You are on page 1of 3

1

Solution: Problem Set - 4


Question 1:
Let w1 , w2 be any matrices in W given as
   
a1 b1 a2 b2
w1 = ∈ W, w2 = ∈W
b1 c1 b2 c2
We need to satisfy the following 3 conditions to prove that W is a subspace.

i) 0 ∈ W
 
0 0
0= ∈W
0 0
ii) w1 + w2 ∈ W
     
a1 b1 a b2 a + a2 b1 + b2
w1 + w2 = + 2 = 1 ∈W
b1 c1 b2 c2 b1 + b2 c1 + c2
iii) α · w1 ∈ W
   
a1 b1 α · a1 α · b1
α · w1 = α = ∈W
b1 c 1 α · b1 α · c1
To verify that given matrices B1 , B2 , B3 form a basis of W , we need to satisfy the following 2 conditions.

i) span{B1 , B2 , B3 } = W
Let a matrix w ∈ W is written as,
       
a b 1 0 0 1 0 0
w= =a +b +c = a · B1 + b · B2 + c · B3
b c 0 0 1 0 0 1
Hence, w ∈ span{B1 , B2 , B3 } ⇒ W = span{B1 , B2 , B3 }.

ii) Linear Independence: a1 · B1 + a2 · B2 + a3 · B3 = 0 only if a1 = a2 = a3 = 0


       
1 0 0 1 0 0 a a2
a1 · B1 + a2 · B2 + a3 · B3 = a1 + a2 + a3 = 1
0 0 1 0 0 1 a2 a3
is 0 if all its entries are 0, hence, a1 = a2 = a3 = 0. This implies linear independence, which verify these matrices to be basis
of W .

Observe
       
1 −2 1 0 0 1 0 0
= (1) + (−2) + (3) = 1 · B1 − 2 · B2 + 3 · B3
−2 3 0 0 1 0 0 1
Hence, the coordinates are (1, −2, 3).

Question 2:
(a) Let B = {b1 , b2 , . . . , bm } be a basis for S i.e., dim(S) = m. Since B ⊂ S ⊂ V and B is linearly independent, V catches
at least m linear independent vectors, which implies dim(V ) ≥ dim(S) = m.

(b) Let n = dim(S) = dim(V ). Let a basis of S is B = {b1 , b2 , . . . , bn }. Since B ⊂ S ⊂ V and B is a set of n linear
independent vectors that are in V as well. Since dim(V ) = n, these vectors must span V as well. That is, B is a basis of V
as well, which implies dim(V ) = dim(S) = n.

(c) A basis for S is a set of independent elements in S. The basis for V is also set of independent elements in V . Then, if
S spans V , it is a basis of V . If not, then there exist some vector y, that can be inserted to enlarge S and making {S, y} a
basis. This process continues until the enlarged S spans V . Hence, every basis of S is contained in some basis of V .

(d) Let V be the vector space of polynomials of degree ≤ 2 and consider the basis {1, 1 + x, x2 } for V . Let S be the subspace
consisting of polynomials of degree ≤ 1, consider the basis {1, x} for S. As {1, x} 6⊂ {1, 1 + x, x2 }, therefore, a basis for V
need not contain a basis for S.
2

Question 3:
(a) We need to satisfy the following conditions to prove that W is a subspace.

i) W = U + V ⇒ 0 + 0 = 0 ∈ W
ii) Let u1 + v 1 , u2 + v 2 ∈ W where u1 , u2 ∈ U and v 1 , v 2 ∈ V .
Since U and V are subspaces, αu1 + βu2 ∈ U and αv 1 + βv 2 ∈ V . Hence,
α(u1 + v 1 ) + β(u2 + v 2 ) = (αu1 + βu2 ) + (αv 1 + βv 2 ) ∈ W
Hence, W is a subspace.

(b) No, it is not always unique. The condition U ∩ V = {0} must be satisfied
Let us first consider the case when U ∩ V 6= {0}. Consider two subspaces
n  n 
a b a b
o o
U= a, b, c ∈ R and V = a, b ∈ R
b c b a
The U ∩ V are the matrices for which c = a.
Example:
     
3 4 1 2 2 2
W = = U1 + V1 = +
4 5 2 3 2 2
     
3 4 2 0 1 4
W = = U2 + V2 = +
4 5 0 4 4 1

Let us now consider the case when U ∩ V = {0}, then every w ∈ W can be expressed in a unique way. Suppose, we can
express w in two ways, given as
w = u1 + v 1 = u2 + v 2
hence,
u1 − u2 = v 1 − v 2
As, U and V are subspaces, we have, u1 − u2 ∈ U and v 1 − v 2 ∈ V . So, u1 − u2 = v 1 − v 2 ∈ U ∩ V .
As U ∩ V = {0}, this means
u1 − u2 = v 1 − v 2 = 0 ⇒ u1 = u2 , v 1 = v 2

(c) Since U and V are subspaces of S, this means p, q ≤ n. As W is also a subspace of S, hence, dim(W ) ≤ n. W is formed
by the addition of U and V , which implies the dimension of S to result in p + q. We have two conditions for this,

i) p+q ≤ n: As basis for U is p-dimensional (u1 , . . . , up ) and basis for V is q dimensional (v1 , . . . , vq ), taking all the combi-
nation to generate basis of W results in dimension of W to be p+q i.e., (u1 +v1 , u1 +v2 , . . . , u1 +vp , u2 +v1 , . . . , . . . , up +vq ).
If p + q ≤ n, the dimension of W will be p + q.

ii) p + q > n: As by (i) dimension of W is p + q. If p + q > n, the dimension of W will be greater than n which contradicts
the condition for subspace. Hence, dimension of W must be less than or equal to dimension of S. In this case the dimension
of W is n.

Combining, (i) and (ii) we have, dim(W ) = min {n, p + q}.

Question 4:
This is false. Let V = R2 and let U1 = span(0, 1), U2 = span(1, 1), W = span(1, 0). Geometrically, U1 , U2 , W are three
mutually distinct lines. Then
U1 + W = U2 + W = R2 but U1 6= U2 .
3

Question 5:
If W1 ⊂ W2 , then since W2 6= V , there must be a α ∈ V such that α 6∈ W2 ( and thus α 6∈ W1 ).
If W2 ⊂ W1 , then since W1 6= V , there must be a α ∈ V such that α ∈6 W1 ( and thus α 6∈ W2 ).
If W1 6⊂ W2 , and W2 6⊂ W1 , then there must exist
α1 ∈ W1 α1 6∈ W2
α2 6∈ W1 α2 ∈ W2
Let α = α1 + α2 . Then, α = α1 + α2 6∈ W1 ∪ W2 . Hence, α 6∈ W1 , α 6∈ W2 .
Because if not, i.e., if α ∈ W1 , then α2 = α − α2 ∈ W1 , a contradiction.

Similarly, for basis b ∈ V . If W1 ⊂ W2 , then since W2 6= V , there must be a b ∈ V such that b 6∈ W2 ( and thus b 6∈ W1 ).
If W2 ⊂ W1 , then since W1 6= V , there must be a b ∈ V such that b 6∈ W1 ( and thus b 6∈ W2 ).
If W1 6⊂ W2 , and W2 6⊂ W1 , then there must exist
b1 ∈ W 1 b1 6∈ W2
b2 6∈ W1 b2 ∈ W2
Let b = b1 + b2 . Then, b = b1 + b2 6∈ W1 ∪ W2 . Hence, b 6∈ W1 , b 6∈ W2 .
Because if not, i.e., if b ∈ W1 , then b2 = b − b2 ∈ W1 , a contradiction.

Yes! It is true for more than 2 nontrivial subspaces. Because, we can extend the earlier parts from 2 sets to more than 2 sets.

Question 6:
Write yi = vi − vi−1 , for i ≤ i ≤ n − 1, yn = vn . Thus, we wish to prove that span(v1 , . . . , vn ) = span(y1 , . . . , yn ).
P P
Since every yi is a linear combination of the vi , every expression ai yi may be rewritten as a sum βi vi . Thus span(v1 , . . . , vn ) ⊂
span(y1 , . . . , yn ).

On the other hand, we may write vn = yn , vn−1 = yn−1 + yn , . . . , v1 = y1 + . . . + yn . Therefore, every vi is a linear
combination of the yi . Thus span(y1 , . . . , yn ) ⊂ span(v1 , . . . , vn ).

Therefore, span(v1 , . . . , vn ) = span(y1 , . . . , yn ) = V . Since dim(V ) = n, {y1 , . . . , yn } must be linearly independent as well.
Hence, it is a basis.

You might also like