EE 605: Error Correction Codes

Quiz 1 Solutions and Marking Scheme

Aug 21st 2023
Duration: 1 hour, Total marks: 20

1. Let G be a group. Prove the following.

(a) Let {Hi }i∈I be a collection of subgroups of G. Then the intersection

\
H= Hi
i∈I

is a subgroup of G. [1 mark]
(b) Consider g ∈ G and let < g > denote the cyclic subgroup generated by g. Show
that < g > is the smallest subgroup containing g. [2 marks]
(c) Show that < g > is abelian. [1 mark]

Answer.
(a) Since H is a subset of group G, for H to be a subgroup it needs to satisfy
i. Closure property
If a, b ∈ H ⇒ a, b ∈ Hi ∀i ∈ I
Since ∀i ∈ I, Hi satisfies closure property
=⇒ for any a, b ∈ H, a ∗ b ∈ H
ii. Inverse Element
If a ∈ H ⇒ a ∈ Hi ∀i ∈ I
Since Hi a subgroup, if a ∈ Hi ⇒ a′ ∈ Hi ∀i ∈ I
=⇒ ∀a ∈ H, ∃ a′ ∈ H s.t. a ∗ a′ = e
(1 mark each awarded for proving properties (i) and (ii))
(b) Consider any subgroup A of G that contains g
Since A is a subgroup, it satisfies closure property
=⇒ g i ∈ A ∀i ∈ {1, . . . , n} where n is the order of < g >
=⇒ < g >⊆ A and hence it is the smallest subgroup containing g
(Full marks awarded for any valid argument using closure property to state that all
elements in < g > are contained in any subgroup containing g)
(c) Consider any i, j ∈ {1, . . . , n} where n is the order of < g >
By utilizing the Associative property of group G we have

gi ∗ gj = x
| ∗ x ∗{z· · · ∗ x} ∗ x
| ∗ x ∗{z· · · ∗ x}
i times j times

| ∗ x ∗{z· · · ∗ x} ∗ x
=x | ∗ x ∗{z· · · ∗ x}
j times i times
j i
=g ∗g

1
 Since < g > is a subgroup that satisfies Commutativity, it is Abelian
(Full marks awarded for any valid proof)

2. Let G be a group with operation ∗ and let H be a subgroup of G. For g ∈ G, the set
gH = {g ∗ h : h ∈ H} is called a left coset of H. Prove the following.

(a) For any a ∈ G, the map f : H → aH defined by f (h) = a ∗ h is one-to-one

between H and aH. [1 mark]
(b) For any a ∈ G, the only left coset of H containing a is aH. [2 marks]
(c) Any two left cosets of H in G are either disjoint or identical. [2 marks]
(d) For any a, b ∈ G, we have aH = bH if and only if a−1 b ∈ H. [1 mark]

Answer.

(a) Let us assume ∃ h1 , h2 s.t. f (h1 ) = f (h2 )

=⇒ a ∗ h1 = a ∗ h2
Pre-multiplying with a−1 on both sides, we obtain h1 = h2
Hence f (h) is one-to-one mapping
(Full marks awarded if the claim f (h1 ) = f (h2 ) ⇐⇒ h1 = h2 is proved)
(b) Let us assume there is some other left coset gH contains a
Then ∃ h ∈ H s.t. g ∗ h = a
=⇒ g = a ∗ h−1
=⇒ gH is identical to aH (because h−1 ∈ H as well)
Hence the only left coset of H containing a is set aH
(Full marks will be awarded even if you have argued that the claim is false by
stating gH also contains a if g ̸= a is an element of aH)
(c) From 2(b), we can claim that if two left cosets contain a common element say
a, they are identical. Hence any two left cosets of H in G are either disjoint or
identical.
(Full marks awarded for any valid argument following similar sequence of steps)
(d) Proving aH = bH =⇒ a−1 b ∈ H:
For any a, b ∈ G if aH = bH
=⇒ ∃ h1 , h2 ∈ H s.t. a ∗ h1 = b ∗ h2
=⇒ h1 ∗ h−1 2 = a
−1
∗b
−1
=⇒ a ∗ b ∈ H

Proving a−1 b ∈ H =⇒ aH = bH:

For any a, b ∈ G if a−1 b ∈ H
=⇒ ∃h ∈ H s.t. b = a ∗ h
=⇒ bH = (a ∗ h)H = aH (using associativity and that H satisfies closure)
(0.5 marks awarded for each direction for any valid argument following similar
sequence of steps)

2
3. Show that when n is not prime, the set Zn = {0, 1, 2, . . . , n−1} with mod−n arithmetic
is not a field. [1 mark]
Answer.
Since n is not a prime, ∃ a, b ∈ {2, . . . , n − 1} s.t. a × b = n. Therefore mod − n multipli-
cation of a and b is not a non-zero element and thus non-zero elements in Zn do not form
an abelian group (as closure is not satisfied). Hence the set Zn = {0, 1, 2, . . . , n − 1}
with mod − n arithmetic is not a field.
(No marks awarded if the claim is proved using a specific value rather than a generalized
non-prime value of n)

4. Prove the following :

(a) For any group G with operation ◦, identity element e, and order m, show that for
each x ∈ G we have
| ◦ x ◦{z· · · ◦ x} = e
x
m times

[2 marks]
(b) For any finite field F with operations (+, ∗), order q, and x ∈ F ,
q
| ∗ x ∗{z· · · ∗ x} = x = x
x
q times

[2 marks]

Answer.

(a) Let x be any element in group G. Consider the sequence {x, x2 , x3 , . . . , xm+1 }.
Since G satisfies the closure property, each of these elements belong to the group.
However, |G| = m, while the sequence has m + 1 elements. Therefore, ∃i, j ∈
{1, . . . , m} such that i < j and xi = xj =⇒ xj−i = e. Let k = j − i.

This means that the sub-sequence {x, x2 , . . . , xk = e} is just the sub-group < x >,
and has order k. By coset decomposition (Lagrange Theorem), we have that the
order of the sub-group divides that of the group, i.e., k|m. Therefore,

n/k
xn = xk = en/k = e

(2 marks awarded for any valid proof making use of cyclic subgroup properties)
(b) F is a field under (+, ∗). Therefore, F \ {0} is an Abelian group under ∗ and
|F \ {0}| = q − 1 =⇒ order of this group is q − 1. Using the result proven in
(a), ∀x ∈ F \ {0} we have that xq−1 = e =⇒ xq = x. When x = 0, using the
property that ∀a ∈ F, a ∗ 0 = 0, we can show that 0q = 0.
(1.5 marks for a valid proof that does not consider the case of x = 0. 2 marks for
a fully valid proof.)

3
5. Let x, y be binary length n vectors. Let x ∗ y = (x1 .y1 , x2 .y2 , ..., xn .yn ) be a vector that
has ones exactly in those positions where both x and y have ones. Prove that wt(x+y) =
wt(x) + wt(y) − 2wt(x ∗ y), where wt(.) denotes the Hamming weight. [2 marks]
Answer.
For any binary vector x, let Hx = {i : xi = 1}, i.e., the set of all indices where the
vector has ones. Therefore we have that wt(x) = |Hx |, wt(y) = |Hy |, and wt(x ∗ y) =
|Hx ∩ Hy |. wt(x + y) represents the number of indices where exactly one codeword has
a 1. Therefore, Hx+y = (Hx \ Hy ) ∪ (Hy \ Hx ).

wt(x + y) = |Hx+y | = |Hx \ Hy | + |Hy \ Hx | − |(Hx \ Hy ) ∩ (Hy \ Hx )|

= |Hx | − |Hx ∩ Hy | + |Hy | − |Hx ∩ Hy | − 0
= wt(x) + wt(y) − 2wt(x ∗ y)

(0.5 mark for arguing about or stating the relation between the weight and the distribu-
tion of 1s for each of the involved vectors. 1.5 mark for proving the relation using any
valid counting argument)

6. Consider a binary code (000, 101, 111). Say, one out of the three codewords is cho-
sen uniformly at random, and is transmitted over a memoryless binary channel with the
following error probabilities:

P(1 received |0 sent ) = 0.3 and P(0 received |1 sent ) = 0.2.

Consider the following possible decoding rules:

Maximum Likelihood Decoding: We find the codeword X that gives the maximum-
likelihood P (Y |X) for the given received word Y
Maximum Aposteriori Decoding: We find the codeword X that gives the maximum-
aposteriori probability P (X|Y ) for the given received word Y
Nearest Distance Decoding: We find the codeword X that has least Hamming distance
with the received word Y .
Use the above decoding rules to decode the received word 010. [3 marks]
Answer.

0.7
0 0
0.2
0.3
1 1
0.8

4
Maximum Likelihood Decoding
We decode to that codeword X in the code such that P (Y |X) for the given channel is
maximum.

P (Y = 010|X = 000) = 0.7 × 0.3 × 0.7 = 0.147

P (Y = 010|X = 101) = 0.2 × 0.3 × 0.2 = 0.012
P (Y = 010|X = 111) = 0.2 × 0.8 × 0.2 = 0.032

X̂ = arg max P (Y |X) = 000

X

Maximum Aposteriori Decoding

We decode to that codeword X in the code such that P (X|Y ) for the given channel is
maximum.

P (Y |X)P (X)
P (X|Y ) =
P (Y )

Note that for any codeword X, P (X) = 1/3 (uniformly chosen). Furthermore, P (Y )
does not depend on the choice of the transmitted codeword X and it just depends on the
code and channel. Therefore P (X|Y ) ∝ P (Y |X) and,

X̂ = arg max P (X|Y ) = arg max P (Y |X) = 000

X X

Nearest Distance Decoding

We decode to that codeword X in the code such that d(X, Y ) for the given channel is
minimum.

d(000, 010) = 1
d(101, 010) = 3
d(111, 010) = 2

X̂ = arg max d(X, Y ) = 000

X

(1 mark for calculating the decoded vector in each decoding scheme)