How can I get inverse matrix at 50 x 50 sparse matrix?

Hello,

I have a square sparse matrix A (50 x 50), and I need to get the inverse matrix of A.

for example, at A x B = C, I know A and C so I have to know matrix B. So I have to get A's inverse matrix.

But if I use inv(A) at the code, every matrix element of inv(A) get 'inf'.

How could I get this inverse matrix? Do I have other method to get matrix B?

[Bruno Luong]

Typically this happens when you try to solve on non invertible matrix. Small example

A=[1 2; 1 2]

A = 2×2

1 2 1 2

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

inv(A)

Warning: Matrix is singular to working precision.

ans = 2×2

Inf Inf Inf Inf

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

B=rand(2);
C=A*B

C = 2×2

1.5628 0.6109 1.5628 0.6109

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

B = A\C

Warning: Matrix is singular to working precision.

B = 2×2

NaN NaN NaN NaN

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Answer 1

[Walter Roberson]

If inv(A) is all inf then chances are that A is singular. You should check rank(A) before proceeding.

You should probably not be forming inv(A) explicitly. You should probably be using

B = A\C;

The \ operator will find a solution that minimizes the sum of squares

Note that the inverse of a sparse matrix is typically a dense matrix. But that isn't going to matter if you use the \ operator.

[Esther Kang]

Thank you for your answer, but matrix B doesn't get proper value. Every elements get 'NaN'...

[Angelo Yeo]

@Esther Kang, can you give us a sample data to reproduce the issue?

[Esther Kang]

Thank you for your reply @Angelo Yeo, I solved this problem by initial condition just before!

I had to calculate power flow using newton Raphson, and I set initial votage to 0. So I couldn't get the inverse of matrix A.(actually it was Jacobian matrix in power flow) I set initial voltage to 1, and now it works!!