Scribd
Upload
66 views2 pages

Lecture Note 5B

1. The row reduced echelon form (RREF) of two row equivalent matrices is unique and independent of the row operations used to obtain it. 2. The RREF of a submatrix of a matrix A is equal to the corresponding columns of the RREF of A. 3. A matrix is invertible if and only if its RREF is the identity matrix.

Uploaded by

RAJ MEENA
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
66 views2 pages

Lecture Note 5B

1. The row reduced echelon form (RREF) of two row equivalent matrices is unique and independent of the row operations used to obtain it. 2. The RREF of a submatrix of a matrix A is equal to the corresponding columns of the RREF of A. 3. A matrix is invertible if and only if its RREF is the identity matrix.

Uploaded by

RAJ MEENA
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/ 2

Linear Algebra through Matrices; Lecture 05 Part B, 07 April

Department of Mathematics and Statistics Indian Institute of Technology - Kanpur

RREF and Inverse

1 Results related to RREF


We now state an important result on the RREF of two row-equivalent matrices. The proof is
omitted as it is beyond our scope.
Theorem 1.1. 1. Let A and B be two row equivalent matrices in RREF. Then A = B.
2. The RREF of a matrix A is unique. This uniqueness implies that RREF(A) is independent
of the choice of the row operations used to get the final matrix which is in RREF.

Remark 1.2. Let A ∈ Mm,n (C).


1. Let B = EA, for some elementary matrix E. Then, RREF(A) = RREF(B).
2. Let F = RREF(A) and B = [A[:, 1], . . . , A[:, s]], for some s, 1 ≤ s ≤ n. Then,

RREF(B) = [F [:, 1], . . . , F [:, s]].

Proof. There exist an invertible matrix P , such that

F = P A = [P A[:, 1], . . . , P A[:, n]] = [F [:, 1], . . . , F [:, n]].

Thus, P B = [P A[:, 1], . . . , P A[:, s]] = [F [:, 1], . . . , F [:, s]]. As F is in RREF, it’s first s
columns are also in RREF. Hence, by Theorem 1.1.2, RREF(P B) = [F [:, 1], . . . , F [:, s]].
Now, a repeated use of Remark 1.2.1 gives RREF(B) = [F [:, 1], . . . , F [:, s]].

Proposition 1.3. Let A ∈ Mn (C). Then, A is invertible if and only if RREF(A) = In , i.e., every
invertible matrix is a product of elementary matrices.

Proof. If RREF(A) = In then In = P A, for some invertible matrix P . As P is invertible, by


definition P A = In = AP . Thus, by definition, again A−1 = P .
Now, let A be invertible. Then there exists an invertible matrix B such that BA = In = AB. If
RREF(A) 6= In , then there exists an invertible matrix P such that RREF(A)= P A 6= In . Thus,
P A is a n × n square matrix in RREF with P A 6= In . Hence, P A can have at "most # n − 1" pivots
#
∗ ∗
and we deduce (P A)[n, :] = 0T . Consequently, P = P In = P (AB) = (P A)B = T B = T , a
0 0
contradiction to P being an invertible matrix. Thus, the required result follows.
2

2 GJE and Inverse of a matrix


Recall that if A ∈ Mn (C) is invertible then
h there exists a matrix
i B suchhthat AB = In = BA. So, wei
want to find a B such that AB = In = e1 e2 · · · en . So, if B = B[:, 1] B[:, 2] · · · B[:, n]
is the matrix of unknowns then using matrix multiplication, we see that we need to solve n-
system of linear equations AB[:, 1] = e1 , AB[:, 2] = e2 and so on where the columns of B are the
unknowns. Thus, we have n-augmented matrices [A|e1 ], [A|e2 ], . . . , [A|en ]. So, in place of solving
the n-augmented matrices separately, the idea of GJE is to consider the augmented matrix
h i h i
A e1 e2 · · · en = A In .
h i h i
Thus, if E is a matrix such that E A In = In B then EA = In and E = B. Hence, we
conclude (even though we haven’t shown AE = In ), B = E = A−1 . This idea is used in the next
result which helps us compute A−1 whenever it exists.

Let A ∈ Mn (C). Suppose we start with C = [A In ] and compute RREF(C). If RREF(C) = [G H]


then using Remark 2, RREF(A) = G. But, either RREF(A) = In or RREF(A) 6= In .
Hence, either G = In or G 6= In . Thus, if G = In then H = A−1 . If G 6= In then, RREF(A) 6= In
and hence A is NOT invertible.
 
0 0 1
 
Example 2.1. Use GJE to find the inverse of A =  0 1 1.

1 1 1
 
0 0 1 1 0 0
 
Solution: Starting with applying E13 to [A | I3 ] = 
 0 1 1 0 1 0
 gives

1 1 1 0 0 1
     
1 1 1 0 0 1 1 1 0 −1 0 1 1 0 0 0 −1 1
E13 (−1),E23 (−2)
E13  0 1 0 −1 1 0 E12→
(−1) 
   
→  0 1 1 0 1 0
 →  
0 1 0 −1 1 0.
 
0 0 1 1 0 0 0 0 1 1 0 0 0 0 1 1 0 0
 
0 −1 1
Thus, A−1 = 
 
−1 1 .
0
1 0 0

You might also like