I have a question regarding Feast generalized eigenvalue solver. I have been using the dfeast_scsrgv for a real valued problem long time and now I wanted to use Feast for a complex problem. And I am using "cfeast_hcsrgv"
So for the following matrices:
A= [1.0 0.0;
0.0 1.0 ]
B= [2.0 0.0;
0.0 2.0i ]
I know the solutions of generalized eigenvalue problem should give 0.5 and -0.5i as eigenvalues.
I tried to obtain these using cfeast_hcsrgv and I am only getting the real valued eigenvalue. The interval that I used is from -3 to 3. And for cfeast_hcsrgv, these interval parameters should be defined as float. So I am not sure how to make them complex valued. Even if I knew it, I am not sure how the define the lower value to include the complex domain as well. My guess is [-3 3] does not include the complex domain.
Attached you could see a source file where I obtain eigenvalues of a real valued case (again very simple 2 by 2 case) and after line 132, the same approach is applied for a complex valued case. Since the original problem is a quite big sparse FEM problem, I had been using some other libraries as well and that's why it might look over complicated for such an easy problem at the first glance. But my problem is basically not being able to covering the complex domain for the guess interval of eigenvalues for the complex part (line 135 and 136)
I would really appreciate if you could help me.