More limiting distributions for eigenvalues of Wigner matrices

Abstract: The Tracy-Widom distributions are among the most famous laws in probability theory, partly due to their connection with Wigner matrices. In particular, for

A = f r a c 1 s q r t n (a_{i j})_{1 l e q i, j l e q n} i n m a t h b b R^{n i m e s n}

symmetric with

(a_{i j})_{1 l e q i l e q j l e q n}

i.i.d. standard normal, the fluctuations of its largest eigenvalue

l a m b d a_{1} (A)

are asymptotically described by a real-valued Tracy-Widom distribution

T W_{1} :

n^{2 / 3} (l a m b d a_{1} (A) - 2) R i g h t a r r o w T W_{1} .

As it often happens, Gaussianity can be relaxed, and this results holds when

m a t h b b E [a_{11}] = 0, m a t h b b E [a_{11}^{2}] = 1,

and the tail of

a_{11}

decays sufficiently fast:

l i m_{x o i n f t y} x^{4} m a t h b b P (| a_{11} | > x) = 0,

whereas when the law of

a_{11}

is regularly varying with index

a l p h a i n (0, 4),

c_{a} (n) n^{1 / 2 - 2 / a l p h a} l a m b d a_{1} (A)

converges to a Fr'echet distribution for

c_{a} : (0, i n f t y) o (0, i n f t y)

slowly varying and depending solely on the law of

a_{11} .

This paper considers a family of edge cases,

l i m_{x o i n f t y} x^{4} m a t h b b P (| a_{11} | > x) = c i n (0, i n f t y),

and unveils a new type of limiting behavior for

l a m b d a_{1} (A) :

a continuous function of a Fr'echet distribution in which

2,

the almost sure limit of

l a m b d a_{1} (A)

in the light-tailed case, plays a pivotal role:

Recommendations

Cites work

Cited in

(9)

This page was built for publication: More limiting distributions for eigenvalues of Wigner matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6102955)