I am trying to run the LiF BSE tutorial on exciting-neon. The calculations seem to progress just fine until it comes to writing out the EPSILON and EXCITON files. In the EPSILON folder, I do not see the file
EPSILON_BSE-singlet-TDA-BAR_SCR-full_OC11.OUT (nor any other file in each folder, for that matter). Consequently, I cannot do any post-processing to visualize the optical spectrum. If I look at my error.log file, it gives me the error message
Error(xhdf5: write_double): size(this%offset) /= rank.
Can someone help me identify/resolve this issue? Thank you!
@abuccheri Hi Alex, do you happen to have any insights into this error? I’m consequently unable to run any BSE calculations using exciting-neon (the dielectric function isn’t being outputted), though if I downgrade it to exciting-fluorine, the calculation proceeds just fine. I would like to use the new SVLO feature and fastbse in exciting-neon since I’m trying to run BSE calculations for spin polarized systems.
Hi @joebiv98,
You should contact Benedikt Maurer, he’s the developer responsible for implementing the ISDF-BSE method in exciting, as well as the HDF5 interface. The error above looks like a bug w.r.t. MPI offsets when using multiple ranks to write.
You can also check if that BSE currently supports spin-polarisation.
Cheers,
Alex