Cholesky vs Eigs speed comparison?

70 views (last 30 days)
Lennart Sinjorgo
Lennart Sinjorgo on 25 Oct 2024 at 15:05
Edited: Bruno Luong on 29 Oct 2024 at 9:50
I have many large (8000 x 8000) sparse PSD matrices for which I would like to verify that their largest eigenvalue is at most some constant C. If A denotes such a matrix, there are two ways to check this.
  1. eigs(A,1) <= C
  2. chol(C*speye(size(A,1)) - A) (and inspect the output flag)
Somehow, method 1 is much faster than 2 (this difference is not due to the computation time of C*speye(size(A,1)) - A) . Why is that? The general consensus is that cholesky decomposition is the fastest way of determining PSD-ness of matrices. Is chol not as fast for sparse matrices as eigs?

Answers (1)

Bruno Luong
Bruno Luong on 25 Oct 2024 at 16:03
Edited: Bruno Luong on 28 Oct 2024 at 14:07
Your test of PSD is not right. You should do:
eigs(A,1, 'smallestreal') > smallpositivenumber % 'sr', 'sa' for older MATLAB
EIGS compute one eigen value by iterative method. It converges rapidly for
eigs(A,1)
  6 Comments
Bruno Luong
Bruno Luong on 28 Oct 2024 at 19:12
Edited: Bruno Luong on 28 Oct 2024 at 22:38
An example with sparse matrix based on Haar basis. The ratio of two matrix norms increases like sqrt(N)
N = 2^14
N = 16384
A = haar(N);
% eigs(A,1) must be 1
norm(A,inf) / eigs(A,1)
ans = 128.0000
function S=haar(N)
if (N<2 || (log2(N)-floor(log2(N)))~=0)
error('The input argument should be of form 2^k');
end
p=[0 0];
q=[0 1];
n=nextpow2(N);
for i=1:n-1
p=[p i*ones(1,2^i)];
t=1:(2^i);
q=[q t];
end
j = 1:N;
I = 1+0*j;
J = j;
V = 1+0*j;
for i=2:N
P=p(1,i); Q=q(1,i);
j = 1+N*(Q-1)/(2^P):N*(Q-0.5)/(2^P);
I = [I, i+0*j];
J = [J, j];
V = [V, 2^(P/2)+0*j];
j = 1+(N*((Q-0.5)/(2^P))):N*(Q/(2^P));
I = [I, i+0*j];
J = [J, j];
V = [V, -(2^(P/2))+0*j];
end
S = sparse(I,J,V,N,N);
S=S/sqrt(N);
end
Bruno Luong
Bruno Luong on 29 Oct 2024 at 9:31
Edited: Bruno Luong on 29 Oct 2024 at 9:50
Here is an example where the ratio is huge (= N), discovered by codinng mistake .;)
N = 2^16
N = 65536
j = 1:N;
I = 1+0*j;
J = j;
V = 1+0*j;
A = sparse(I,J,V,N,N);
norm(A,inf) / eigs(A,1)
ans = 6.5536e+04

Sign in to comment.

Categories

Find more on Linear Algebra in Help Center and File Exchange

Products


Release

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!