Summary. We point out a simple argument relying on hypercontractivity to describe tail inequalities on the distribution of the largest eigenvalues of random matrices at the rate given by the Tracy-Widom distribution. The result is illustrated on the known
2MichelLedoux
For xedN,asaLipschitzfunctionoftheGaussianentriesofMN,thelargesteigenvalueλNmaxsatis estheconcentrationinequalityarounditsmean
N 2N Pλmax E(λmax) r 2e 2Nr(1)
foreveryr 0(cf.[15]).Thisresulthoweverdoesnotyieldthe uctuationrateN2/3thatrequiresmorere nedtools,relyingusuallyondelicatePlancherel–RotachasymptoticsforHermitepolynomialsinvolvingtheAiryfunction.Theaimofthisnoteisactuallytopointoutasimpleargument,basedonhyper-contractivity,toreachthenormalizationN2/3andtorecovertailinequalitiesforthelargesteigenvaluesofsomeinvariantensemblesofinterest.
Thestartingpointisthewell-knownfact(see[17,5])thatthedistributionNoftheeigenvalues(λN1,...,λN)oftheGUEhasdensity
1
N 2xdxfdet2hi 1(xj)1 i,j Ne |x|/2
Ni=1N
2
Inparticularthus,foreveryε 0,√NN 1 k=0h2k(x)γ(dx).(3)
N[1+ε,∞)=PλN 1+ε Nmmax ∞√2
Interpolated inequalities between exponential and Gaussian, Orlicz hypercontractivity and i We introduce and study a notion of Orlicz hypercontractive semigroups....
Interpolated inequalities between exponential and Gaussian, Orlicz hypercontractivity and i We introduce and study a notion of Orlicz hypercontractive semigroups....
热门文档