From c18d153f20ac9a4c5e6e4aa9023c4f7fa5c9ada6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Fabien=20P=C3=A9an?= Date: Mon, 7 Aug 2023 22:01:47 +0000 Subject: [PATCH] Update FAQ --- README.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/README.md b/README.md index 56d47e9..af40e4c 100644 --- a/README.md +++ b/README.md @@ -343,6 +343,13 @@ Note: Make sure to update `CMAKE_MODULE_PATH` env variable (otheriwse, `find_pac You need to copy this eigen vector estimate in `v` (not `resid`) and set `info` to 1 before calling aupd methods. The `v` vector targets a non-null vector such that `resid = 0`, that is, such that `A*v = lambda*v`. +- Using PARPACK, I get incorrect eigen values. + + Make sure each MPI processor handles a subpart of the eigen system (matrices) only. + ARPACK handles and solves the whole eigen problem (matrices) at once. + PARPACK doesn't: each MPI processor must handle and solve a subpart of the eigen system (matrices) only (independently from the other processors). + See examples for Fortran in folder `PARPACK/EXAMPLES/MPI`, and for C/C++ examples in `PARPACK/TESTS/MPI/icb_parpack_c.c` and `PARPACK/TESTS/MPI/icb_parpack_cpp.cpp` + ## Using MKL instead of BLAS / LAPACK How to use arpack-ng with Intel MKL: