高级微观经济学 黄有光 AdMicroL1MathTools.docx
《高级微观经济学 黄有光 AdMicroL1MathTools.docx》由会员分享,可在线阅读,更多相关《高级微观经济学 黄有光 AdMicroL1MathTools.docx(13页珍藏版)》请在冰豆网上搜索。
高级微观经济学黄有光AdMicroL1MathTools
ECC5650MicroeconomicTheory
Topic1:
Set,Topology,RealAnalysisandOptimization
(withthecomplementofAssociateProf.MichaelLi)
Readings:
JR-Chapters1&2,supplementedbyDL-AppendixC
1.1Introduction
Inthislecture,wewillquicklygothroughsomebasicmathematicalconceptsandtoolsthatwillbeusedthroughouttherestofthecourse.Asthisisareviewsession,theattentionwillbemainlyonrefreshingonthelanguage,styleandrigorofmathematicalreasoning.Ineconomicanalysis,especiallymicroeconomicanalysis,mathematicsisalwaystreatedasatool,nevertheend.Ontheotherhand,byintegratingeconomicswithrigorousmathematics,wewillbeabletodevelopthetheoreticalexpositionsinasoundandlogicalmanner,whichiswhyeconomicsisalsoknownaseconomicscience.Notmanyothertraditionallyknownassocialsciencefieldsmanagetopassthiscriticalstage.Butitisimportanttorememberthatasaneconomist,wemustgobeyondthenormalmathematicaltreatmentandtheunderlyingeconomicsandtheirpolicyimplicationsarefarmoreimportantandinteresting.
Theplanofthislecturegoeslikethis.First,wewillreviewthebasicsettheory.Wethenmoveontoabitoftopology.Afterreviewingbasicelementsofrealanalysis,wewillcoversomekeyresultsinoptimization.
1.2BasicsofSetTheory
1.2.1BasicConcepts
∙set:
acollectionofelements
∙setsoperations:
union,intersection
∙realsets:
(thenotionofvectors)
∙:
forany;:
thereexists;:
suchthat;:
belongsto;isanelementof
1.2.2Convexity&Relations
ConvexSet:
∙AsetSRnisconvexif
∙Intuitively,asetisaconvexsetifandonlyif(iff)wecanconnectanytwopointsinastraightlinethatliesentirelywithintheset.
∙Convexsethasnoholes,nobreaks,noawkwardcurvaturesontheboundaries;theyareconsideredas“nicesets”.
∙Theintersectionoftwoconvexsetsremainsconvex.
Relations
∙Foranytwogivensets,SandT,abinaryrelationRbetweenSandTisacollectionoforderedpairs(s,t)withsSandtT.
∙ItisclearthatRisasubsetofST:
(s,t)RorsRt.
PropertiesofRelations:
∙Completeness
∙RSSiscompleteiffforallxandy(xy)inS,xRyoryRx.
∙Reflexivity
∙RSSisreflexiveifforallxinS,xRx.
∙Transitivity
∙RSSistransitiveifforallx,y,zinS,xRyandyRzimpliesxRz.
1.3Topology
Topologyattemptstostudythefundamentalpropertiesofsetsandmappings.OurdiscussionwillbemainlyontherealspaceRn.
∙Arealtopologicalspaceisnormallydenotedas(Rn,d),wheredisthemetricdefinedontherealspace.Intuitivelyspeaking,disadistancemeasurebetweentwopointsintherealspace.
∙EuclideanspacesarespecialrealtopologicalspacesassociatedwiththeEuclideanmetricdefinedasfollows:
1.3.1SetsonaRealTopologicalSpace
-Balls
∙Open-Ballforapointx0:
for>0,
∙Closed-Ballforapointx0:
for>0,
OpenSets
∙SRnisopensetif,xS,>0suchthat()B(x)S.
∙PropertiesofOpenSets:
∙Theemptysetandthewholesetareopenset
∙Unionofopensetsisopen;intersectionofopensetsisopentoo.
∙Anyopensetcanberepresentedasaunionofopenballs:
where
.
ClosedSets
∙Sisaclosedsetifitscomplement,Sc,isanopenset.
∙ApointxSisaninteriorpointifthereissome-ballcenteredatxthatisentirelycontainedinS.ThecollectionofallinteriorpointsofSisdenotedbyintS,knownastheinteriorofS.
∙PropertiesofClosedSets:
∙Theemptysetandthewholesetareclosed;
∙Unionofanyfinitecollectionofclosedsetsisaclosedset;
∙Intersectionofclosedsetsisaclosedset.
CompactSets
∙AsetSisboundedif>0suchthat():
SB(x)forsomexS.
∙AsetinRnthatisclosedandboundediscalledacompactset.
1.3.2Functions/MappingsonRn
∙LetDRm,f:
DRn.Wesayfiscontinuousatthepointx0Dif
>0,>0
∙SpecialCase:
DR,f:
DR.fiscontinuousatx0Dif
>0,>0
PropertiesofContinuousMappings:
∙LetDRm,f:
DRn.Then
∙fiscontinuousforallopenballBRn,f--1(B)isopeninD
forallopensetSRn,f--1(S)isopeninD
∙IfSDiscompact(closedandbounded),thenitsimagef(S)iscompactinRn.
1.3.3WeierstrassTheorem&TheBrouwerFixed-PointTheorem
Thesetwotheorems,knownasexistencetheorems,areveryimportantinmicroeconomictheory.“Anexistencetheorem”specifiesconditionsthat,ifmet,somethingexists.Inthemeantime,pleasekeepinmindthattheconditionsintheexistencetheoremsarenormallysufficientconditions,meaningthatiftherequiredconditionsareNOTmet,itdoesnotmeanthenonexistenceofsomething–itmaystillexist.Theexistencetheoremssayverylittleaboutexactlocationofthissomething.Inotherwords,existencetheoremsarepowerfultoolsforshowingthatsomethingisthere;butitisnotsufficientinactuallyfindingtheequilibrium.
WeierstrassTheorem–ExistenceofExtremeValues
∙Thisisafundamentalresultinoptimizationtheory.
∙(WeierstrassTheorem)Letf:
SRbeacontinuousreal-valuedmappingwhereSisanonemptycompactsubsetofRn.Thenaglobalmaximumandaglobalminimumexist,namely,
TheBrouwerFixed-PointTheorem
Manyprofoundquestionsaboutthefundamentalconsistencyofmicroeconomicsystemshavebeenansweredbyreformulatingthequestionasoneoftheexistenceofafixedpoint.Examplesinclude:
∙Theviewofacompetitiveeconomyasasystemofinterrelatedmarketsislogicallyconsistentwiththissetting;
∙Thewell-knownMinimaxTheoremingametheory
∙(BrouwerFixed-PointTheorem)LetSRnbeanonemptycompactandconvexset.Letf:
SSbecontinuousmapping.ThenthereexistsatleastonefixedpointoffinS.Thatis,x*Ssuchthatx*=f(x*).
1.4Real-ValuedFunctions
∙Bydefinition,areal-valuedfunctionisamappingfromanarbitrarysetD(domainset)ofRntoasubsetRofthereallineR(rangeset).
∙f:
DR,withDRn&RR.
Increasing/DecreasingFunctions:
∙Increasingfunction:
f(x0)f(x1)wheneverx0x1;
∙Strictlyincreasingfunction:
f(x0)>f(x1)wheneverx0>x1;
∙Stronglyincreasingfunction:
f(x0)>f(x1)wheneverx0x1andx0x1
∙Similarly,wecandefinethethreetypesofdecreasingfunctions.
ConcavityofReal-ValuedFunctions
∙Assumptionf:
DR,withDRnisconvexsubsetofRn&RR.
∙f:
DRisconcaveifforallx1,x2D,
∙Intuitivelyspeaking,afunctionisconcaveiffforeverypairofpointsonitsgraph,thechordbetweenthemliesonorbelowthegraph.
∙f:
DRisstrictconcaveifforallx1x2inD,
∙f:
DRisquasiconcaveifforallx1,x2D,
∙f:
DRisstrictlyquasiconcaveifforallx1x2inD,
ConvexityofReal-ValuedFunctions
∙Afterthediscussionofconcavefunctions,wecantakecareoftheconvexfunctionsbytakingthenegativeofaconcavefunction.
∙f:
DRisconvexifforallx1,x2D,
∙f:
DRisstrictconvexifforallx1x2inD,
∙f:
DRisquasiconvexifforallx1,x2D,
∙f:
DRisstrictlyquasiconvexifforallx1x2inD,
PropertiesofConcave/ConvexFunctions
∙f:
DRisconcavethesetofpointsbeneaththegraph,i.e.,{(x,y)|xD,f(x)y}isaconvexset.
∙f:
DRisconvexthesetofpointsabovethegraph,i.e.,{(x,y)|xD,f(x)y}isaconvexset.
∙f:
DRisquasiconcavesuperiorsets,i.e.,{x|xD,f(x)y}areconvexforallyR.
∙f:
DRisquasiconvexinferiorsets,i.e.,{x|xD,f(x)y}areconvexforallyR.
∙Iffisconcave/convexfisquasiconcave/quasiconvex;
∙f(strictly)concave/quasiconcave-f(strictly)convex/quasiconvex.
∙Letfbeareal-valuedfunctiondefinedonaconvexsubsetDofRnwithanonemptyinterioronwhichfisatwicedifferentiablefunction,thenthefollowingstatementsareequivalent:
∙Iffisconcave.
∙TheHessianmatrixH(x)isnegativesemidefiniteforallxinD.
∙Forallx0D,f(x)f(x0)+f(x0)(x–x0),xD.
HomogeneousFunctions
∙Areal-valuedfunctionf(x)iscalledhomogeneousofdegreekif
.
PropertiesofHomogeneousFunctions:
∙fishomogeneousofdegreek,itspartialderivativesarehomogeneousofdegreek–1.
∙(Euler’sTheorem)f(x)ishomogeneousofdegreekiff
1.5IntroductiontoOptimization
∙Wewillfocusonreal-valuedfunctionsonly.
MainConceptsofOptima
∙Localminimum/maximum
∙Globalminimum/maximum
∙Interiormaxima,boundarymaxima
1.5.1UnconstrainedOptimization
First-Order(Necessary)ConditionforLocalInteriorOptima
∙Ifthedifferentiablefunctionf(x)reachesonalocalinteriormaximumorminimumatx*,thenx*solvesthesystemofsimultaneousequations:
f(x*)=0.
Second-Order(Necessary)ConditionforLocalInteriorOptima
Letf(x)betwicedifferentiable.
1.Iff(x)reachesalocalinteriormaximumatx*,thenH(x*)isnegativesemidefinite.
2.Iff(x)reachesalocalinteriorminimumatx*,thenH(x*)ispositivesemidefinite.
Notes:
∙Thereisasimplemethodincheckingwhetheramatrixisanegative(positive)semidefinte,whichistoexaminethesignsofthedeterminantsoftheprincipleminorsforthegivenmatrix.
Local-GlobalOptimizationTheorem
∙Foratwicecontinuouslydifferentiablereal-valuedconcavefunctionfonD,thefollowingthreestatementsareequivalent,wherex*isaninteriorpointofD:
1.f(x*)=0.
2.fachievesalocalmaximumatx*.
3.fachievesaglobalmaximumatx*.
StrictConcavity/ConvexityandUniquenessofGlobalOptima
∙Ifx*maximizesthestrictlyconcave(convex)functionf,thenx*istheuniqueglobalmaximizer(minimizer).
1.5.2ConstrainedOptimi