-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Refactor config parameters into UEP opt and read tools #128 #130
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hi there,
Following our meeting of 14/03/2023, here is a summary of my comments.
- I think we can remove the keyword "supercell" from the generation parameters
- We need to fix the den_fmt data type, as Galaxy does not recognise this data type. The problem seems to be that, when we developed the data type, we used a non-polarised calculation and the header of the file: 'END header: data is " charge spin" in units of electrons/grid_point * number of grid_points' lacks the word 'spin' in it.
- We should be able to have H:mu as the default symbol for the muon
- in the PyMuonSuite AIRSS Cluster tool we need to modify the options for the 'Clustering Save Type' parameter. The option 'input' is unnecessary.
- In the same PyMuonSuite AIRSS Cluster tool we need to add an extra option. If we choose the option 'Clustering Save Type' =structure, it would be useful to provide the capacity to create a .cell supercell file.
Have created galaxyproject/galaxy#15796 and galaxyproject/galaxy#15797 to address this. |
In my previous comment about the PyMuonSuite AIRSS Cluster tool, I mentioned that we need to modify the options for the 'Clustering Save Type' parameter, so that the option 'input' is removed. I also mentioned that, in the same PyMuonSuite AIRSS Cluster tool we need to add an extra option. If we choose the option 'Clustering Save Type' =structure, it would be useful to provide the capacity to create a .cell supercell file. Furthermore, that supercell file should look like this: ####################################################### %BLOCK LATTICE_CART %BLOCK POSITIONS_ABS %BLOCK SPECIES_GAMMA %BLOCK SPECIES_MASS FIX_ALL_CELL: True i.e.: it must have the %BLOCK SPECIES_GAMMA, %BLOCK SPECIES_MASS and FIX_ALL_CELL options included in it. Finally: the spin in this lines is wrong. It should be 1 and -1 |
This is already mentioned in passing on muon-spectroscopy-computational-project/pymuon-suite#61:
It's an upstream issue, PyMuonSuite is using
And it is NOT using it:
Any changes will need to happen in PyMuonSuite (I'd recommend any further discussion on the existing issue linked above). |
For The other methods will then be clustering supercells anyway. To output supercells for UEP we would need to extend PyMuonSuite to use the supercell parameter at this stage of the process, iff the calculator is UEP (otherwise we would expand the cell twice). Once again this is an upstream issue, tracked by muon-spectroscopy-computational-project/pymuon-suite#78 |
With changes to PyMuonSuite (muon-spectroscopy-computational-project/pymuon-suite/pull/79 muon-spectroscopy-computational-project/pymuon-suite/pull/81), and some minor changes in the xml to account for them, it's possible to generate the following for the
I believe this addresses all the above concerns, but if not it's likely any further changes will need to be in PyMuonSuite rather than here. It's also worth saying that until we release the changes to PyMuonSuite in a new version this PR will be blocked. |
…pectroscopy-computational-project/muon-galaxy-tools into patrick/128_stopping_site_changes
Have updated the tools to use the new 0.3.0 version of PyMuonSuite, which should enable the tools to produce clustered cell files with supercells and supplementary muon information, e.g. https://github.com/muon-spectroscopy-computational-project/muon-galaxy-tools/blob/patrick/128_stopping_site_changes/pm_muairss_read/test-data/Si_uep_min_cluster_1.cell Following discussion with @leandro-liborio, have also altered how we run the UEP optimisation so that all sites are calculated in parallel rather than sequentially. On my development instance this resulted in a speedup from around 3 seconds per site to 1 second per site for a simple Si example. The exact impact will depend on how we configure our job submission in production (i.e. how many cores we allocate to UEP jobs). Attempted to bump the MSS version as well, but ran into muon-spectroscopy-computational-project/muspinsim/issues/86 so left it at 2.2.1 until that can be fixed. |
NB: This change is larger in scope than #126, but builds on the decision there to add the clustering_save parameters to the read tool rather than the config tool. If this is merged, that PR can be closed without merging. If this PR is rejected/changed significantly, #126 should still be considered as an active PR.
Closes #99, closes #128
Context
Motivation was to make the stopping site calculations using UEP (the most likely calculator for users to want to use) more streamlined, by removing options which are not relevant to UEP calculations from the UI.
Our tools wrappers for the stopping site workflow were split into 4 stages/tools:
Impact of change
As of this change, the core UEP muAIRSS workflow only requires:
Having the parameters only feature at the point where they are relevant (rather than in a dedicated "config" tool) is more similar to the other PyMuonSuite tools (pm_nq, pm_asephonons), and has other benefits (and drawbacks) as described in #99, #126. Hopefully this change also makes the UI more intuitive for a user, however this is speculation at this point.
It's also worth noting that these (or other significant changes to the stopping site tools) will mean changes are required to the WIP training materials.
Alternative approaches
Other possible approaches/changes I considered that might be preferable instead of / in addition to this change:
Comments and opinions on this change are welcome, as it was not fully discussed in the project meeting.
Have deployed these updated versions under the Tools Under Development section on the dev instance.