Open mbercx opened 3 years ago
Pinging @sphuber for comments, since this should probably still be added to v3.3.2.
I don't understand. Didn't we run the GeTe test case with RelaxType.VOLUME
? That worked just fine. Is this a limitation of a particular version of QE? Which version were you running?
Edit: in fact, the GeTe example is run with RelaxType.POSITIONS_SHAPE
not the volume. The volume is actually kept constant in this example. It is confirmed that optimizing the volume is currently only supported for ibrav=1
i.e. simple cubic systems.
Which version were you running?
I'm running QE v6.6. I was also surprised to run into this issue, as I thought I had field tested all the different RelaxType
s (also with QE v6.6, if I remember correctly). I must have not been thorough enough in my testing after all, maybe I didn't actually run all the different possibilities, but just test the get_builder_from_protocol()
method?
I bumped into this issue now when I was trying to test all the RelaxType
s for the aiida-common-workflows
project. E.g. try running:
aiida-common-workflows launch relax -r volume -S Al -- quantum_espresso
And you'll get a PwCalculation
that fails with exit code 305, with the above error message in the stdout
. I then tested this in aiida-quantumespresso
itself using the develop
branch, and was able to reproduce the same error.
Support for the VOLUME
and POSITIONS_VOLUME
RelaxType
s was temporarily removed in #660, until we sort out how to deal with this properly.
Currently, trying to run the
PwRelaxWorkChain
with theget_builder_from_protocol
builder andRelaxType.VOLUME
orRelaxType.POSITIONS_VOLUME
doesn't work, since the defaultibrav
is set to0
, which results in the following error:If it's true that you can only run with
ibrav == 1
, I suppose we can just set this in theget_builder_from_protocol
method. However, in case a non-BBC structure is run, the work chain will except:So perhaps we should include a validator for the
PwRelaxWorkChain
instead.As a side note, I'm not sure why you can't run e.g. FCC structures with
cell_dofree == 'volume'
. Is this error message correct?