How to perturb datas for quadratic optimization wiht CVXPY

Question

I have the same problem described here CVXPY Quadratic Programming; ArpackNoConvergence error .

I would like to try the pertubation/trasformation solution, but i have no idea how to perform it.

I have a quadratic problem like this https://www.cvxpy.org/examples/basic/quadratic_program.html , and all the matrix P, q, G, h, A, b.

The solver works as long as the matrix are small (like 150 items). But with larger problems i get the problem above.

Would really appreciate if someone could explain me how to perturb datas (with Python).

Thank you.

Answer 1

What do you mean by not working? Does it say the Q is not positive semi-definite?

If that is the case you may want to read

https://docs.mosek.com/whitepapers/qmodel.pdf