook) https://doi.org/10.1007/978-3-319-69107-7 LibraryofCongressControlNumber:2018932366 MathematicsSubjectClassification(2010):15,65 ©SpringerInternationalPublishingAG,partofSpringerNature2017 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology nowknownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional claimsinpublishedmapsandinstitutionalaffiliations. Printedonacid-freepaper ThisSpringerimprintispublishedbytheregisteredcompanySpringerInternationalPublishingAGpart ofSpringerNature. Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland To mydaughterPamela Preface Thisisthesecondvolumeinathree-volumebookaboutscientificcomputing.The primary goal in these volumes is to present many of the important computational topics and algorithms used in applications such as engineering and physics, together with the theory needed to understand their proper operation. However, a secondarygoalin thedesign ofthis bookisto allowreadersto experimentwith a numberof interactiveprogramswithin the book,so thatreaderscan improvetheir understandingoftheproblemsandalgorithms.Thisinteractivityisavailableinthe HTMLformofthebook,throughJavaScriptprograms. Theintendedaudienceforthisbookareupperlevelundergraduatestudentsand beginninggraduatestudents.Duetotheself-containedandcomprehensivetreatment of the topics, this book should also serve as a useful reference for practicing numericalscientists. Instructors could use this book for multisemester courses on numericalmethods.Theycouldalsouseindividualchaptersforspecializedcourses suchasnumericallinearalgebra,constrainedoptimization,ornumericalsolutionof ordinarydifferentialequations. Inordertoreadallvolumesofthisbook,readersshouldhaveabasicunderstand- ing of both linear algebra and multivariable calculus. However, for this volume it willsufficetobefamiliarwithlinearalgebraandsinglevariablecalculus.Someof thebasicideasforbothoftheseprerequisitesarereviewedinthistext,butatalevel thatwouldbeveryhardtofollowwithoutpriorfamiliaritywiththosetopics.Some experience with computer programming would also be helpful, but not essential. Studentsshouldunderstandthepurposeofacomputerprogram,androughlyhowit operatesoncomputermemorytogenerateoutput. Manyofthe computerprogrammingexampleswilldescribetheuse ofa Linux operating system. This is the only publicly available option in our mathematics department, and it is freely available to all. Students who are using proprietary operating systems, such as Microsoft and Apple systems, will need to replace statementsspecifictoLinuxwiththecorrespondingstatementsthatareappropriate totheirenvironment. This book also references a large number of programs available in several programming languages, such as C, CCC, Fortran and JavaScript, as well as vii viii Preface MATLABmodules.Theseprogramsshouldprovideexamplesthatcantrainreaders to developtheir own programs,fromexistingsoftware wheneverpossibleor from scratchwhenevernecessary. Chaptersbeginwithanoverviewoftopicsandgoals,followedbyrecommended books and relevant software. Some chapters also contain a case study, in which the techniques of the chapter are used to solve an important scientific computing problemindepth. Chapter1presentsnumericalmethodsforfindingeigenvaluesandeigenvectors. These mathematical tools are important in such diverse problem areas as funda- mentalmodesofvibrationinengineering,orgrowthanddecayratesineconomics. Laterin thischapter,we developthesingularvaluedecompositionmorecarefully. Attheendofthechapter,wealsoshowhowtouseeigenvaluesandeigenvectorsto solvelinearrecurrences,ortocomputefunctionsofmatrices,suchasexponentials orsquareroots.ThischapterdependsonthematerialinChap.6ofVolumeI. Chapter 2 examines some important methods for solving large and typically sparsesystemsoflinearequations.Thesenumericalmethodsareespeciallyuseful for implementing implicit methods for partial differential equations (which are discussed in two other books by this author [172, 173]). This chapter depends on thematerialinChap.1. Chapter 3 discusses the numerical solution of systems of nonlinear equations. Here the mathematicalanalysisdependsstrongly on multivariablecalculus. There aresomeusefulgeneralizationsofmethodsfromChap.5ofVolumeItothesolution ofproblemsinthischapter.Therearealsosomeusefulgeneralizationsofiterative methodsforlinearsystemsthatextendtononlinearsystems.Thechapterendswith a case study involving the Lennard-Jones potential from chemistry. This chapter dependsonthematerialinChap.5ofVolumeIandChap.2. Chapter 4 examines constrained optimization methods. This subject is usually left to books that are devoted solely to this topic. The chapter begins with linear programmingproblems,advances to quadratic programmingproblems, and ends with general nonlinear programmingproblems. This chapter depends on the materialinChap.3andChap.6ofVolumeI. Insummary,thisvolumecoversmathematicalandnumericalanalysis,algorithm selection, and software development. The goal is to prepare readers to build programsfor solving important problems in their chosen discipline. Furthermore, theyshoulddevelopenoughmathematicalsophisticationtoknowthelimitationsof the pieces of their algorithmand to recognizewhen numericalfeaturesare due to programmingbugsratherthanthecorrectresponseoftheirproblem. Iamindebtedtomanyteachersandcolleagueswhohaveshapedmyprofessional experience.IthankJimDouglasJr.forintroducingmetonumericalanalysisasan undergrad. (Indeed, I could also thank a class in category theory for motivating me to look for an alternative field of mathematical study.) John Dennis, James Bunch,andJorgeMoréallprovidedafirmfoundationformytraininginnumerical analysis, while Todd Dupont, Jim Bramble, and Al Schatz gave me important training in finite element analysis for my PhD thesis. But I did not really learn to program until I met Bill Gragg, who also emphasized the importance of classical Preface ix analysis in the development of fundamental algorithms. I also learned from my students,particularlyRandyLeVeque,whowasinthefirstnumericalanalysisclass Ievertaught.Finally,IwanttothankBillAllardformanyconversationsaboutthe deficienciesinnumericalanalysistexts.Ihopethatthisbookmovesthefielda bit inthedirectionthatBillenvisions. Mostofall,Iwanttothankmyfamilyfortheirloveandsupport. Durham,NC,USA JohnA.Trangenstein July7,2017 Contents 1 EigenvaluesandEigenvectors .............................................. 1 1.1 Overview............................................................... 2 1.2 FundamentalConcepts ................................................ 3 1.2.1 Examples..................................................... 3 1.2.2 CharacteristicPolynomial................................... 5 1.2.3 InversesandHermitians ..................................... 9 1.2.4 FundamentalTheoremofAlgebra .......................... 10 1.2.5 GerschgorinCircleTheorem................................ 12 1.2.6 TraceandDeterminant....................................... 14 1.2.7 InvariantSubspace........................................... 17 1.2.8 ChangeofBasis.............................................. 18 1.2.9 SimilarityTransformations.................................. 20 1.2.10 SpectralRadius............................................... 21 1.2.11 DistanceBetweenSubspaces................................ 22 1.3 HermitianMatrices.................................................... 38 1.3.1 Theory........................................................ 38 1.3.2 PerturbationAnalysis........................................ 44 1.3.3 SymmetricTridiagonalMatrices............................ 53 1.3.4 SturmSequences............................................. 55 1.3.5 Bisection...................................................... 65 1.3.6 PowerMethod................................................ 67 1.3.7 QLandQRAlgorithms...................................... 73 1.3.8 DivideandConquer.......................................... 80 1.3.9 dqds........................................................... 84 1.3.10 SolverComparison........................................... 89 1.3.11 HouseholderTridiagonalization............................. 90 1.3.12 LanczosProcess.............................................. 92 1.3.13 SymmetricEigenvalueProblems............................ 97 1.4 GeneralMatrices....................................................... 98 1.4.1 Theory........................................................ 98 1.4.2 PerturbationAnalysis........................................ 123 xi

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.