Convergence of the probability of large deviations in a model of correlated random variables having compact-supportQ-Gaussians as limiting distributions

have the same distribution. This distribution, which was introduced by Rodr'iguez et al (2008) and then generalized by Hanel et al (2009), is scale-invariant and depends on a real parameter

u > 0

(

u o i n f t y

implies independence). Putting

S_{n} = X_{1} + c d o t s + X_{n}

, the distribution of

S_{n} - n / 2

approaches a

Q

-Gaussian distribution with compact support (

Q = 1 - 1 / (u - 1) < 1

) as

n

increases, after appropriate scaling. In the present article, we show that the distribution of

S_{n} / n

converges, as

n o i n f t y

, to a beta distribution with both parameters equal to

u

. In particular, the law of large numbers does not hold since, if

0 l e x < 1 / 2

, then

m a t h b b P (S_{n} / n l e x)

, which is the probability of the event

S_{n} / n l e x

(large deviation), does not converges to zero as

n o i n f t y

. For

x = 0

and every real

u > 0

, we show that

m a t h b b P (S_{n} = 0)

decays to zero like a power law of the form

1 / n^{u}

with a subdominant term of the form

1 / n^{u + 1}

. If

0 < x l e 1

and

u > 0

is an integer, we show that we can analytically find upper and lower bounds for the difference between

m a t h b b P (S_{n} / n l e x)

and its (

n o i n f t y

) limit. We also show that these bounds vanish like a power law of the form

1 / n

with a subdominant term of the form

1 / n^{2}

.

Recommendations

Cites work

Cited in

(6)

This page was built for publication: Convergence of the probability of large deviations in a model of correlated random variables having compact-supportQ-Gaussians as limiting distributions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5178181)