Open moabe84 opened 8 months ago
Hi Mostafa, after CREST version 2.12, I updated the source code so that the GFN2-xTB optimization was made optionally if a different method should be used and is then only be performed when using the "--fin_opt_gfn2" flag. This feature is currently only available in the Continous release version or an other version compiled with the current source code.
Thank you for your reply. If I want to do everything at the GFN-FF level, do I also need to specify the "--enslvl gfnff" keyword: _crest mol1.xyz --qcg water.xyz --nsolv 100 --gfnff --enslvl gfnff --T 60 --gsolv --nclus 5 --mreset 4 --alpb water --nofix
I just ran the calculations with the Continous release version (precompiled binaries) using the following command line and it crashed in the "Multilevel Ensemble Optimization" step with this error:
crest pentanol.xyz --qcg benzene.coord --nsolv 25 --gfnff --T 120 --gsolv --nclus 4 --fscal 0.65 --gbsa benzene
_input file name : crest_rotamers_0.xyz
output file name : crest_rotamers_1.xyz
reference state Etot : -62.1430257400000
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
crest 0000000002659710 Unknown Unknown Unknown
crest 00000000006E1484 cregenconffile 2510 cregen.f90
crest 00000000006C01EC newcregen_ 260 cregen.f90
crest 00000000005AE3A4 confgchk3 913 confscript2_misc.f90
crest 00000000005AE1C8 sort_andcheck 1052 confscript2_misc.f90
crest 0000000000517B0D crestmultilevel 396 search_conformers.f90
crest 0000000000512C61 crest_search_imtd 129 search_conformers.f90
crest 00000000004214E5 qcgensemble 1029 solvtool.f90
crest 0000000000406350 crestsolvtool 102 solvtool.f90
crest 00000000004041F8 MAIN__ 262 crestmain.f90
crest 000000000040336D Unknown Unknown Unknown
crest 000000000265B9E0 Unknown Unknown Unknown
crest 000000000040324E Unknown Unknown Unknown
To apply GFN-FF in every step, both the --gfnff and --enslvl gfnff options need to be used. Thank you for reporting the bug. There have been some bug fixes recently that should also address this issue.
Thank you for the updates.
Hi everyone, I'm conducting Solvation Free Energy calculations using QCG for several macromolecules in 100 solvent molecules. I'm curious if it's feasible to employ the GFN-FF method for the Final Geometry Optimization. The default method is GFN2-xTB, which appears to be quite costly for my systems. Here's my command line for the calculations: _crest mol1.xyz --qcg water.xyz --nsolv 100 --gfnff --T 60 --gsolv --nclus 5 --mreset 4 --alpb water --nofix
I'm using CREST version 2.12.
Thanks, Mostafa