Limit theory for the largest eigenvalues of sample covariance matrices with heavy-tails

Abstract: We study the joint limit distribution of the

k

largest eigenvalues of a

p i m e s p

sample covariance matrix

X X^{T}

based on a large

p i m e s n

matrix

X

. The rows of

X

are given by independent copies of a linear process,

X_{i t} = s u m_{j} c_{j} Z_{i, t - j}

, with regularly varying noise

(Z_{i t})

with tail index

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

. It is shown that a point process based on the eigenvalues of

X X^{T}

converges, as

n o i n f t y

and

p o i n f t y

at a suitable rate, in distribution to a Poisson point process with an intensity measure depending on

a l p h a

and

s u m c_{j}^{2}

. This result is extended to random coefficient models where the coefficients of the linear processes

(X_{i t})

are given by

c_{j} (h e t a_{i})

, for some ergodic sequence

(h e t a_{i})

, and thus vary in each row of

X

. As a by-product of our techniques we obtain a proof of the corresponding result for matrices with iid entries in cases where

p / n

goes to zero or infinity and

a l p h a i n (0, 2)

.

Recommendations

Cites work

Cited in

(23)

This page was built for publication: Limit theory for the largest eigenvalues of sample covariance matrices with heavy-tails