Workflow for Quasi-harmonic approximation (forcefields and VASP) #903

JaGeo · 2024-06-28T15:47:59Z

Continues #902 .

ToDo:

JaGeo · 2024-06-29T06:47:57Z

I tracked it down. As soon as I initialize PhonopyQHA, I cannot jsanitanize the document. Even in cases when none of the outputs are used for the document. This is an extremely weird error and I might have to go through the phonopy source to understand where this might be coming from.

JaGeo · 2024-06-29T07:15:44Z

I think it is this line here: https://github.com/phonopy/phonopy/blob/04f912a77e3efbac00a9987b545fc4f5e8927eb9/phonopy/qha/eos.py#L134

Afterwards all warnings are turned into exceptions and none of the tests are running as before anymore.

… QHA api

tests/common/schemas/test_qha.py

JaGeo · 2024-07-10T14:26:25Z

@tpurcell90 this one would be nearly ready. If you have time, you could have a look here.

tpurcell90 · 2024-07-10T14:28:19Z

Yes do you want me to do directly in this PR or in a separate branch that gets merged in (not sure if there are concurrent changes going on)?

JaGeo · 2024-07-10T14:31:59Z

@tpurcell90 i have stopped working on it for now. Tests have been passing and I am waiting for further feedback from @utf .

But you might want to use your own branch starting from this one.

tpurcell90 · 2024-07-10T14:35:44Z

Okay I will start my own branch and then merge things in once done just in case

tpurcell90 · 2024-07-10T14:44:31Z

It might make more sense for #889 to be fully merged in before I adapt this one. Since that one seems to just need a final merge. Since this also uses eos workflows I don't want to conflict with what @ansobolev did

JaGeo · 2024-07-10T14:49:04Z

Ah, I see! @tpurcell90

JaGeo · 2024-07-10T14:50:12Z

@tpurcell90 to add, i don't rely on the full eos workflow as it would be too hard to modify. Therefore, I would not expect many conflicts.

src/atomate2/vasp/flows/qha.py

JaGeo · 2024-07-23T09:40:46Z

src/atomate2/forcefields/flows/qha.py

+        -------
+        .PhononMaker
+        """
+        logging.log(level=0, msg="test")


JaGeo · 2024-08-01T18:47:34Z

src/atomate2/common/flows/qha.py

+        self.eos = CommonEosMaker(
+            initial_relax_maker=self.initial_relax_maker,
+            eos_relax_maker=self.eos_relax_maker,
+            static_maker=None,
+            postprocessor=None,
+            number_of_frames=self.number_of_frames,
+        )


@utf I am currently wondering about how to initalize and reuse the EosMaker in the best way. Unfortunately, starting solely from the EosMaker for the qha workflow would take too much time. This is why I only partly reuse and extend it here. Should I do it similarly to the Grüneisen workflow and add the EosMaker as an argument for the overall qhamaker?

@utf ? any opinions?

I'm a little confused - what do you mean by:

starting solely from the EosMaker for the qha workflow would take too much time

Either way, I'm open to both approaches:

Adding the EosMaker as an argument would reduce the number of options in the CommonQhaMaker class but make it more difficult to set properties such as the number of frames. I don't think this is too much of an issue if we could include an example of how to configure the settings in the docstring/documentation page.

The current internal use of CommonEosMaker makes it easier to configure settings but is slightly less clean code. By the way, I noticed the linear_strain option is not currently used.

I'll let you decide!

Thanks! I think I will go with 2.

What I meant: instead of relying on phonopy to deal with the equation of state fitting, to reuse the fitting options from the EosStateMaker. (EDIT: there was another comment here. I mixed something up.)

utf

Thanks for this @JaGeo, it looks in a very nice state. Just some minor points from me.

utf · 2024-08-23T08:37:46Z

src/atomate2/common/flows/qha.py

+        self.eos = CommonEosMaker(
+            initial_relax_maker=self.initial_relax_maker,
+            eos_relax_maker=self.eos_relax_maker,
+            static_maker=None,
+            postprocessor=None,
+            number_of_frames=self.number_of_frames,
+        )


I'm a little confused - what do you mean by:

starting solely from the EosMaker for the qha workflow would take too much time

Either way, I'm open to both approaches:

Adding the EosMaker as an argument would reduce the number of options in the CommonQhaMaker class but make it more difficult to set properties such as the number of frames. I don't think this is too much of an issue if we could include an example of how to configure the settings in the docstring/documentation page.

The current internal use of CommonEosMaker makes it easier to configure settings but is slightly less clean code. By the way, I noticed the linear_strain option is not currently used.

I'll let you decide!

utf · 2024-08-23T08:37:54Z

src/atomate2/vasp/flows/qha.py

+    def prev_calc_dir_argname(self) -> str:
+        """Name of argument informing static maker of previous calculation directory.
+
+        As this differs between different DFT codes (e.g., VASP, CP2K), it
+        has been left as a property to be implemented by the inheriting class.
+
+        Note: this is only applicable if a relax_maker is specified; i.e., two
+        calculations are performed for each ordering (relax -> static)
+        """
+        return "prev_dir"


Is this actually needed any more? VASP, CP2K etc all have the same dir name now prev_dir. I guess we coud remove this from any other workflows that have it?

I think yes because abinit has different names.

See

atomate2/src/atomate2/abinit/jobs/base.py

Line 140 in 1ebc9a7

def make(

utf · 2024-08-23T08:39:09Z

src/atomate2/common/jobs/qha.py

+    phonon_jobs = []
+    outputs = []
+    for istructure, structure in enumerate(eos_output["relax"]["structure"]):
+        phonon_job = phonon_maker.make(structure)


If possible, it would be nice to pass the prev_dir from the relaxation? That way we can make use of auto_ispin and make the calculations cheaper.

Did this now by extending the EOS dictionaries.

JaGeo · 2024-09-03T05:03:37Z

@utf I would be really happy if this could be merged. I would like to show this in an upcoming presentation (~2 weeks).

JaGeo added 9 commits June 28, 2024 17:46

add structures to eos output

195a7cc

fix defaults in phonon schema

aa291da

add qha flows and one full workflow test

faf7633

really add qha flows

a84161b

fix description of qha schema file

dc74e72

remove tests

adab362

add schema test instead

c6b01a2

new test files

2f641da

remove future import

3c059e3

JaGeo added 12 commits June 29, 2024 10:07

Fix tests by changing the warning treatment back to default after the…

34640d5

… QHA api

fix linting

6480df8

add switch to eos types

580ccab

change how warnings are treated

fdfb338

add safer code for literal

f26fe13

add more tests

1b86ed9

fix tests add kwargs

71ef7b8

Fix tests

d7900ae

Fix tests

e21a679

reformat

3fe1597

linting

7145376

update prev calc

a991a6c

JaGeo commented Jul 1, 2024

View reviewed changes

tests/common/schemas/test_qha.py Outdated Show resolved Hide resolved

JaGeo and others added 6 commits July 2, 2024 00:47

Apply suggestions from code review

80c4a79

fix test tolerance

7c07011

fix test tolerance

0f08b5a

Merge branch 'main' into qha_fresh

89db46f

draft a first vasp workflow

c948542

fix return

a85ff08

Merge branch 'main' into qha_fresh

b1ad4f7

JaGeo commented Jul 23, 2024

View reviewed changes

src/atomate2/vasp/flows/qha.py Outdated Show resolved Hide resolved

JaGeo commented Jul 23, 2024

View reviewed changes

src/atomate2/forcefields/flows/qha.py Outdated

-------

.PhononMaker

"""

logging.log(level=0, msg="test")

Copy link

Member Author

JaGeo Jul 23, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remove

change to phonon static

f2974bd

JaGeo commented Aug 1, 2024

View reviewed changes

JaGeo and others added 3 commits August 23, 2024 09:39

change some initialisations

e0b4536

Merge branch 'main' into qha_fresh

ca24d1c

change phonon maker

de97216

utf reviewed Aug 23, 2024

View reviewed changes

JaGeo added 9 commits August 23, 2024 11:47

fix tests

ae34913

fix test

502b7dc

update qha_fresh

b8af794

fix comments

8b0417d

fix comments

b58857c

connect flows

eb35145

fix linting

9dd7056

correct documentation

f70f7cc

document workflows, fix imports

0bd4a37

Merge branch 'main' into qha_fresh

9596811

JaGeo enabled auto-merge (squash) September 9, 2024 20:18

JaGeo merged commit f1894c1 into materialsproject:main Sep 9, 2024
6 checks passed

utf added the feature A new feature being added label Sep 26, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Workflow for Quasi-harmonic approximation (forcefields and VASP) #903

Workflow for Quasi-harmonic approximation (forcefields and VASP) #903

JaGeo commented Jun 28, 2024 •

edited

Loading

JaGeo commented Jun 29, 2024

JaGeo commented Jun 29, 2024 •

edited

Loading

JaGeo commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

JaGeo commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

JaGeo commented Jul 10, 2024

JaGeo commented Jul 10, 2024

JaGeo Jul 23, 2024

JaGeo Aug 1, 2024 •

edited

Loading

JaGeo Aug 23, 2024

utf Aug 23, 2024

JaGeo Aug 23, 2024 •

edited

Loading

utf left a comment

utf Aug 23, 2024

utf Aug 23, 2024

JaGeo Aug 23, 2024

JaGeo Aug 23, 2024

utf Aug 23, 2024

JaGeo Aug 23, 2024

JaGeo commented Sep 3, 2024

Workflow for Quasi-harmonic approximation (forcefields and VASP) #903

Workflow for Quasi-harmonic approximation (forcefields and VASP) #903

Conversation

JaGeo commented Jun 28, 2024 • edited Loading

JaGeo commented Jun 29, 2024

JaGeo commented Jun 29, 2024 • edited Loading

JaGeo commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

JaGeo commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

tpurcell90 commented Jul 10, 2024

JaGeo commented Jul 10, 2024

JaGeo commented Jul 10, 2024

Choose a reason for hiding this comment

JaGeo Aug 1, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JaGeo Aug 23, 2024 • edited Loading

Choose a reason for hiding this comment

utf left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JaGeo commented Sep 3, 2024

JaGeo commented Jun 28, 2024 •

edited

Loading

JaGeo commented Jun 29, 2024 •

edited

Loading

JaGeo Aug 1, 2024 •

edited

Loading

JaGeo Aug 23, 2024 •

edited

Loading