Closed smujahed closed 3 years ago
I got the same problem. I could generate GGA basis, but failed for LDA basis. I used LibXC 3.3.8, and couldn't solve this problem yet.
Ayako Nakata
I have been trying to generate some LDA basis sets and have found that for all elements and basis set sizes MakeIonFiles will SEGFAULT. The error isn't obvious to me, but I have included all files necessary to replicate the issue assuming it is not an issue with my binary or linked libraries. I have only included samples of the issue with H and Si.
The pseudopotentials and inputs are the same as those found in: https://github.com/OrderN/CONQUEST-release/tree/master/pseudo-and-pao
Without the same Intel compiler, FFTW and LibXC versions I have, do you successfully generate the basis? If so, could you let me know what compilers and libraries you're using? It's possible there is something wrong with my binary or libraries, but I have had success running CONQUEST in the past with these.
Code versions and compile environment:
- BasisGeneration/Conquest version 1.0.3 (8dc94a8 https://github.com/OrderN/CONQUEST-release/commit/8dc94a8907b6a21b25f6a4340aeb990c8daa9bbf)
- Intel compiler version 2019.3.199
- FFTW version 3.3.8
- LibXC version 4.3.4
- CPU info: Intel Xeon E5-2665
bad-basis-gen.tar.gz https://github.com/OrderN/CONQUEST-release/files/4905816/bad-basis-gen.tar.gz
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/OrderN/CONQUEST-release/issues/63, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADIRNHRKN3SHTSKIQ6FPN2TR26MXZANCNFSM4OW7Z4BA.
@ayakon Thank you for confirming.
I have no problem using GCC 9.3.0, FFTW 3.3.8 and LibXC 4.3.4.
@smujahed Where does it segfault? Please run with Iprint 6 and IO.WriteOutToFile F to help us identify where the fault is. @ayakon Which compiler did you use?
@davidbowler How strange.
I have run with Iprint 6 and WriteOutToFile F, although this last flag seems to have been ignored (I can't see it in the input.log). Please find the output attached again for H and Si.
N.B. The directory in the zip file has the same name as the previous version, so take care when choosing an extraction location, sorry!
OK - we both made mistakes! It should have been:
IO.WriteOutToFile F
but this is only for Conquest. Please save output (pipe to tee).
@davidbowler Ah, I see! All output produced by MakeIonFiles is in the out-debug
files, i.e.
~/bin/MakeIonFiles &> out-debug
OK - the problem is here:
We should have exc_array
declared always. Fix coming soon.
I will test this now.
Thank you for the quick patch!
Please use the new branch:
https://github.com/OrderN/CONQUEST-release/tree/bugfix-ionfile-xc-error
I have been trying to generate some LDA basis sets and have found that for all elements and basis set sizes MakeIonFiles will SEGFAULT. The error isn't obvious to me, but I have included all files necessary to replicate the issue assuming it is not an issue with my binary or linked libraries. I have only included samples of the issue with H and Si.
The pseudopotentials and inputs are the same as those found in: https://github.com/OrderN/CONQUEST-release/tree/master/pseudo-and-pao
Without the same Intel compiler, FFTW and LibXC versions I have, do you successfully generate the basis? If so, could you let me know what compilers and libraries you're using? It's possible there is something wrong with my binary or libraries, but I have had success running CONQUEST in the past with these.
Code versions and compile environment:
bad-basis-gen.tar.gz