As the columns of
AT
A
are simply the rows of
AA
we call
RaAT
Ra
A
the row space of
AT
A
. More precisely
- Definition 1:
Row Space
The row space of the m-by-n matrix
A
A is simply the span of
its rows, i.e.,
RaAT≡
{ATy|y∈
ℝ
m
}
Ra
A
A
y
y
ℝ
m
This is a subspace of
ℝ
n
ℝ
n
.
Let us examine the matrix:
A=0100-10100001
A
0100
-1010
0001
(1)
The row space of this matrix is:
RaAT={
y
1
0100+
y
2
-1010+
y
3
0001|y∈
ℝ
3
}
Ra
A
y
ℝ
3
y
1
0100
y
2
-1010
y
3
0001
(2)
As these three rows are linearly independent we may go no
further. We "recognize" then
RaAT
Ra
A
as a three dimensional
subspace
of
ℝ
4
ℝ
4
.
Regarding a basis for
RaAT
Ra
A
we recall that the rows of
A
red
A
red
, the
row reduced form
of the matrix
A
A,
are merely linear combinations of the rows of
A
A and hence
RaAT=Ra
A
red
Ra
A
Ra
A
red
(3)
This leads immediately to:
- Definition 2:
A Basis for the Row Space
Suppose
A
A is m-by-n. The pivot rows of
A
red
A
red
constitute a basis for
RaAT
Ra
A
.
With respect to our example,
0100-10100001
0100
-1010
0001
(4)
comprises a basis for
RaAT
Ra
A
.