Get Variational calculus with engineering applications constantin udriste PDF ebook with Full Chapte

Page 1


https://ebookmass.com/product/variational-calculus-withengineering-applications-constantin-udriste/

Instant digital products (PDF, ePub, MOBI) ready for you

Download now and discover formats that fit your needs...

Calculus with Applications 11th Edition, (Ebook PDF)

https://ebookmass.com/product/calculus-with-applications-11th-editionebook-pdf/

ebookmass.com

Microeconomics: Theory and Applications with Calculus, 5th Global Edition Jeffrey M. Perloff

https://ebookmass.com/product/microeconomics-theory-and-applicationswith-calculus-5th-global-edition-jeffrey-m-perloff/

ebookmass.com

Elementary Vector Calculus and Its Applications with MATLAB Programming (River Publishers Series in Mathematical, Statistical and Computational Modelling for Engineering) 1st Edition Nita H. Shah

https://ebookmass.com/product/elementary-vector-calculus-and-itsapplications-with-matlab-programming-river-publishers-series-inmathematical-statistical-and-computational-modelling-forengineering-1st-edition-nita-h-shah/ ebookmass.com

Planning and Installing Micro-Hydro Systems: A Guide for Designers, Installers and Engineers – Ebook PDF Version

https://ebookmass.com/product/planning-and-installing-micro-hydrosystems-a-guide-for-designers-installers-and-engineers-ebook-pdfversion/ ebookmass.com

Every Season Is Soup Season: 85+ Souper-Adaptable Recipes to Batch, Share, Reinvent, and Enjoy Shelly Westerhausen Worcel

https://ebookmass.com/product/every-season-is-soup-season-85-souperadaptable-recipes-to-batch-share-reinvent-and-enjoy-shellywesterhausen-worcel/

ebookmass.com

American Higher Education, Second Edition: A History 2nd Edition, (Ebook PDF)

https://ebookmass.com/product/american-higher-education-secondedition-a-history-2nd-edition-ebook-pdf/

ebookmass.com

Epoxy Composites Jyotishkumar Parameswaranpillai

https://ebookmass.com/product/epoxy-composites-jyotishkumarparameswaranpillai/

ebookmass.com

The Science of Psychology: An Appreciative View 5th Edition Laura A. King

https://ebookmass.com/product/the-science-of-psychology-anappreciative-view-5th-edition-laura-a-king/

ebookmass.com

Psychiatric Case Studies for Advanced Practice First Edition

https://ebookmass.com/product/psychiatric-case-studies-for-advancedpractice-first-edition/

ebookmass.com

USMLE Step 2 Secrets 5th

https://ebookmass.com/product/usmle-step-2-secrets-5th-editiontheodore-oconnell/

ebookmass.com

VariationalCalculuswithEngineeringApplications

Variational Calculus with Engineering Applications

Thiseditionfirstpublished2023 © 2023JohnWiley&SonsLtd

Allrightsreserved.Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,or transmitted,inanyformorbyanymeans,electronic,mechanical,photocopying,recordingorotherwise, exceptaspermittedbylaw.Adviceonhowtoobtainpermissiontoreusematerialfromthistitleis availableat http://www.wiley.com/go/permissions.

TherightofConstantinUdristeandIonelTevytobeidentifiedastheauthorsofthisworkhasbeen assertedinaccordancewithlaw.

RegisteredOffice(s)

JohnWiley&Sons,Inc.,111RiverStreet,Hoboken,NJ07030,USA

JohnWiley&SonsLtd,TheAtrium,SouthernGate,Chichester,WestSussex,PO198SQ,UK

EditorialOffice(s)

9600GarsingtonRoad,Oxford,OX42DQ,UK

TheAtrium,SouthernGate,Chichester,WestSussex,PO198SQ,UK

Fordetailsofourglobaleditorialoffices,customerservices,andmoreinformationaboutWileyproducts visitusat www.wiley.com.

Wileyalsopublishesitsbooksinavarietyofelectronicformatsandbyprint-on-demand.Somecontent thatappearsinstandardprintversionsofthisbookmaynotbeavailableinotherformats.

LimitofLiability/DisclaimerofWarranty

Whilethepublisherandauthorshaveusedtheirbesteffortsinpreparingthiswork,theymakeno representationsorwarrantieswithrespecttotheaccuracyorcompletenessofthecontentsofthiswork andspecificallydisclaimallwarranties,includingwithoutlimitationanyimpliedwarrantiesof merchantabilityorfitnessforaparticularpurpose.Nowarrantymaybecreatedorextendedbysales representatives,writtensalesmaterialsorpromotionalstatementsforthiswork.Thefactthatan organization,website,orproductisreferredtointhisworkasacitationand/orpotentialsourceoffurther informationdoesnotmeanthatthepublisherandauthorsendorsetheinformationorservicesthe organization,website,orproductmayprovideorrecommendationsitmaymake.Thisworkissoldwith theunderstandingthatthepublisherisnotengagedinrenderingprofessionalservices.Theadviceand strategiescontainedhereinmaynotbesuitableforyoursituation.Youshouldconsultwithaspecialist whereappropriate.Further,readersshouldbeawarethatwebsiteslistedinthisworkmayhavechanged ordisappearedbetweenwhenthisworkwaswrittenandwhenitisread.Neitherthepublishernor authorsshallbeliableforanylossofprofitoranyothercommercialdamages,includingbutnotlimitedto special,incidental,consequential,orotherdamages.

AcataloguerecordforthisbookisavailablefromtheLibraryofCongress

HardbackISBN:9781119944362;ePDFISBN:9781119944379;epubISBN:9781119944386; oBookISBN:9781119944423

Coverimage: © enjoynz/GettyImages

CoverdesignbyWiley

Setin9.5/12.5ptSTIXTwoTextbyIntegraSoftwareServicesPvt.Ltd,Pondicherry,India

Contents

Preface ix

1ExtremaofDifferentiableFunctionals 1

1.1DifferentiableFunctionals 1

1.2ExtremaofDifferentiableFunctionals 6

1.3SecondVariation;SufficientConditionsforExtremum 14

1.4OptimumwithConstraints;thePrincipleofReciprocity 17

1.4.1IsoperimetricProblems 18

1.4.2TheReciprocityPrinciple 19

1.4.3ConstrainedExtrema:TheLagrangeProblem 19

1.5MapleApplicationTopics 21

2VariationalPrinciples 23

2.1ProblemswithNaturalConditionsattheBoundary 23

2.2SufficiencybytheLegendre-JacobiTest 27

2.3UnitemporalLagrangianDynamics 30

2.3.1NullLagrangians 31

2.3.2InvexityTest 32

2.4LavrentievPhenomenon 33

2.5UnitemporalHamiltonianDynamics 35

2.6ParticularEuler–LagrangeODEs 37

2.7MultitemporalLagrangianDynamics 38

2.7.1TheCaseofMultipleIntegralFunctionals 38

2.7.2InvexityTest 40

2.7.3TheCaseofPath-IndependentCurvilinearIntegralFunctionals 41

2.7.4InvexityTest 44

2.8MultitemporalHamiltonianDynamics 45

2.9ParticularEuler–LagrangePDEs 47

2.10MapleApplicationTopics 48

3 Optimal Models Based on Energies 53

3.1 Brachistochrone Problem 53

3.2 Ropes, Chains and Cables 55

3.3 Newton’s Aerodynamic Problem 56

3.4 Pendulums 59

3.4.1 Plane Pendulum 59

3.4.2 Spherical Pendulum 60

3.4.3 Variable Length Pendulum 61

3.5 Soap Bubbles 62

3.6 Elastic Beam 63

3.7 The ODE of an Evolutionary Microstructure 63

3.8 The Evolution of a Multi-Particle System 64

3.8.1 Conservation of Linear Momentum 65

3.8.2 Conservation of Angular Momentum 66

3.8.3 Energy Conservation 67

3.9 String Vibration 67

3.10 Membrane Vibration 70

3.11 The Schrödinger Equation in Quantum Mechanics 73

3.11.1 Quantum Harmonic Oscillator 73

3.12 Maple Application Topics 74

4 Variational Integrators 79

4.1 Discrete Single-time Lagrangian Dynamics 79

4.2 Discrete Hamilton’s Equations 84

4.3 Numeric Newton’s Aerodynamic Problem 87

4.4 Discrete Multi-time Lagrangian Dynamics 88

4.5 Numerical Study of the Vibrating String Motion 92

4.5.1 Initial Conditions for Infinite String 94

4.5.2 Finite String, Fixed at the Ends 95

4.5.3 Monomial (Soliton) Solutions 96

4.5.4 More About Recurrence Relations 100

4.5.5 Solution by Maple via Eigenvalues 101

4.5.6 Solution by Maple via Matrix Techniques 102

4.6 Numerical Study of the Vibrating Membrane Motion 104

4.6.1 Monomial (Soliton) Solutions 105

4.6.2 Initial and Boundary Conditions 108

4.7 Linearization of Nonlinear ODEs and PDEs 109

4.8 Von Neumann Analysis of Linearized Discrete Tzitzeica PDE 113

4.8.1 Von Neumann Analysis of Dual Variational Integrator Equation 115

4.8.2 Von Neumann Analysis of Linearized Discrete Tzitzeica Equation 116

4.9 Maple Application Topics 119

5 Miscellaneous Topics 123

5.1 Magnetic Levitation 123

5.1.1 Electric Subsystem 123

5.1.2ElectromechanicSubsystem 124

5.1.3StateNonlinearModel 124

5.1.4TheLinearizedModelofStates 125

5.2TheProblemofSensors 125

5.2.1SimplifiedProblem 126

5.2.2ExtendingtheSimplifiedProblemofSensors 128

5.3TheMovementofaParticleinNon-stationaryGravito-vortexField 128

5.4GeometricDynamics 129

5.4.1Single-timeCase 129

5.4.2TheLeastSquaresLagrangianinConditioningProblems 130

5.4.3Multi-timeCase 133

5.5TheMovementofChargedParticleinElectromagneticField 134

5.5.1UnitemporalGeometricDynamicsInducedbyVectorPotential A 135

5.5.2UnitemporalGeometricDynamicsProducedbyMagneticInduction B 136

5.5.3UnitemporalGeometricDynamicsProducedbyElectricField E 136

5.5.4PotentialsAssociatedtoElectromagneticForms 137

5.5.5PotentialAssociatedtoElectric1-form E 138

5.5.6PotentialAssociatedtoMagnetic1-form H 138

5.5.7PotentialAssociatedtoPotential1-form A 138

5.6WindTheoryandGeometricDynamics 139

5.6.1PendularGeometricDynamicsandPendularWind 141

5.6.2LorenzGeometricDynamicsandLorenzWind 142

5.7MapleApplicationTopics 143

6NonholonomicConstraints 147

6.1ModelsWithHolonomicandNonholonomicConstraints 147

6.2RollingCylinderasaModelwithHolonomicConstraints 151

6.3RollingDisc(Unicycle)asaModelwithNonholonomicConstraint 152

6.3.1NonholonomicGeodesics 152

6.3.2GeodesicsinSleighProblem 155

6.3.3UnicycleDynamics 156

6.4NonholonomicConstraintstotheCarasaFour-wheeledRobot 157

6.5NonholonomicConstraintstothe N-trailer 158

6.6FamousLagrangians 160

6.7SignificantProblems 160

6.8MapleApplicationTopics 163

7Problems:FreeandConstrainedExtremals 165

7.1SimpleIntegralFunctionals 165

7.2CurvilinearIntegralFunctionals 169

7.3MultipleIntegralFunctionals 171

7.4LagrangeMultiplierDetails 174

7.5SimpleIntegralFunctionalswithODEConstraints 175

7.6SimpleIntegralFunctionalswithNonholonomicConstraints 181

7.7SimpleIntegralFunctionalswithIsoperimetricConstraints 184

7.8MultipleIntegralFunctionalswithPDEConstraints 186

7.9MultipleIntegralFunctionalsWithNonholonomicConstraints 188

7.10MultipleIntegralFunctionalsWithIsoperimetricConstraints 189

7.11CurvilinearIntegralFunctionalsWithPDEConstraints 191

7.12CurvilinearIntegralFunctionalsWithNonholonomicConstraints 193

7.13CurvilinearIntegralFunctionalswithIsoperimetricConstraints 195

7.14MapleApplicationTopics 197

Bibliography 203 Index 209

Preface

TheVariationalCalculuswithEngineeringApplicationswasandisbeingtaughtto fourth-yearengineeringstudents,intheFacultyofAppliedSciences,Mathematics-InformaticsDepartment,fromtheUniversityPolitehnicaofBucharest,byProf.EmeritusDr. ConstantinUdriste.Certaintopicsaretaughtatotherfacultiesofouruniversity,especiallyasmaster’sordoctoralcourses,beingpresentinthepapersthatcanbepublished injournalsnowcategorizedas“ISI”.

Thechapterswerestructuredaccordingtotheimportance,accessibility,andimpactof thetheoreticalnotionsabletooutlineafuturespecialistbasedonmathematicaloptimizationtools.Theprobingandintermediatevariantslastedforanumberofsixteenyears, leadingtotheselectionofthemostimportantmanageablenotionsandreachingmaturity throughthisvariantthatwedecidedtopublishatWiley.

Nowthetopicsofthebookincludesevenchapters:ExtremaofDifferentiableFunctionals;VariationalPrinciples;OptimalModelsBasedonEnergies;VariationalIntegrators; MiscellaneousTopics;ExtremalswithNonholonomicConstraints;Problems:Freeand ConstrainedExtremals.Tocovermodernproblem-solvingmethods,eachchapterincludes Mapleapplicationtopics.Thescientificauthoritywehaveasprofessorsintechnicaluniversityhasallowedustoimposeourpointofviewonthesubjectandonthetypesof reasoningthatdeservetobeofferedtoreadersofvariationalcalculustexts.Everything hasbeenstructuredforthebenefitoffunctionaloptimizationthatdescribessignificant physicalorengineeringphenomena.Tounlockthestudents’mindswepreferredasimplifiedlanguage,specifictoappliedmathematics,sendingreadersofpuremathematicstothe bibliography.Longexperienceasmathematicsteachersintechnicaluniversitieshaspreparedustofluentlyexposeideasundressedbyexcessiveformalizations,withgreatimpact onthestudents’understandingofthenaturalphenomenaencounteredinengineeringand economicsaswellastheirdressinginmathematicalclothing.Inthissense,wepreferred mathematicalmodelingasascientificsupportforintelligentpresentationoftherealworld andweavoidedthehintofabstractnotionsspecifictopuremathematics.

Topicsthatdistinguishthisbookfromexistingbooksinclude:Mapleapplicationtopics,invexitytests,caseofpath-independentcurvilinearintegralfunctionals,variational integrators,discretizationofHamiltonODEs,numericalstudyofvibratingstringmotion, numericalstudyofvibratingmembranemotion,linearizationofnonlinearODEsand

x Preface

PDEs,geometricdynamicsandwind,extremalswithnonholonomicconstraints,freeand constrainedextremals.Infact,thisisthefirstbookthatstudiesthecurvilinearintegral functionalsatthelevelofstudentcourses.Thenoveltyissueswerefinalizedfollowing repeateddiscussionswiththeprofessorsfromourfacultyandfromFacultyofMathematics inWestUniversityofTimisoara,especiallywiththoseinterestedinfunctionaloptimization,towhomwebringthanksandthoughtsofappreciation.Weareopentoanycomments orcriticismswhichbringdidacticbenefitsforvariationalcalculustheories.

ThemottosarestanzasfromtwopoemsbyTudorArghezi(1880–1967),aRomanian writer,knownforhiscontributiontothedevelopmentofRomanianlyricpoetryunder theinfluenceofBaudelaireanism.Hispoeticwork,ofexemplaryoriginality,represents anothersignificantageofRomanianliterature.

BorninBucharestofpeasantstock,TudorArgheziwasawardedRomania’sNational PoetryPrizein1946andtheStatePrizeforPoetryin1956.

TheauthorsthankAssociateProf.Dr.Oana-MariaPastae,“ConstantinBrancusi” UniversityofTg-Jiu,fortheEnglishimprovementofthemanuscript.Weareindebtedto Prof.Dr.DumitruOprisfromtheWestUniversityofTimisoaraforthesolutionsofferedto somediscretizationproblems.

Bucharest,April2022

ExtremaofDifferentiableFunctionals

Motto: “Iwrotethemwithmynailontheplaster

Onawallofemptycracks, Inthedark,inmysolitude, Unaidedbythebulllionvulture ofLuke,MarkandJohn.” TudorArghezi– FlowersofMildew

Variationalcalculusaimstogeneralizetheconstructionsofclassicalanalysis,tosolveextrema problemsforfunctionals.Inthisintroductorychapter,wewillstudytheproblemofdifferentiablefunctionalsdefinedonvariousclassesoffunctions.Thenewsreferstopathintegral functionals.Fordetails,see[4, 6, 7, 8, 14, 18, 21, 30, 58].

1.1DifferentiableFunctionals

Letusconsiderthepoint ��=(��1,��2,...,����) andthefunction ��∶��⊂ R�� → R, ��↦��(��). Ifthisfunctionhascontinuouspartialderivativeswithrespecttoeachofthevariables ��1,��2,...,����,thenincreasingtheargumentwith ℎ=(ℎ1,ℎ2,...,ℎ��) produces ��(��+ℎ)−��(��)= ��

��,ℎ).

Thefirsttermontheright-handsiderepresentsthedifferentialofthefunction �� atthe point ��,alinearformofargumentgrowth,thevector ℎ;thesecondtermisthedeviation fromthelinearapproximationandissmallinrelationto ℎ,inthesensethat lim ℎ→0 ��(��,ℎ) ‖ℎ‖ =0.

Let ��,�� betwonormedvectorspaces.Thepreviousdefinitioncanbeextendedimmediatelytothecaseoffunctions ��∶��⊂��→��.If �� isaspaceoffunctions,and ��= R, theninsteadoffunctionweusetheterm functional

VariationalCalculuswithEngineeringApplications,FirstEdition.ConstantinUdristeandIonelTevy. © 2023JohnWiley&SonsLtd.Published2023byJohnWiley&SonsLtd.

Definition1.1: Thefunction �� iscalleddifferentiableatapoint ��∈�� ifthereexista linearandcontinuousoperator ����(��, ) andacontinuousfunction ��(��, ) suchthatfor anyvector ℎ∈�� tohave

10 ��(��+ℎ)−��(��)=����(��,ℎ)+��(��,ℎ),

20 lim ℎ→0 ��(��,ℎ) ‖ℎ‖ =0.

Thelinearcontinuousoperator����(��,⋅)iscalledthederivativeofthefunction�� atgiven point ��.

Foragivennonzerovector ℎ∈�� and ��>0,thevector ��(��,ℎ)=lim ��→0 ��(��+��ℎ)−��(��) �� ∈��, ifthelimitexists,iscalledthe variationorderivativeofthefunction �� atthepoint �� along thedirection ℎ

Adifferentiablefunctionofrealvariableshasaderivativealonganydirection.Thepropertyisalsokeptforfunctionsbetweennormedvectorspaces.Specifically,thenextone takesplace:

Proposition1.1: Ifthefunction ��∶��⊂��→�� isdifferentiableatthepoint ��∈��,then foranynonzerovector ℎ∈�� thefunction ��(��+��ℎ),ofrealvariable ��≥0,isderivablewith respectto ��,for ��=0 and

(��

Proof. Thederivativesoughtisobviously ����(��,ℎ) andthen

Example1.1: Anylinearcontinuousoperator �� ∶ �� → �� is,obviously,adifferentiable functionatanypoint �� and ����(��,ℎ)=����(��,ℎ)=��(ℎ) since ��(��+ℎ)−��(��)=��(ℎ).

Let �� beafunctional.Thesimplestexamplesof functionals aregivenbyformulas:(i) evaluationfunctional(“applicationofafunctiononthevalueatapoint”), �� ↦ ��(��0),

where ��0 isafixedpoint;(ii)definiteintegration(functionaldefinedbydefiniteintegral); (iii)numericalquadraturedefinedbydefiniteintegration:

where ���� meansthesetofallpolynomialsofdegreeatmost ��;and(iv)distributionsin analysis(aslinearfunctionalsdefinedonspacesoftestfunctions).Wewillcontinueto dealwithdefiniteintegraltypefunctionals,asthereasoningismorefavorabletous.

Example1.2: Letusconsiderthefunctional

definedonthespace ��0[��,��] ofthecontinuousfunctionsonthesegment [��,��],endowed withthenormofuniformconvergence.TheLagrangianofthefunctional(i.e.,thefunction ��(��,��))ispresumedcontinuousandwithcontinuouspartialderivativesofthefirstorderin thedomain ��≤��≤��, −∞<��<+∞.

Letusdeterminethevariationofthefunctional ��(��(⋅)) whentheargument ��(��) increases with ℎ(��):

SincetheLagrangian �� isadifferentiablefunction,wehave

(��,��+ℎ)−��(��,��)=

Therefore

Inthisway,accordingtothepreviouspropositionandthederivationformulaforintegrals withparameters,wefind

Example1.3: Letusconsiderthefunctional

(��(⋅))=∫

(��,��(��),��′(��))����,

definedonthespace��1[��,��]offunctionswithacontinuousderivativeonthesegment[��,��], endowedwiththenormofuniformconvergenceofderivatives.TheLagrangian��(��,��,��)ofthe functionalissupposedtohavefirst-ordercontinuouspartialderivatives.Towritetheintegral functional,weuseinfactthepullbackform ��(��,��(��),��′(��)) oftheLagrangian.

Applyingthederivationformulaoftheintegralswithparameterweobtain

Analogously,wecanextendtheresulttothefunctional

definedonthespace ����[��,��] offunctionswithcontinuousderivativesuptotheorder �� inclusive,onthesegment [��,��].Inthiswaywehave

Example1.4: Letusconsiderafunctionalwhichdependsonseveralfunctionvariables,for example

(��1(⋅),...,����(⋅))=

definedonthespace (��1[��,��])��,whoseelementsarevectorfunctions ��(��)= (��1(��),...,����(��)),��∈[��,��],withthenormofuniformconvergenceofderivatives ‖��‖=max ��∈[��,��] [|��

(�� meansthederivativewithrespectto ��).

SupposingthattheLagrangian �� hascontinuouspartialderivatives,denoting ℎ(��)= (ℎ1(��),...,ℎ��(��)),thevariationofthefunctional �� is ����(��,ℎ)=∫ �� �� [

(sumovertheindex ��).

Now,letusconsiderfunctionalswhoseargumentsarefunctionsofseveralrealvariables.

Example1.5: Let Ω ⊂ R2 beacompactdomain.Asanexamplewewilltake,first,the doubleintegralfunctional

wherewenotedforabbreviation

Thefunctionalisdefinedonthespace ��1(Ω) ofallfunctionswithcontinuouspartialderivatives;thenormofthespace ��1(Ω) isgivenby

SupposingthattheLagrangian �� hascontinuouspartialderivatives,thevariationofthe functional ��,astheargument �� growswith ℎ(��,��),is

Example1.6: Anotherexampleisthecurvilinearintegralfunctional

where Γ isapiecewise ��1 curvewhichjoinstwofixedpoints ��,�� inacompactdomain Ω⊂ R2 . Supposetheargument ��(��,��) growswith ℎ(��,��).Thenthevariationofthefunctionalis

1.2ExtremaofDifferentiableFunctionals

Letusconsiderthefunctional ��∶��⊂��→ R,definedonasubset �� ofanormedvector space �� offunctions.Bydefinition,thefunctional �� hasa (relative)minimum atthepoint ��0 in ��,ifthereexistsaneighborhood ��,ofpoint ��0,suchthat

��.

Ifthepoint ��0 hasaneighborhood �� onwhichtheoppositeinequalitytakesplace

(��0)≥��(��), ∀��∈��∩��,

wesaythat ��0 isapointof localmaximum forthefunctional ��.Theminimumand maximumpointsarecalled relativeextrema points.

Inclassicalanalysis,theextremapointsofadifferentiablefunctionareamongthecritical points,thatis,amongthepointsthatcancelsfirst-orderderivatives.Asimilarproperty occursinthecaseoffunctionalonesonnormedvectorspacesoffunctions,onlyinthis casetheextremapointsarefoundbetweentheextremals(criticalpoints).

Proposition1.2: Ifthefunction�� isanextremumpointforthefunctional��,interiorpoint oftheset �� andif �� isdifferentiableatthispoint,then ����(��,ℎ)=0 foranygrowth ℎ

Proof. Letℎbeagrowthoftheargument��(function);since��isaninteriorpointoftheset ��,thefunction��(��+��ℎ)ofrealvariable�� isdefinedonholeinterval[−1,1].Thisfunction hasanextremumpointat �� =0 andisderivableatthispoint.Thenitsderivativemust vanishat ��=0.Itfollows ����(��,ℎ)=���� ����(

Anypoint �� atwhichthevariation ����(��,ℎ) ofthefunctional �� iscanceledidentically withrespectto ℎ iscalledeitherthe stationarypoint or criticalpoint or extremal ofthe functional.

Hence, forestablishingtheextremapointsofafunctional,thevariation ����(��,ℎ) mustbe expressed,determinethecriticalpoints(thoseatwhichthevariationiscanceledidentically withrespectto ℎ)andthenchoosefromthesetheminimumpointsorthemaximumpoints.

CommentaryThevariation����(��,ℎ)ofthefunctional��isacontinuouslinearfunctional onthenormedvectorspace �� onwhichisdefined ��.Theset ��∗ ofallcontinuouslinear functionalson �� iscalled dualspace.Thustheevaluationofthevariationofafunctional imposestheexistenceofsomerepresentationtheoremsforthedualspace.

Forexample,foranylinearfunctional��∶ R�� → Rthereexistsavector��=(��1,...,����)∈ R�� suchthat

��(ℎ)=��1ℎ1+ +����ℎ�� =⟨��,ℎ⟩, ∀ℎ∈ R��

Also,anycontinuousandlinearfunctional��∶��0[��,��]→ R,onthespaceofcontinuous functions,canbewrittenintheform

��(ℎ)=∫ �� �� ℎ(��)������(��),

where ���� isafunctionwithboundedvariationandcontinuousattheright(F.Riesz’s Theorem).

Todeterminetheextremapoints,weneedtheidenticalcancelationofthefirstvariation. Forthatweuse

Lemma1.1: (Fundamentallemmaofvariationalcalculus) Let �� bearealcontinuousfunctionontheinterval[��,��].Ifforanycontinuousfunctionℎ,vanishingat�� and��,the equality

�� ��(��)ℎ(��)����=0, istrue,then ��(��)=0, ∀��∈[��,��]

Proof. Supposethereexistsapoint ��0 ∈[��,��] atwhich ��(��0) ≠ 0.Let ��(��0) > 0.Then thereexistsaneighborhood �� =(��0 ��,��0 +��) of ��0 where ��(��) > 0.Letusconsider acontinuousfunction ℎ,strictlypositiveontheneighborhood �� andnulloutsideit(for example, ℎ(��)=(�� ��0+��)(��0+�� ��) for ��∈��).Then

�� �� ��(��

whichcontradictsthehypothesisandconcludestheproofofthelemma.

Remark1.1: 1)Forthesetofcontinuousrealfunctions ��0[��,��],theintegral ⟨��,��⟩ =

�� �� ��(��)ℎ(��)���� isascalarproduct.Consequently,

implies ��(��)=0.

2)Thesetoffunctions ℎ youneedforcheckingtheequalityinthelemmacanoften bereduced.Ascanbeseenfromtheproofofthefundamentallemma,wecanimpose derivabilityconditionsofanyordertothefunctions ℎ andcancelationattheendofthe intervaletc.Itisessentialthatforany �� ∈[��,��] andany ��> 0 toexist,inthesetof testfunctions,astrictlypositivefunctionontheinterval (��0 ��,��0 +��) andnullfunction otherwise.

3)Inasimilarwording,thefundamentallemmaofthevariationalcalculusremainsin forceforfunctionsofseveralvariables.

4)Wecanalsoseethatinpreviousconstructionsitissufficientthatthefunctional �� be definedonalinearvariety(affinesubspace)oratleastonaconvexset:withtwopoints �� and ��+ℎ tobeinthedomainof �� allpointsoftheform ��+��ℎ forany ��∈ R oratleast ��∈[−1,1]too.Specifically,wecanimposerestrictions(conditions)onextremapointsand variations ℎ,butremaininginalinearvariety.

Problem1.2.1: a) Let �� bea ���� functionontheinterval [��,��].Ifforany ℎ∈����[��,��], nullat �� and ��,togetherwithitsfirst ��−1 derivatives,theequality

�� �� ��(��)ℎ(��)(��)����=0

holds,then �� isapolynomialof ������≤��−1.

b) Thepreviousstatementremainsinforceevenif �� isonlyacontinuousfunction.

Solution. a) Theproperty“�� isapolynomialof ������ ≤��−1”isequivalentto“��(��) = 0,forany �� ∈[��,��]”.Weareinspiredbytheproofofthefundamentallemma.Suppose thereexistsapoint ��0 ∈[��,��] atwhich ��(��)(��0)≠0.Let ��(��)(��0)>0.Thenthereexistsa neighborhood ��=(��0 ��,��0+��),of ��0 where ��(��)(��)>0.Letusconsiderthe ���� function ℎ(��)=(�� ��0+��)��(��0+�� ��)�� for ��∈�� andnulloutside ��.Assumingthat ��⊂(��,��),the functionℎ,togetherwithitsfirst��−1derivativesvanishesin�� and��.Integrating�� times bypartsoneobtains

(>0 or <0 as �� isevenorodd)whichcontradictsthehypothesis.

b) Firstly,letusprovethefollowingresult:

Afunction ��,continuouson [��,��] istheorder �� derivativeofa ���� function ℎ which, togetherwithitsfirst ��−1 derivatives,vanishesin �� and ��, ��=ℎ(��),ifandonlyif

Obviously,if ��=ℎ(��),theproperty(P)resultsbyintegratingbypartsasmanytimesas necessary.

Conversely,letusdenote ℎ(��)=

Thenℎ(��) =��andℎ(��)=ℎ′(��)=...ℎ(��−1)(��)=0,andtheequalitiesintheproperty(��),in whichweput��=ℎ(��) andintegratebyparts,giveussuccessivelyℎ(��−1)(��)= =ℎ′(��)= ℎ(��)=0

Returningtotheproposedproblem,letusnote

(��)=��(��)−(��0+��1��+ +����−1

)

anddeterminethecoefficients ��0,��1,...����−1 sothatthefunction �� verifiestheequalitiesof theproperty (��)

TheorthogonalizationpropertiesinaEuclideanvectorspaceensurethepossibilityof thisdetermination.Thenwecantake ℎ(��) =�� andget

Theresultis ��(��)=0, ∀��∈[��,��],i.e., �� isapolynomialofdegreeatmost ��−1.

Remark1.2: Thepreviousresultillustratesonaconcretesituationtheequality ��⊥⊥ = ��,foranyclosedsubspaceofaHilbertspace.

Letusexaminethedetectionofcriticalpoints(extremals)forthefunctionalswhose variationsweredeterminedintheprevioussection.

Example1.7: Obviously,anonzerocontinuouslinearfunctional �� ∶ �� → R hasno criticalpoints.

Example1.8: Forthefunctionalsoftype

(��())=∫

(��,��(��))����, thevariationisgivenbyformula(1).Then,applyingthefundamentallemmaweobtain

(7)

Equation(7)canbesolved,accordingtotheimplicitfunctiontheorem,obtainingsolutions ��=��(��),aroundthepoints (��0,��0) atwhich

Example1.9: Aspecialplace,genericinthevariationalcalculus,hasthefunctional

where �� isa ��2 Lagrangian.Thevariationofthisfunctionalisgiveninformula(2):

Integratingthesecondtermbyparts,thevariationtakestheshape

For �� tobeanextremumwemusthave ����(��,ℎ)=0,identicalwithrespectto ℎ(��).Since amongthevariations ℎ arealsothoseforwhich ℎ(��)=ℎ(��)=0,usingtheremarksatthe fundamentallemma,weobtainthatthefunction ��(��) whichperformstheextremummust checktheODE(atmostofsecondorderintheunknown ��(��))

ThisevolutionODEiscalledEuler–LagrangeODE;itssolutions ��(��) arecalledextremals (criticalpoints)ofthegivenfunctional.

Sothefunctions ��(��) thatperformanextremumofthefunctionalarefoundamongthe extremalsofthefunctional.

TheEuler–Lagrangeequation(8)iswrittenexplicitly

IfthisODEwiththeunknown ��(��) isofsecondorder,thenthegeneralsolutiondependson twoarbitraryconstants,soastochooseacertainsolutionwemustimposeconditionsatthe boundary.

Fixedendsextremals.Weimposetoextremalsgivenvaluesattheendsoftheintegration interval: ��(��)= ��1,��(��)= ��2.Inthiscaseweonlyhave ℎ(��)= ℎ(��)=0 andthenecessaryconditionofextremumisreducedtoEuler–LagrangeODE(8).This,togetherwiththe conditionsattheends,generallydeterminesauniqueextremal.

Extremalswithfreeordinatesattheends. Letussupposethatonlythepoint ��(��)= ��1 isfixed.Then,theEuler–LagrangeODEimposesthecondition ���� ����′ ℎ(��)=0.Since ℎ(��)

isarbitrary,itisobtainedasecondcondition ���� ����′(��,��(��),��′(��))=0.Inthiswayweobtaina uniqueextremal.Thisequalityiscallednaturalboundarycondition. Obviously,ifweleavetheordinatesfreeatbothends,wewillgettwonaturalboundary conditions.

Extremalswithendsongivencurves. Letussupposeagain ��(��)=��1,andtheright endoftheextremalbevariableonagivencurve ��=��(��).Inthiscasethefunctional ��(��(⋅)) becomesanintegralwithavariableupperlimit.Forthefunctionalvariationweobtain, applyingtheformulaforderivinganintegralwithaparameter,

Toevaluate

Takingthederivativewithrespectto �� andsetting ��=0,weobtain

Then,afteranintegralbypartsandselecting ℎ(��)=0,wefind

Sincethevariation ℎ isarbitrary,weobtain,theEuler–LagrangeODE(8)andthenthe transversalitycondition

,i.e.,theextremalandthecurve ��=��(��) aretransversalatcontactpoint.)

Obviously,iftheleftendoftheextremalgraphisvariableonagivencurve��=��(��),wewill haveatthisendaswellasimilarconditionoftransversality

1ExtremaofDifferentiableFunctionals

Theresultsextendtofunctionalwithhigher-orderderivatives,

startingfromtheexpression(3)ofvariation.Integratingbypartsseveraltimesandusingthe fundamentallemma,theEuler–Lagrangeequationtakestheshape

anODEoforderatmost 2��,withunknown ��(��).

Example1.10: Forafunctionalthatdependonseveralfunctionvariables

theextremals ��

aresolutionsofEuler–Lagrangesystem

Theproblemstowhichtheextremalsareconstrainedattheendsoftheintervalareposedand solvedasinthepreviouscase.

Example1.11: Incaseofdouble(multiple)integralfunctional

thevariationisgivenbytherelation(5).IntegratingbypartsandusingGreenformulawe obtain

where ��Ω representstheboundaryof Ω Ifthefunction �� isanextremumpoint,wemusthave ����(��,ℎ)=0,identicallywith respectto ℎ.Consideringthatwehavenullvariations ℎ ontheboundary ��Ω,weapplythe

fundamentallemmaaswellastheaccompanyingremarks.Wefindthatthefunction��(��,��) mustverifytheEuler–Lagrange(Euler-Ostrogradski)PDE

ThesolutionsofthisPDEarecalledextremalsurfaces;anyextremumpointisanextremal surface.

Example1.12: Nowletexaminetheextremalsofcurvilinearintegralfunctional

where Γ isapiecewise ��1 curvewhichjoinstwofixedpoints ��,�� inacompactdomain Ω⊂ R2.Thevariationofthisfunctionalisgivenbyformula(6)which,afterintegrationbyparts, becomes

Tohaveanextremumitisnecessarytocancelthevariation ����(��,ℎ) identicallywithrespect to ℎ.Forthispurposeweusevariations ℎ,with ℎ(��)=ℎ(��)=0,forwhich

Theabovesystemalwayshasasolutionbeingoftypeintegralfactor.Forlargerthan 2 dimensions,thesituationisabitmorecomplicatedandwillbesolvedinthefollowing chapters.

Theconditionofcriticalpointbecomes

foranyvariation ℎ inthepreviousconditions.

Wewillhavetodistinguishtwosituationsasintegralistakenonafixedcurveortheintegral ispath-independent.Wewillmake,forabbreviation,thefollowingnotations:

1,����2).

a)IfthecurvilinearintegralistakenonafixedpathΓ∶��=��(��),��=��(��),and��=(��,��) isthevelocityfieldofthecurve,theconditionofcriticalpointistheorthogonalityon Γ ofthe vectorfield ���� and ��,i.e.,

b)Ifthecurvilinearfunctionalispath-independent,theconditionofcriticalpoint(13)must besatisfiedforanyvector ��=(̇ ��, ̇ ��) andthenweobtainthePDEssystem

withtheunknownfunction ��(��,��).

1.3SecondVariation;SufficientConditionsforExtremum

Thethirdstepinsolvingvariationalproblemsisthechoicebetweenthecriticalpointsof thosewhoactuallyachieveanextremum.Inconcretesituations,comingfromgeometry, physics,engineeringoreconomics,therealizationofextremaisimposedbythenatureof theproblem.

Theoreticallyspeaking,weneedtofindsufficientconditionsforextremum.Tothisend wedefinethesecondvariationofafunctional.

Letusgobacktothedefinitionofdifferentiability,Definition1.1,andsupposethatthe function�� isderivableoveritsentiredomainofdefinition��.Thenthederivativeoperator ����(��, ) becomesafunctionbetweentwonormedspaces

��(⋅ , ⋅)∶��⊂��→ℒ(��,��),��↦����(��,⋅), where ℒ(��,��) isthenormedspaceoflinearboundedoperatorsfrom �� to ��.Ifthisfunctionisdifferentiableatapoint ��∈��,thederivativeat �� ofthefunction ���� iscalled the secondderivativeofthefunction �� atthepoint ��.Thesecondderivativeisabilinearand continuousoperatorfrom ��×�� to �� or,byidentifyingvariations,aquadraticoperator. Specifically,inthecaseofthefunctionalswedealtwithpreviously,wehavethefollowing results.

1.3SecondVariation;SufficientConditionsforExtremum 15

Definition1.2: Ifthevariation ����(��,ℎ) isdifferentiableasfunctionof ��,itsvariation ��(����(��,ℎ))(��) iscalledthesecondvariationofthefunctional �� andisdenotedby ��2��(��;ℎ,��)

Proposition1.3: Ifthefunctional �� istwicedifferentiable,then ��2��(��;ℎ,ℎ)= ��2 ����2 ��(��+��ℎ)| | | | |��=0 .

Ontheotherhandsupposethatthedeviationfromthelinearapproximation, ��(��,ℎ), smallrelativeto ℎ,canbeapproximatedwithasquareoperator,i.e., ��(��,ℎ)=��(ℎ,ℎ)+��1(��,ℎ), lim ℎ→0 ��1(��,ℎ) ‖ℎ‖2 =0, where �� isasymmetricbilinearoperator.Thenthefollowingpropertyholds: Second-orderTaylorformula Wehavetheequality ��(ℎ,ℎ)=1 2 ��2��(��;ℎ,ℎ) andconsequently

△��(��(⋅))=��(��+ℎ)−��(��)=����(��,ℎ)+1 2 ��2��(��;ℎ,ℎ)+��1(��,ℎ)

Nextwewillabbreviate ��2��(��;ℎ,ℎ) by ��2��(��,ℎ) If �� isacriticalpoint(anextremal),i.e., ����(��,ℎ)=0,thenatureofthecriticalpointis givenbythequadraticfunction ��2��(��,ℎ).Asinthecaseofrealfunctionsofrealvariables, itholdsthefollowing

Theorem1.1: a)Necessarycondition Iftheextremal �� isaminimumpointofthe functional ��,then ��2��(��,ℎ)≥0 forany ℎ b)Sufficientcondition Iffortheextremal �� thecondition ��2��(��,ℎ)>��‖ℎ‖2 (��>0 ����������) holdsforany ℎ,then �� istheminimumpointforthefunctional ��

Remark1.3: Conditionb)cannotbeweakenedby ��2��(��,ℎ) > 0.Forinstance,ifwe considerthefunctional ��(��(⋅))=∫ 1 0 ��2(��)(�� ��(��))���� onthespace ��0[0,1],thenthenullfunction ��(��)=0,∀��∈[0,1],isacriticalpointand ��2��(��=0,ℎ)=∫ 1 0 2��ℎ2(��)����>0

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.
Get Variational calculus with engineering applications constantin udriste PDF ebook with Full Chapte by Education Libraries - Issuu