0% found this document useful (0 votes)
60 views4 pages

Solution To Homework #1: (A) (B) (C) (D)

This document contains the solutions to 6 problems from Homework #1 of the ECE 7520 course in Spring 2011. The problems cover topics like entropy of functions, measures of correlation, joint entropy, mixing distributions, relative entropy, and discrete entropies. The solutions provide justifications and calculations to support the answers in concise mathematical steps.

Uploaded by

london
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
0% found this document useful (0 votes)
60 views4 pages

Solution To Homework #1: (A) (B) (C) (D)

This document contains the solutions to 6 problems from Homework #1 of the ECE 7520 course in Spring 2011. The problems cover topics like entropy of functions, measures of correlation, joint entropy, mixing distributions, relative entropy, and discrete entropies. The solutions provide justifications and calculations to support the answers in concise mathematical steps.

Uploaded by

london
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
You are on page 1/ 4

ECE 7520, Spring 2011

Solution to Homework #1

1. Entropy of functions of a random variable. Let X be a discrete random variable. Show


that the entropy of a function of X is less than or equal to the entropy of X by justifying the
following steps (a), (b), (c) and (d):
(a)
H(X, g(X)) = H(X) + H(g(X)|X)
(b)
= H(X)
(c)
H(X, g(X)) = H(g(X)) + H(X|g(X))
(d)
≥ H(g(X))

Hence H(g(X)) ≤ H(X).


Solution:
(a) By the chain rule for entropies.
(b) Given X, g(X) has a fixed value. Hence
X X
H(g(X)|X) = p(x)H(g(X)|X = x) = 0 = 0.
x x

(c) By the chain rule for entropies.


(d) Follows because the (conditional) entropy of a discrete random variable is nonnegative, i.e.,
H(X|g(X)) ≥ 0, with equality iff g(X) is a one-to-one mapping.

2. A measure of correlation. Let X1 and X2 be identically distributed, but not necessarily


independent. Let
H(X1 |X2 )
ρ =1− .
H(X1 )

I(X1 ;X2 )
(a) Show ρ = H(X1 ) .

(b) Show 0 ≤ ρ ≤ 1.
(c) When is ρ = 0?
(d) When is ρ = 1?
Solution:
H(X1 )−H(X1 |X2 ) I(X1 ;X2 )
(a)ρ = H(X1 ) = H(X1 )

(b) 0 ≤ ρ ≤ 1 follows easily because 0 ≤ H(X1 |X2 ) ≤ H(X1 ).

1
(c) ρ = 0 iff I(X1 ; X2 ) = 0, i.e., X1 and X2 are independent.
(d) ρ = 1 iff H(X1 |X2 ) = 0, i.e., X1 is a function of X2 .

3. Example of joint entropy. Let p(x, y) be given by

Y
X 0 1

0 1 1
3 3
1
1 0
3

Find
(a) H(X), H(Y ).
(b) H(X|Y ), H(Y |X).
(c) H(X, Y ).
(d) H(Y ) − H(Y |X).
(e) I(X; Y ).
(f) Draw a Venn diagram for the quantities in (a) through (e).
Solution:
2 3 1 2
(a) H(X) = 3 log 2 + 3 log 3 = log 3 − 3 = 0.918 bits = H(Y ).
(b) H(X|Y ) = 13 H(X|Y = 0) + 23 H(X|Y = 1) = 1
3 ·0+ 2
3 ·1= 2
3 = 0.667 bits = H(Y |X).
1
(c) H(X, Y ) = 3 × 3 log 3 = log 3 = 1.585 bits. Alternatively, H(X, Y ) = H(X) + H(Y |X) =
1.585 bits.
2 2 4
(d) H(Y ) − H(Y |X) = log 3 − 3 − 3 = log 3 − 3 = 0.251 bits.
(e) I(X; Y ) = H(Y ) − H(Y |X) = 0.251 bits.
(f)

H(X) H(Y)

H(X|Y) I(X;Y) H(Y|X)

4. Mixing increases entropy. Show that the entropy of the probability distribution

(p1 , . . . , pi , . . . , pj , . . . , pm )

2
is less than the entropy of the distribution
pi + pj pi + pj
(p1 , . . . , ,..., , . . . , pm ).
2 2

(hint: loge x ≤ x − 1 for x > 0. )


Solution:
pi +pj pi +pj
Let P1 = (p1 , . . . , pi , . . . , pj , . . . , pm ) and P2 = (p1 , . . . , 2 ,..., 2 , . . . , pm ).

pi + pj pi + pj
He (P1 ) − He (P2 ) = −pi ln pi − pj ln pj + 2 ln
2 2
pi + pj pi + pj
= pi ln + pj ln
2pi 2pj
   
pi + pj pi + pj
≤ pi − 1 + pj −1
2pi 2pj
= 0.

5. Relative entropy is not symmetric. Let the random variable X have three possible
outcomes {a, b, c}. Consider two distributions p(x) and q(x) on this random variable

Symbol p(x) q(x)


a 1/2 1/3
b 1/4 1/3
c 1/4 1/3

Calculate H(p), H(q), D(pkq) and D(qkp). Verify that in this case D(pkq) 6= D(qkp).
Solution:

1 1 1
H(p) = log 2 + log 4 + log 4 = 1.5 bits
2 4 4
1
H(q) = 3 × log 3 = 1.58496 bits
3
1 3 1 3 1 3
D(pkq) = log + log + log = log 3 − 1.5 = 0.08496 bits
2 2 4 4 4 4
1 2 1 4 1 4 5
D(qkp) = log + log + log = − log 3 + = 0.0817 bits
3 3 3 3 3 3 3

It is clear that D(pkq) 6= D(qkp).

6. Discrete entropies. Let X and Y be two independent integer-valued random variables. Let
X be uniformly distributed over {1, 2, . . . , 8}, and let P r{Y = k} = 2−k , for k = 1, 2, 3, . . .
(a) Find H(X).

3
(b) Find H(Y ).
(c) Find H(X + Y, X − Y ).
hint: For (b), the following expression may be useful

X r
nr n = .
n=0
(1 − r)2

For (c), if you do not find a direct way, try to use

I(X, Y ; X + Y, X − Y )
= H(X, Y ) − H(X, Y |X + Y, X − Y )
= H(X + Y, X − Y ) − H(X + Y, X − Y |X, Y ).

Solution:
(a) H(X) = log 8 = 3 bits.
1/2
(b) H(Y ) = k 2−k log 2k = k k2−k =
P P
(1−1/2)2
= 2 bits.

(c) Since (X, Y ) → (X + Y, X − Y ) is a one-to-one mapping,

H(X + Y, X − Y ) = H(X, Y ) = H(X) + H(Y ) = 5 bits.

Alternatively, use the hint. It is clear H(X, Y |X + Y, X − Y ) = 0 and H(X + Y, X − Y |X, Y ) = 0.


Hence H(X + Y, X − Y ) = H(X, Y ) = H(X) + H(Y ) = 5 bits.

You might also like