Scribd
Upload
10 views1 page

Homework 6: December 13, 2023

The document discusses properties of characteristic functions and their convergence, leading to conclusions about expectations and variances of random variables. It applies theorems related to the strong law of large numbers and the Lindeberg-Feller theorem to establish relationships between sums of random variables and chi-squared distributions. Additionally, it includes a derivation of conditional probabilities using integrals and expectations.

Uploaded by

rxzhen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
10 views1 page

Homework 6: December 13, 2023

The document discusses properties of characteristic functions and their convergence, leading to conclusions about expectations and variances of random variables. It applies theorems related to the strong law of large numbers and the Lindeberg-Feller theorem to establish relationships between sums of random variables and chi-squared distributions. Additionally, it includes a derivation of conditional probabilities using integrals and expectations.

Uploaded by

rxzhen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Homework 6

December 13, 2023

1. The ch.f. of Xn is φn (t) = exp(−σn2 t2 /2). Since Xn ⇒ X, by Continuity theorem, φn (t) → φ(t)
for all t. And therefore φ is continuous. Hence there exists s > 0 such that |φ(s) − φ(0)| <
1/2. As φn (s) < 1, we have 1/2 < φ(s) ≤ 1. φn (s) = exp(−σn2 s2 /2) → φ(s). Hence, σn2 =
2 2
− 2 log φn (s) → − 2 log φ(s) ∈ [0, ∞).
s s
2. limt↓0 (φ(−t) − 1)/t2 = limt↓0 (φ(t) − 1)/t2 = limt↓0 (φ(t) − 1)/t2 = c. Hence,

φ(t) − 2φ(0) + φ(−t)


lim = 2c > −∞.
t↓0 t2

Then E|X|2 < ∞. By Theorem 3.3.20, φ(t) = 1 + itEX − t2 E(X 2 )/2 + o(t2 ). Hence,

iEX = lim(φ(t) − 1)/t = lim t · (φ(t) − 1)/t2 = 0,


t↓0 t↓0

E|X|2 = −2 lim(φ(t) − 1)/t2 = −2c < ∞.


t↓0

Inparticular, if φ(t) = 1 + o(t2 ), limt↓0 (φ(t) − 1)/t2 = limt↓0 o(t2 )/t2 = 0. Hence, EX = 0 and
E(X 2 ) = 0. X = 0 a.s. Hence, φ(t) ≡ 1.

3. By Theorem 3.4.1, Pnm=1 Xm /σn1/2 ⇒ χ. By the strong law of large numbers


P Pn 2 2
m=1 Xm /σ n→1
1/2 n 2 1/2
P n P n 2 1/2
a.s. Hence, σn /( m=1 Xm ) → 1 a.s. By Converging Together Lemma, m=1 Xm /( m=1 Xm ) ⇒
χ.
p
4. As |Xi | ≤ M , var(Xi ) and var(Sn ) are finite. Let Yn,m = (Xm − EXm )/ var(Sn ), then Yn,m , 1 ≤
m ≤ n are independent random variables with EYn,m = 0.
Pn 2
Pn var(Xm )
m=1 EYn,m = m=1 = 1 > 0. For ϵ > 0, there exists an N such that for n ≥
var(Sn ) p p
2
N, var(Sn ) > (2M/ϵ)
Pn . Hence for n ≥ N , |y n,m | = |X m − EXm |/ var(Sn ) ≤ 2M/ var(Sn ) < ϵ.
2
Hence, limn→∞ m=1 p E(|Yn,m | ; |Yn,m | > ϵ) = 0. By The Lindeberg-Feller theorem, Yn,1 + · · · +
Yn,n = (Sn − ESn )/ var(Sn ) ⇒ χ.

5. R R R
P (A ∩ G) 1A dP E(1 A |G) dP P (A|G) dP
P (G|A) = = RG = RG = RG
P (A) 1 dP
Ω A Ω
E(1A |G) dP Ω
P (A|G) dP

You might also like