Thanks. I will post on their GitHub
I think @matterhorn103 said this was the appropriate issue: [BUG] - QCG Terminates with Backtrace · Issue #357 · crest-lab/crest · GitHub
Hi @ripes,
As @ghutchis said, the underlying issue is a CREST bug unfortunately. I’ve been grappling with it myself while trying to add a solvation runtype to the plugin.
The thread Geoff linked to is the one I’ve posted in, and I saw your comment there too just now, but there’s also this issue and this one from May which look to be the same problem.
You can try using the bleeding-edge version of xtb and that solves some of the issues that are cropping up some of the time with QCG, but in general I would expect the QCG runtypes to be broken until the CREST devs fix whatever is going wrong behind the scenes.
The next release of the plugin will have the solvate command in it despite this, it will just display a warning if a broken CREST/xtb version combination is detected.
I’m also interested to know how you’re even running these CREST calculations? I am planning to add a “Run CREST” command to match the “Run xtb” one but I hadn’t got round to it yet…
In general you should be warned that while the “Run xtb” command can essentially start any xtb calculation (it just runs whatever you tell it to on the command line using your Avogadro geometry as input), you shouldn’t expect any of the results to be passed back to Avogadro other than the things the plugin already knows how to look for (i.e. the final geometry, frequencies, orbitals).
To be honest I wonder if it would be more user friendly to remove the “Run xtb” command and just allow extra command line options to be provided in the config menu. As in, I provided the command mainly so that if you want to use e.g. GBSA solvation with --gbsa or to change the SCC accuracy with --acc it’s possible without me adding explicit support. Running non-supported runtypes with the command is a bit pointless because the plugin doesn’t know how to parse the results appropriately.
Hi @matterhorn103
to be honest, I am running these latter crest calculations by command line, but because till this afternoon I was not able to find the xtb forum on github - eheh - I thought I could stalk you here 
I see the problem about extra runtypes, I agree that it may be useless to launch complex commands by the interface. In my (short) experience, the single thing I would benefit most of, is the ability to open conformers.xyz file combined with crest.out to display both the structures and populations seamlessly. But probably this is too much…
Thnaks again for your help, I’ll keep following the discussion on github.
What exactly do you want? The conformers file should have all of them with relative energies.
Are you saying you want the new conformers window to show populations? That was also a request from @matterhorn103 but it’ll take a bit, since I need to do some energy conversions.
It should already show relative energies – I just haven’t done a lot of testing on crest output yet.
I have nothing to do with the xtb or crest projects, just so that you’re aware!
Also reads to me like that’s what was meant!
Note @ripes that since the populations directly follow from the relative conformer energies, the plugin is actually already passing all the information needed to show this! It just requires some mathematical treatment, which as @ghutchis says might take some time to implement. But the infrastructure is there 
I’ll do a bit in the coming days.
@matterhorn103 @ghutchis thanks both.
I look forward to the implementation, I’ll be happy to beta-test anything, just ask.
I have nothing to do with the xtb or crest projects, just so that you’re aware!
Yeah, I had this (wrong) impression till yesterday, sorry for that.
Not a problem at all, just want to make sure there’s clarity
1 Like