Compute only desired several eigenvalue for a generalized eigenvalue problem

[Wu]

Hello, I have a generalized eigenvalue problem Aq = wBq. Sparse matrix A and B are shown below, neither are symmetry.

matrix A:

matrix B:

My question is how can I compute only several interested eigenvalue? I used the following code, where the sigma is the desired a eigenvalue, and I want to compute for example 15 eigenvalues around Sigma. As I tried, the code seems to compute the largest imaginary of the overall eigenvalue spectrum.

import sys, slepc4py
slepc4py.init(sys.argv)
from petsc4py import PETSc    
from slepc4py import SLEPc

opts = PETSc.Options()
kk = 15

# create petsc sparse matrix (createAIJ) and assigh lhs and rhs data
A = PETSc.Mat().createAIJ(size=LHS.shape,
                               csr=(LHS.indptr, LHS.indices,
                                    LHS.data),comm=PETSc.COMM_WORLD)
A.assemble()
B = PETSc.Mat().createAIJ(size=RHS.shape,
                               csr=(RHS.indptr, RHS.indices,
                                    RHS.data),comm=PETSc.COMM_WORLD)
B.assemble()

# create SLEPc Eigenvalue solver
E = SLEPc.EPS().create(PETSc.COMM_WORLD)
E.setOperators(A,B)
E.setDimensions(kk,PETSc.DECIDE)     # set number of  eigenvalues to compute
E.setWhichEigenpairs(E.Which.LARGEST_IMAGINARY)

E.setProblemType(SLEPc.EPS.ProblemType.GNHEP)    # generalized non-Hermitian eigenvalue problem

E.setTolerances(tol,500)   # set tolerance 
if sigma is not None:
    E.setTarget(sigma)     # set the desired eigenvalue 
E.setTarget(sigma) 
if v0 is not None:
    E.setInitialSpace(v0)  # set the initial vector

st = E.getST()
st.setType('sinvert')

E.setFromOptions()
E.solve()

[Lisandro Dalcin]

This issue is misplaced. petsc4py has no business computing eigenvalue problems.