The cost of solving a reactor core neutronics problem is reduced by accelerating the iterative process. Special techniques are used to estimate the eigenvalue of the dominant error vector to support acceleration. Eigenvalue estimates can be tested for reliability to help make acceleration effective.