From: Rick F. <ri...@ar...> - 2020-02-07 23:03:42
|
After looking at the original paper on which the chow_patel_icc preconditioner is based, it seems that convergence for the preconditioner is guaranteed theoretically if a fixed point exists for the matrix. However, overflow can occur in practice which will lead to lack of convergence. In lieu of more detailed investigation, it seems likely that my matrix is inducing overflow during the preconditioner calculation which leads to lack NaNs during the CG solve. If any experts could comment on this possibility, please do so. Have others encountered this issue? |