OPTIMIZATION OF THE SUBSPACE ITERATION METHOD FOR THE LAMBDA-MODE DETERMINATION OF A NUCLEAR-POWER REACTOR

In this paper, we propose different alternative approaches to optimize the subspace iteration method which obtains the dominant lambda-modes of a nuclear power reactor. We present a variational acceleration for the iterative process, a coarse grain parallelizatization approach, a nd a multistep algorithm to choose efficient initial estimations for t he subspace iteration method. We have tested the efficiency of the dev eloped algorithms with a 3D realistic problem, corresponding to a stat ic configuration of the reactor Ringhals I.