Skip to main content

Convergence rate of extremes from Maxwell sample

Abstract

For the partial maximum from a sequence of independent and identically distributed random variables with Maxwell distribution, we establish the uniform convergence rate of its distribution to the extreme value distribution.

MSC:62E20, 60E05, 60F15, 60G15.

1 Introduction

One interesting problem in extreme value theory is to consider the convergence rate of some extremes. For the uniform convergence rate of extremes under the second-order regular variation conditions, see Falk [1], Balkema and de Haan [2], de Haan and Resnick [3] and Cheng and Jiang [4]. For the extreme value distributions and their associated uniform convergence rates for given distributions, see Hall and Wellner [5], Hall [6], Peng et al. [7], Lin and Peng [8] and Lin et al. [9].

In this note, we discuss the uniform convergence rate of extremes from a sequence of independent and identically distributed (iid) random variables with Maxwell distribution (MD). The probability density function of MD is given by

f(x)= 2 π x 2 σ 3 exp ( − x 2 2 σ 2 ) ,x>0.
(1.1)

The MD and the convergence rate of extremes from Maxwell sample have been widely used in the field of physics. We establish the uniform convergence rate of its distribution to the extreme value distribution and give an improved proof for the pointwise convergence rate of MD.

Throughout this paper, let ( ξ n ,n≥1) be a sequence of iid random variables with common distribution F(x)= ∫ 0 x f(t)dt with a probability density function f(x) given by (1.1), and let M n = max 1 ≤ k ≤ n ξ k be the partial maximum. Liu and Fu [10] proved that

lim n → ∞ P ( α n − 1 ( M n − β n ) ≤ x ) =exp ( − exp ( − x ) ) :=Λ(x)

with the normalizing constants α n and β n given by

α n = σ ( 2 log n ) 1 2 , β n = ( 2 σ 2 log n ) 1 2 + σ log ( 2 log n ) + σ log 2 π 2 ( 2 log n ) 1 2 .
(1.2)

By arguments similar to those of Hall [6], Peng et al. [7] and Lin et al. [9], the appropriate normalizing constants a n and b n can be given by the following equations:

a n = σ 2 b n − 1
(1.3)

and

π 2 b n σ exp ( b n 2 2 σ 2 ) =n.
(1.4)

By arguments similar to those of Example 2 of Resnick [11], we have

b n = ( 2 σ 2 log n ) 1 2 + σ log ( 2 log n ) + σ log 2 π 2 ( 2 log n ) 1 2 +o ( ( log n ) − 1 / 2 ) .

Hence

α n / a n →1,( β n − b n )/ a n →0,

implying

lim n → ∞ P( M n ≤ a n x+ b n )= lim n → ∞ F n ( a n x+ b n )=Λ(x),

cf. Leadbetter et al. [12] or Resnick [11].

This paper is organized as follows. Section 2 gives some auxiliary results. In Section 3, we present the main result. Related proofs are given in Section 4.

2 Auxiliary results

To establish the uniform convergence of F n ( a n x+ b n ) to its extreme value distribution Λ(x), we need some auxiliary results. The first result is the decomposition of F(x), which is the following result.

Lemma 1 Let F(x) be the Maxwell distribution function. Then, for x>0, we have

1−F(x)= 2 π x σ ( 1 + σ 2 x 2 ) exp ( − x 2 2 σ 2 ) −r(x)
(2.1)

with

0<r(x)= 2 π ∫ x ∞ σ y 2 exp ( − y 2 2 σ 2 ) dy< 2 π σ 3 x 3 exp ( − x 2 2 σ 2 ) .
(2.2)

For simplicity, throughout this paper, let C be a generic positive constant whose value may change from line to line, and let C i , C i j (i∈N, j∈N) be absolute positive constants.

For the normalizing constants a n , b n defined by (1.3) and (1.4), respectively, let

a n ∗ = a n r n , b n ∗ = b n + a n δ n ,
(2.3)

where r n →1, δ n →0, n→∞. So, a n ∗ / a n →1, ( b n ∗ − b n )/ a n →0, implying F n ( a n ∗ x+ b n ∗ )→Λ(x). For large n, we have the following result.

Lemma 2 Let a n ∗ , b n ∗ be defined by (2.3). For fixed x∈R and sufficiently large n, we have

F n ( a n ∗ x + b n ∗ ) − Λ ( x ) = Λ ( x ) e − x ( ( x 2 2 − x − 1 ) a n b n − 1 + ( r n − 1 ) x + δ n + O [ ( a n b n − 1 ) 2 + ( r n − 1 ) 2 + δ n 2 ] ) .
(2.4)

Proof Note that b n ∼σ ( 2 log n ) 1 2 , which means

a n b n − 1 ∼ 1 2 log n →0.

For large n, we have

2 π a n ∗ x + b n ∗ σ exp ( − ( a n ∗ x + b n ∗ ) 2 2 σ 2 ) = 2 π b n σ ( 1 + a n b n − 1 ( r n x + δ n ) ) exp ( − b n 2 2 σ 2 ) × exp ( − a n 2 ( r n 2 x 2 + δ n 2 + 2 r n δ n x ) 2 σ 2 − ( r n − 1 ) x − x − δ n ) = n − 1 e − x ( 1 − ( x 2 2 − x ) a n b n − 1 − ( r n − 1 ) x − δ n + O [ ( a n b n − 1 ) 2 + ( r n − 1 ) 2 + δ n 2 ] ) .

Since

σ 2 ( a n ∗ x + b n ∗ ) 2 = a n b n − 1 −2x ( a n b n − 1 ) 2 +O ( ( a n b n − 1 ) 3 ) ,

we have

σ 4 ( a n ∗ x + b n ∗ ) 4 = ( a n b n − 1 ) 2 +O ( ( a n b n − 1 ) 3 ) .

Similarly,

σ 5 ( a n ∗ x + b n ∗ ) 5 exp ( − ( a n ∗ x + b n ∗ ) 2 2 σ 2 ) =O ( n − 1 ( a n b n − 1 ) 2 ) .

Hence,

1 − F ( a n ∗ x + b n ∗ ) = n − 1 e − x ( 1 − ( x 2 2 − x − 1 ) a n b n − 1 − ( r n − 1 ) x − δ n + O [ ( a n b n − 1 ) 2 + ( r n − 1 ) 2 + δ n 2 ] ) .
(2.5)

So,

F n ( a n ∗ x + b n ∗ ) − Λ ( x ) = ( 1 − n − 1 e − x ( 1 − ( x 2 2 − x − 1 ) a n b n − 1 − ( r n − 1 ) x − δ n + O [ ( a n b n − 1 ) 2 + ( r n − 1 ) 2 + δ n 2 ] ) ) n − Λ ( x ) = Λ ( x ) e − x ( ( x 2 2 − x − 1 ) a n b n − 1 + ( r n − 1 ) x + δ n + O [ ( a n b n − 1 ) 2 + ( r n − 1 ) 2 + δ n 2 ] ) ,

which is the desired result. □

3 Main results

In this section we present the pointwise convergence rate and the uniform convergence rate of F n (â‹…) to its extreme value distribution under different normalizing constants. The first result is the pointwise convergence of extremes under the normalizing constants given by (1.2).

Theorem 1 Let { ξ n ,n≥1} be a sequence of independent identically distributed random variables with common distribution MD. Then

F n ( α n x+ β n )−Λ(x)∼Λ(x) e − x ( log ( 2 log n ) ) 2 16 log n ,
(3.1)

for large n, where α n , β n are defined in (1.2).

Recently Liu and Fu [10] proved the result, we present an improved proof for the pointwise convergence rate in Section 4.

The following is the uniform convergence rate of extremes under the appropriate normalizing constants a n and b n given by (1.3) and (1.4), which shows that the optimal convergence rate is proportional to 1/logn.

Theorem 2 Let ( ξ n ,n≥1) be a sequence of independent identically distributed random variables with common distribution MD. For large n, there exist absolute constants 0< d 1 < d 2 such that

d 1 log n < sup x ∈ R | F n ( a n x + b n ) − Λ ( x ) | < d 2 log n ,
(3.2)

where a n and b n are defined by (1.3) and (1.4), respectively.

4 Proofs

Proof of Theorem 1 Firstly, we derive the following asymptotic expansions of b n defined by (1.4)

b n = β n +o ( ( log n ) − 1 2 )
(4.1)

and

b n = β n − σ ( log ( 2 log n ) + log 2 π ) 2 16 2 ( log n ) 3 2 + ( σ log 4 log n + log ( 2 log n ) + log 2 π 4 log n ) / ( 2 log n ) 1 2 + O ( ( log ( 2 log n ) ) 2 ( log n ) 5 2 ) .
(4.2)

Setting b n = β n + θ n and substituting into (1.4), we obtain by taking logarithms that

log π 2 +logσ−log( β n + θ n )+ β n 2 2 σ 2 + β n θ n σ 2 + θ n 2 σ 2 =logn.

So,

( log ( 2 log n ) + log 2 π ) 2 16 log n − log 4 log n + log ( 2 log n ) + log 2 π 4 log n + β n θ n σ 2 + θ n 2 σ 2 − log ( 1 + θ n β n ) = 0 ,
(4.3)

therefore

β n θ n σ 2 ∼− ( log ( 2 log n ) + log 2 π ) 2 16 log n +log 4 log n + log ( 2 log n ) + log 2 π 4 log n ,
(4.4)

which implies

θ n ∼ − σ ( log ( 2 log n ) + log 2 π ) 2 16 2 ( log n ) 3 2 + ( σ log 4 log n + log ( 2 log n ) + log 2 π 4 log n ) / ( 2 log n ) 1 2 .
(4.5)

Once again let

θ n =− σ ( log ( 2 log n ) + log 2 π ) 2 16 2 ( log n ) 3 2 + ( σ log 4 log n + log ( 2 log n ) + log 2 π 4 log n ) / ( 2 log n ) 1 2 + ν n ,

where ν n =o( ( log ( 2 log n ) ) 2 ( log n ) 3 2 ). By similar arguments, we can obtain (4.2).

Note that a n = σ 2 b n , we have

r n −1= α n a n −1∼ log ( 2 log n ) 4 log n , δ n = β n − b n a n ∼ ( log ( 2 log n ) ) 2 16 log n .

Noting a n b n − 1 ∼ 1 2 log n , by Lemma 2, we have

F n ( α n x+ β n )−Λ(x)∼Λ(x) e − x ( log ( 2 log n ) ) 2 16 log n .

The proof is complete. □

Proof of Theorem 2 Letting r n =1, δ n =0 in (2.3) and noting a n b n − 1 ∼ 1 2 log n , and by Lemma 2, there exists an absolute constant d 1 >0 such that

sup x ∈ R | F n ( a n x + b n ) − Λ ( x ) | > d 1 log n .
(4.6)

Thus, in order to obtain the upper bound, we need to prove

(a) sup − c n ≤ x < 0 | F n ( a n x + b n ) − Λ ( x ) | < D 1 a n b n − 1 ,
(4.7)
(b) sup 0 ≤ x ≤ d n | F n ( a n x + b n ) − Λ ( x ) | < D 2 a n b n − 1 ,
(4.8)
(c) sup d n ≤ x < ∞ | F n ( a n x + b n ) − Λ ( x ) | < D 3 a n b n − 1 ,
(4.9)
(d) sup − ∞ < x ≤ − c n | F n ( a n x + b n ) − Λ ( x ) | < D 4 a n b n − 1 ,
(4.10)

where D i >0 (i=1,2,3,4), and

c n =:loglog b n 2 σ 2 >0, d n =:−loglog b n 2 b n 2 − σ 2 >0.

Obviously,

σ ( 2 log n ) 1 2 < b n <σ ( 2 log n ) 1 2 (1+ C 0 )

and

b n − a n c n = b n ( 1 − σ 2 b n 2 c n ) = b n ( 1 − σ 2 b n 2 log log b n 2 σ 2 ) >0.

Define Ψ n (x)=1−F( a n x+ b n ), then

n log ( 1 − F ( a n x + b n ) ) = − n Ψ n ( x ) + n Ψ n ( x ) + n log ( 1 − Ψ n ( x ) ) = − n Ψ n ( x ) − R n ( x ) .
(4.11)

By the following inequality

−x− x 2 2 ( 1 − x ) <log(1−x)<−x(0<x<1),

we have

0< R n (x)=− ( n Ψ n ( x ) + n log ( 1 − Ψ n ( x ) ) ) < n Ψ n 2 ( x ) 2 ( 1 − Ψ n ( x ) ) .

First, suppose that x≥− c n . By (2.1), we have

Ψ n ( x ) ≤ Ψ n ( − c n ) = 1 − F ( b n − a n c n ) < 2 π b n − a n c n σ ( 1 + σ 2 ( b n − a n c n ) 2 ) exp ( − ( b n − a n c n ) 2 2 σ 2 ) < 2 2 π b n σ ( 1 − a n b n − 1 c n ) exp ( − ( b n ) 2 2 σ 2 + c n − a n b n − 1 c n 2 2 ) < 2 n − 1 e c n = 2 n − 1 log b n 2 σ 2 < sup n ≥ n 0 2 log ( C 1 log n ) n < C 11 < 1

with C 1 =2 ( 1 + C 0 ) 2 , implying

inf x ≥ − c n ( 1 − Ψ n ( x ) ) >1− C 11 >0.

Therefore,

0 < R n ( x ) ≤ n Ψ n 2 ( x ) 2 ( 1 − C 11 ) ≤ n Ψ n 2 ( − c n ) 2 ( 1 − C 11 ) < n − 1 ( log ( C 1 log n ) ) 2 2 ( 1 − C 11 ) = n − 1 ( log ( C 1 log n ) ) 2 a n b n − 1 2 ( 1 − C 11 ) a n b n − 1 < n − 1 ( log ( C 1 log n ) ) 2 4 ( 1 − C 11 ) log n a n b n − 1 < C a n b n − 1 .

By 1− e − x <x, x>0, we have

| exp ( − R n ( x ) ) − 1 | < R n (x)<C a n b n − 1 .

Setting A n (x)=exp(−n Ψ n (x)+ e − x ), B n (x)=exp(− R n (x)), we obtain

| F n ( a n x + b n ) − Λ ( x ) | = Λ ( x ) | A n ( x ) B n ( x ) − 1 | = Λ ( x ) | A n ( x ) B n ( x ) − B n ( x ) + B n ( x ) − 1 | ≤ Λ ( x ) | A n ( x ) − 1 | + | B n ( x ) − 1 | < Λ ( x ) | A n ( x ) − 1 | + C a n b n − 1 .
(4.12)

By (2.1) and (2.2), we have

− n Ψ n ( x ) + e − x = − n [ 2 π b n + a n x σ ( 1 + σ 2 ( b n + a n x ) 2 ) exp ( − ( b n + a n x ) 2 2 σ 2 ) − r ( a n x + b n ) ] + e − x = ( 1 + a n b n − 1 x ) e − x C n ( x ) ,

where

C n (x)= ( − 1 − σ 2 ( b n + a n x ) 2 + σ 4 ( b n + a n x ) 4 δ n ( a n x + b n ) ) exp ( − a n b n − 1 x 2 2 ) + ( 1 + a n b n − 1 x ) − 1

with 0< δ n ( a n x+ b n )<1. To prove (4.7), we consider the case of − c n ≤x<0. By e − x >1−x, x>0, we have

C n ( x ) < ( 1 − a n b n − 1 x 2 2 ) ( ( − 1 + σ 4 ( b n + a n x ) 4 ) δ n ( a n x + b n ) ) + ( 1 + a n b n − 1 x ) − 1 < ( 1 − a n b n − 1 x 2 2 ) { − 1 + ( a n b n − 1 ) 2 ( 1 + a n b n − 1 x ) − 4 } + ( 1 + a n b n − 1 x ) − 1 = ( ( 1 + a n b n − 1 x ) − 4 + x 2 2 − x ( 1 + a n b n − 1 x ) − 1 ) a n b n − 1
(4.13)

and

C n ( x ) > ( − 1 − σ 2 ( b n + a n x ) 2 ) exp ( − a n b n − 1 x 2 2 ) + ( 1 + a n b n − 1 x ) − 1 > ( − 1 − σ 2 ( b n + a n x ) 2 ) + ( 1 + a n b n − 1 x ) − 1 > ( − ( 1 + a n b n − 1 x ) − 2 − x ( 1 + a n b n − 1 x ) − 1 ) a n b n − 1 > − ( 1 + a n b n − 1 x ) − 2 .
(4.14)

Hence, for − c n ≤x<0, by combining (4.13) and (4.14) together, we have

| C n ( x ) | < ( ( 1 + a n b n − 1 x ) − 4 + x 2 2 − x ( 1 + a n b n − 1 x ) − 1 + ( 1 + a n b n − 1 x ) − 2 ) a n b n − 1 < ( ( 1 − a n b n − 1 c n ) − 4 + c n 2 2 + c n ( 1 − a n b n − 1 c n ) − 1 + ( 1 − a n b n − 1 c n ) − 2 ) a n b n − 1 < C 21 .

Furthermore, for − c n ≤x<0, we have

| − n Ψ n ( x ) + e − x | < ( 1 + a n b n − 1 x ) e − x | C n ( x ) | < ( ( 1 + a n b n − 1 x ) − 4 + x 2 2 − x ( 1 + a n b n − 1 x ) − 1 + ( 1 + a n b n − 1 x ) − 2 ) e − x a n b n − 1 < ( ( 1 − a n b n − 1 c n ) − 4 + c n 2 2 + c n ( 1 − a n b n − 1 c n ) − 1 + ( 1 − a n b n − 1 c n ) − 2 ) e c n a n b n − 1 < C 22 .

Noting that 0<| e x −1|<|x|( e x +1), x∈R and e − x >1−x+ x 2 /2 for − c n ≤x<0, we have

Λ ( x ) | A n ( x ) − 1 | = Λ ( x ) | exp ( − n Ψ n ( x ) + e − x ) − 1 | < Λ ( x ) | − n Ψ n ( x ) + e − x | ( exp ( − n Ψ n ( x ) + e − x ) + 1 ) < ( e C 22 + 1 ) ( ( 1 + a n b n − 1 x ) − 4 + x 2 2 − x ( 1 + a n b n − 1 x ) − 1 + ( 1 + a n b n − 1 x ) − 2 ) × a n b n − 1 exp ( − e − x − x ) < C 23 a n b n − 1 .

Together with (4.12), we establish (4.7).

Second, we prove (4.8). Note that

C n ( x ) < ( − 1 + σ 4 ( b n + a n x ) 4 δ n ( a n x + b n ) ) ( 1 − a n b n − 1 x 2 2 ) + ( 1 + a n b n − 1 x ) − 1 < ( a n b n − 1 ) 2 ( 1 + a n b n − 1 x ) − 4 + x 2 2 a n b n − 1 − a n b n − 1 x ( 1 + a n b n − 1 x ) − 1 < ( a n b n − 1 ) 2 ( 1 + a n b n − 1 x ) − 4 + x 2 2 a n b n − 1 < ( 1 + x 2 2 ) a n b n − 1
(4.15)

and

C n (x)> ( − ( 1 + a n b n − 1 x ) − 2 − x ( 1 + a n b n − 1 x ) − 1 ) a n b n − 1 .
(4.16)

By (4.15) and (4.16), for 0≤x< d n , we have

| C n ( x ) | < ( 1 + x 2 2 + ( 1 + a n b n − 1 x ) − 2 + x ( 1 + a n b n − 1 x ) − 1 ) a n b n − 1 < ( 2 + x + x 2 2 ) a n b n − 1 .

Hence,

| − n Ψ n ( x ) + e − x | < ( 1 + a n b n − 1 x ) e − x | C n ( x ) | < ( 1 + a n b n − 1 x ) e − x ( 2 + x + x 2 2 ) a n b n − 1 < C 31 a n b n − 1 < C 32 .

Therefore

Λ ( x ) | A n ( x ) − 1 | < Λ ( x ) | − n Ψ n ( x ) + e − x | ( exp ( − n Ψ n ( x ) + e − x ) + 1 ) < C 31 ( e C 32 + 1 ) Λ ( d n ) a n b n − 1 < C 33 a n b n − 1 .
(4.17)

Combining (4.12) and (4.17) together, we can derive that

sup 0 ≤ x ≤ d n | F n ( a n x + b n ) − Λ ( x ) | <( C 12 + C 33 ) a n b n − 1 =: D 2 a n b n − 1 .

Hence (4.8) is proved.

Third, for x≥ d n , we have

sup x ≥ d n ( 1 − Λ ( x ) ) ≤1−Λ( d n )= a n b n − 1 .
(4.18)

By 1− e x <−x, x∈R, we have

1 − F n ( a n d n + b n ) = 1 − exp ( n log F ( a n d n + b n ) ) < − n log F ( a n d n + b n ) = n Ψ n ( d n ) + R n ( d n ) .
(4.19)

By (2.1) and log(1+x)<x, 0<x<1, we have

n Ψ n ( d n ) = n ( 1 − F ( a n d n + b n ) ) < ( 1 + a n b n − 1 d n ) e − d n ( 1 + a n b n − 1 ( 1 + a n b n − 1 d n ) − 2 ) < 2 ( 1 + a n b n − 1 d n ) log b n 2 b n 2 − σ 2 < 2 ( d n + a n − 1 b n ) σ 2 b n 2 − σ 2 a n b n − 1 = ( 1 + σ 2 b n 2 − σ 2 − σ 2 b n 2 − σ 2 log log ( σ 2 b n 2 − σ 2 ) ) a n b n − 1 < C 41 a n b n − 1 .
(4.20)

Noting that R n ( d n )< C 12 a n b n − 1 , and combining (4.18), (4.19), (4.20) and (4.14) together, we have

sup x ≥ d n | F n ( a n x + b n ) − Λ ( x ) | < sup x ≥ d n ( 1 − F n ( a n x + b n ) ) + sup x ≥ d n ( 1 − Λ ( x ) ) < n Ψ n ( d n ) + R n ( d n ) + a n b n − 1 < ( C 41 + C 12 + 1 ) a n b n − 1 = : D 3 a n b n − 1 ,

which is (4.9).

Finally, consider the case of −∞<x<− c n . If a n x+ b n ≤0, then F n ( a n x+ b n )=0. By Λ(−x)< 1 x , x>1, we have

sup x ≤ − b n / a n | F n ( a n x + b n ) − Λ ( x ) | = sup x ≤ − b n / a n Λ(x)≤Λ ( − b n 2 / σ 2 ) < σ 2 b n 2 = a n b n − 1 .

So, we only need to consider the case of a n x+ b n >0. By using the monotonicity of Λ(x), we have

sup x ≤ − c n Λ(x)≤Λ(− c n )= a n b n − 1 .
(4.21)

Noting log(1−x)<−x, 0<x<1 and e − x >1−x, x∈R, and combining (2.1) and (2.2) together, we have

sup x ≤ − c n F n ( a n x + b n ) ≤ F n ( b n − a n c n ) < ( 1 − n − 1 ( 1 − a n b n − 1 c n ) ( 1 − ( a n b n − 1 ) 2 ( 1 − a n b n − 1 c n ) − 4 ) exp ( c n − a n b n − 1 c n 2 2 ) ) n < exp ( − e c n ( 1 − a n b n − 1 c n ) ( 1 − ( a n b n − 1 ) 2 ( 1 − a n b n − 1 c n ) − 4 ) exp ( − a n b n − 1 c n 2 2 ) ) < exp ( − e c n ( 1 − ( a n b n − 1 c n + ( a n b n − 1 ) 2 ( 1 − a n b n − 1 c n ) − 3 + a n b n − 1 c n 2 2 ) ) ) < exp ( − e c n ) exp ( ( a n b n − 1 c n + ( a n b n − 1 ) 2 ( 1 − a n b n − 1 c n ) − 3 + a n b n − 1 c n 2 2 ) e c n ) < C 51 a n b n − 1 .

Together with (4.21), we have

sup − ∞ < x ≤ − c n | F n ( a n x + b n ) − Λ ( x ) | ≤ sup − ∞ < x < − c n F n ( a n x + b n ) + sup − ∞ < x < − c n Λ ( x ) ≤ F n ( b n − a n c n ) + Λ ( − c n ) < ( C 12 + 1 ) a n b n − 1 = : D 4 a n b n − 1 .

This is (4.10). The proof is complete. □

References

  1. Falk A: Rate of uniform convergence of extreme order statistics. Ann. Inst. Stat. Math. 1986, 38(2):245–262.

    Article  MathSciNet  MATH  Google Scholar 

  2. Balkema AA, de Haan L: A convergence rate in extreme-value theory. J. Appl. Probab. 1990, 27: 577–585. 10.2307/3214542

    Article  MathSciNet  MATH  Google Scholar 

  3. de Haan L, Resnick SI: Second order regular variation and rates of convergence in extreme value theory. Ann. Probab. 1996, 24: 97–124.

    Article  MathSciNet  MATH  Google Scholar 

  4. Cheng S, Jiang C: The Edgeworth expansion for distributions of extreme values. Sci. China Ser. A 2001, 4: 427–437.

    Article  MathSciNet  MATH  Google Scholar 

  5. Hall WJ, Wellner JA: The rate of convergence in law of the maximum of an exponential sample. Stat. Neerl. 1979, 33: 151–154. 10.1111/j.1467-9574.1979.tb00671.x

    Article  MathSciNet  MATH  Google Scholar 

  6. Hall P: On the rate of convergence of normal extremes. J. Appl. Probab. 1979, 16: 433–439. 10.2307/3212912

    Article  MathSciNet  MATH  Google Scholar 

  7. Peng Z, Nadarajah S, Lin F: Convergence rate of extremes for the general error distribution. J. Appl. Probab. 2010, 47: 668–679. 10.1239/jap/1285335402

    Article  MathSciNet  MATH  Google Scholar 

  8. Lin F, Peng Z: Tail behavior and extremes of short-tailed symmetric distribution. Commun. Stat., Theory Methods 2010, 39: 2811–2817. 10.1080/03610920903132087

    Article  MathSciNet  MATH  Google Scholar 

  9. Lin F, Zhang X, Peng Z, Jiang Y: On the rate of convergence of STSD extremes. Commun. Stat., Theory Methods 2011, 40: 1795–1806. 10.1080/03610921003695726

    Article  MathSciNet  MATH  Google Scholar 

  10. Liu B, Fu Y: The pointwise rate of extremes for Maxwell distribution. J. Southwest Univ., Nat. Sci. 2013, 5: 86–99.

    Google Scholar 

  11. Resnick SI: Extreme Values, Regular Variation and Point Processes. Springer, New York; 1987.

    Book  MATH  Google Scholar 

  12. Leadbetter MR, Lindgren G, Rootzen H: Extremes and Related Properties of Random Sequences and Processes. Springer, New York; 1983.

    Book  MATH  Google Scholar 

Download references

Acknowledgements

The research was supported by the National Natural Science Foundation of China (11171275) and the Fundamental Research Funds for the Central Universities (XDJK2012C045).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Chuandi Liu.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and permissions

About this article

Cite this article

Liu, C., Liu, B. Convergence rate of extremes from Maxwell sample. J Inequal Appl 2013, 477 (2013). https://doi.org/10.1186/1029-242X-2013-477

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2013-477

Keywords