I want to calculate the eigenvalues form the very huge sparse mass and stiffness matrix.

I have mass and stiffness matrix in sparse format since the size of the matrix is too big.

I want to claculate the eigenvalues from them in the sparse format only. How to do so?

[Steven Lord]

How big is "too big"?

How many of the eigenvalues do you want to compute?

[HIMANSHU]

It is a square matrix with 3 lakh rows.

I need eigenvalue which has positive real part.

Answer 1

[Aquatris]

If eig() function is taking too long, you can use eigs() function.

Example from mathworks eigs() site:

A = delsq(numgrid('C',15));
size(A)

ans = 1x2

139 139

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

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

eigs(A)' % by default 6 largest eigenvalues

ans = 1x6

7.8666 7.7324 7.6531 7.5213 7.4480 7.3517

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

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

eigs(A,20)' % optional argument to define how many eigenvalues you want

ans = 1x20

7.8666 7.7324 7.6531 7.5213 7.4480 7.3517 7.2848 7.1478 7.1093 7.0644 6.9295 6.8257 6.7737 6.7395 6.6472 6.5914 6.4252 6.4118 6.3598 6.3196

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

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

eigs(A,5,'smallestabs')' % optional argument to get the smallest eigenvalues instead

ans = 1x5

0.1334 0.2676 0.3469 0.4787 0.5520

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

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