As the title says I have a large sparse matrix, 262000 by 262000, and i want to know how the number of times that an eigenvalue is degenerate. I can get the eigenvalue using the arnoldi method but after that i am clueless in what should i do. I can't calculate all of the eigenvalues of the matrix since i run out of memory.

  • 1
    $\begingroup$ Use Tally on the resulting list of eigenvalues? Since they are numerical, you may want to Round them to an appropriate precision first, to ease the comparison. $\endgroup$
    – MarcoB
    Commented May 23, 2023 at 17:00
  • $\begingroup$ The problem is: i can't calculate all of them without running out of memory. $\endgroup$
    – LittleBlue
    Commented May 23, 2023 at 17:03
  • $\begingroup$ That's an important part of your problem. Please edit your question to add that. $\endgroup$
    – MarcoB
    Commented May 23, 2023 at 17:08
  • 1
    $\begingroup$ Oh ok, I thought that was obvious since i said that was a large matrix. $\endgroup$
    – LittleBlue
    Commented May 23, 2023 at 17:22
  • $\begingroup$ provide minimal real example. just reduce your matrix and insert ur question. and what u tried with Large matrix that lead you to memory problem! $\endgroup$
    – nufaie
    Commented May 23, 2023 at 17:29

1 Answer 1


If your memory is big enough to compute the orthogonal projector to the complement of each eigenvector and look for another eigenvalue of the same value in the projected matrix

   A . e = \lambda e
   P = TensorProduct[e, Transpose@Conjugate@e]
   (1-P) e = 0
   Solve[ (1-P).A.(1-P) . x   == lambda x,x]
  • $\begingroup$ I very much doubt that symbolic methods will be of any use for a matrix this size. $\endgroup$ Commented May 24, 2023 at 11:44
  • $\begingroup$ This is a minutes test over some numbers. Depends on how sparse the matrix is. I tested it for a digaonal matrix and its unitary transformed with randoms of dimension 6. With these given, dimensions it looks like a social media on pairs problem of mid town size. $\endgroup$
    – Roland F
    Commented May 24, 2023 at 12:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.