Closed tbigot closed 11 months ago
Dear @tbigot,
Can you please let me know which HyPhy version you are using (hyphy --version
or HYPHYMPI --version
)?
Try adding `ENV="TOLERATE_NUMERICAL_ERRORS=1;" to your comman line call and see if that lets the job run to completion.
Would you be filling to share the input files with me (privately if needed) so that I could reproduce the error and try to resolve it more systematically?
Best, Sergei
Thanks! I’m using the last GIT commit (41306280822033d0b6d9432cee8f7ed4051ee4d5
) HYPHY 2.5.48(MPI) for Linux on x86_64
.
It is public data, you’ll find them at: https://dl.pasteur.fr/fop/8NIIUfny/nt.aln.fasta.gz
I can also provide the intermediate file generated by Gard, with the checkpoint data, it makes the software end a lot faster (because it re-uses the already computed results), if it helps! 🙂
Dear @tbigot,
Sorry for letting this drop; can you please re-share the data.
Best, Sergei
Stale issue message
Hi,
I’m having a problem running HYPHY with GARD. I have this kind of error:
Internal error in _LikelihoodFunction::ConjugateGradientDescent. The function evaluated at current parameter values [-6227879.055521984] does not match the last recorded LF maximum [-5227974.756559206]
My command:
mpirun HYPHYMPI gard --type nucleotide --rv GDD --model JTT --rate-classes 4 --mode Faster --alignment nt.aln.fasta --output nt.aln.json
I can provide data or more complete logs if needed. Thanks in advance, Thomas