INvestigation Of The Biconjugate Gradient Algorithm For The Inversion Of Fermion Matrices

An algorithm for the numerical inversion of large matrices, the biconjugate gradient algorithm (BGA), is investigated in view of its use for Monte Carlo simulations of fermionic field theories. It is compared with the usual conjugate gradient algorithm (CGA) and the minimal residue algorithm (MRA) within a Higgs-Yukawa model including mirror fermions. For this model it can be shown that BGA represents an improvement under certain circumstances where the other two slow down.