Arpack.jl

Arpack.jl copied to clipboard

Hi, I am trying to obtain the lowest eigenvalue of a sparse hermitian matrix with one or more zero pivots. For example, my matrix has the following form:

ham

6×6 SparseMatrixCSC{Float64, Int64} with 10 stored entries:
5.0   ⋅   1.0  -1.0   ⋅     ⋅
   ⋅    ⋅    ⋅     ⋅    ⋅     ⋅
  1.0   ⋅    ⋅     ⋅    ⋅    1.0
 -1.0   ⋅    ⋅     ⋅    ⋅   -1.0
   ⋅    ⋅    ⋅     ⋅    ⋅     ⋅
   ⋅    ⋅   1.0  -1.0   ⋅    5.0

. When I execute the “eigs” function for the above matrix, I get the following error message:

energy, ψ = eigs(ham, nev=1, which=:SM)

LoadError: ZeroPivotException: factorization encountered one or more zero pivots. Consider switching to a pivoted LU factorization.
ZeroPivotException: factorization encountered one or more zero pivots. Consider switching to a pivoted LU factorization.

According to the massage, it suggest the use of a pivoted LU factorization, can I chose it? If not, is this designed with a purpose? It would be nice to have it work for matrices that contain zero pivots.

Since you have a Hermitian matrix, try eigs(ham, nev=1, which=:SR) i.e use the real part option.