HW Solutions, 2.1
2.1, 22 Show that if the columns of B are linearly dependent, then so are the columns of AB.
SOLUTION 1: Let B = [b1 , . . . bk ]. If the columns of B are linearly dependent, then
there is a set of constants c1 , . . . , ck , not all zero, so that
c1 b1 + c2 b2 + · · · + ck bk = 0
We note that AB is formed as the matrix:
AB = A[b1 , . . . bk ] = [Ab1 , Ab2 , . . . Abk ]
so that these are the columns of AB. Now, if we multiply the previous equation by A,
we must have constants c1 , . . . , ck , not all zero, so that:
c1 Ab1 + c2 Ab2 + · · · + ck Abk = 0
Therefore, the columns of AB are linearly dependent as well. And, as we noted in class,
this implies that the dependence relation in the columns of B are exactly the same
as in AB.
Alternative Solution: We could use a theorem about linear dependence instead of the
definition. In that case, if the columns of B are linearly dependent, there is a nontrivial
solution x so that
Bx = 0.
Multiply both side by A, and we see that x is a nontrivial solution to:
(AB)x = 0
so the columns of AB are linearly dependent.
2.1, 24 Suppose that AD = Im . Show that for any b ∈ IRm , the equation Ax = b has a solution.
SOLUTION: We show that x = Db:
Ax = A(Db) = (AD)b = Ib = b
1