28 Jun 2024 |
Hasith Vattikuti | In reply to @nw18rica:matrix.tu-darmstadt.de Is there a way to get auto completion working? (Using pycharm) As long as you have ASE (or any module for that matter, even a local one) autocomplete should be working unless you specifically turned off some setting to do so. Double check that you installed ase in your python environment as well as your pycharm settings | 16:50:07 |
Hasith Vattikuti | In reply to @nw18rica:matrix.tu-darmstadt.de Is there a way to get auto completion working? (Using pycharm) * As long as you have ASE (or any module for that matter, even a local one) autocomplete should be working unless you specifically turned off some setting to do so. Double check that you installed ase in your python environment as well as your pycharm settings. If you did your due diligence, DM me as I highly doubt this is relevant to ASE specifically. | 16:53:09 |
Nils Phillip Winkelmann (TUDa) | I installed it via a conda in intellij via the GUI of intellij | 16:53:44 |
3 Jul 2024 |
mjwaters | I just submitted a 1-character bug fix which required 3 characters to be changed in a test. This might be my smallest MR that isn't spelling related typo. | 04:36:51 |
Ask Hjorth Larsen | Adam Jackson: Hi. How's the plugins MR? Is there something I can do? | 13:07:50 |
Adam Jackson | I managed to look at it a few times last week, am still trying to fully understand it. Would be easier with more type annotation and docstrings to describe arguments, but I appreciate the code is officially still WIP. I think it didn't help that the modules are alphabetically almost in reverse order of what makes sense for understanding the structure 😅 | 13:12:20 |
Ask Hjorth Larsen | We can also have a videochat at some point if we want to discuss more in depth | 13:12:50 |
Adam Jackson | * I managed to look at it a few times last week, am still trying to fully understand it. Would be easier with more type annotation and examples in docstrings, but I appreciate the code is officially still WIP. I think it didn't help that the modules are alphabetically almost in reverse order of what makes sense for understanding the structure 😅 | 13:13:00 |
Adam Jackson | Sounds good, I think I need one more session to look through the actual "plugins" module and then hopefully it will come together a bit more clearly | 13:13:57 |
Adam Jackson | There are also perhaps some fundamental questions about e.g. how we handle conflicts and multiple implementations of the same format/calculator | 13:14:34 |
Ask Hjorth Larsen | I can talk at almost any time. We could scroll through and try to understand at a higher level. I mean I did something like that at some point, but it's a while ago and probably wise to review well :) | 13:15:06 |
Ask Hjorth Larsen | Those are good questions indeed. | 13:16:26 |
4 Jul 2024 |
Ask Hjorth Larsen | Bar is open over in #ase-bar:matrix.org | 18:58:08 |
5 Jul 2024 |
Adam Jackson | Could have a chat about plugins this afternoon? | 10:11:17 |
Ask Hjorth Larsen | Yes. Maybe 15:00? | 10:15:21 |
Ask Hjorth Larsen | CEST, so that must be 14:00 UK | 10:15:43 |
Ask Hjorth Larsen | Adam Jackson: Should we use the bar? | 12:57:50 |
Adam Jackson | Sure! Am on a new laptop so will take a minute to sign in | 12:58:27 |
8 Jul 2024 |
mjwaters | Andrew Rosen: There's another bug in the vasp calculator when setting case sensitive tags for vtst. | 06:11:56 |
mjwaters | when I want to set 'DRotMax' by setting drotmax=2, it writes the incorrect string "DROTMAX" to the INCAR file. | 06:13:47 |
mjwaters | I'm not sure how this is supposed to be handled in the vasp calculator. | 06:14:45 |
mjwaters | I did notice that drotmax is in the float section but it's an integer tag. | 06:15:14 |
mjwaters | I could fix that, but the tag still won't work without fixing the first thing. | 06:16:04 |
mjwaters | anyway, it's late, somebody remind me this week to make a issue on gitlab for this. | 06:17:14 |
Ask Hjorth Larsen | Maybe it would be wise to start a simple vasp calculator, one which has minimal awareness of keywords.
A good starting point would be a simplified vasp writer which speaks Vasp language but doesn't know about ASE. Then there can be an explicit conversion layer between ASE (atoms, kpts) and Vasp keywords which is outside the writer. And finally that can be integrated with a calculator -- or with other things.
| 11:57:14 |
Adam Jackson | What layer of abstraction do you have in mind for "vasp language"? Presumably it is higher than {"POSCAR": list[str], "INCAR": list[str], "KPOINTS": list[str], "POTCAR": list[str]} | 14:05:33 |
Adam Jackson | I guess the answer for any code here is "a reasonable mapping from Python data types to the input syntax", but then we are 90% of the way towards an ASE-specific syntax | 14:07:25 |
Adam Jackson | Maybe that is still useful as most of the hard parts lie in that other 10% 🤔 | 14:08:01 |
mjwaters | In reply to @askhl:matrix.org
Maybe it would be wise to start a simple vasp calculator, one which has minimal awareness of keywords.
A good starting point would be a simplified vasp writer which speaks Vasp language but doesn't know about ASE. Then there can be an explicit conversion layer between ASE (atoms, kpts) and Vasp keywords which is outside the writer. And finally that can be integrated with a calculator -- or with other things.
the ugly and compatibility breaking solution would be stop using the lowercase string versions of tags. | 15:46:55 |
mjwaters | wait, don't we have mechanism for passing custom tags in the vasp calculator? | 15:47:28 |