[WIP] Multi step lattice dynamics workflow using hiPhive #559

hrushikesh-s · 2023-10-10T00:14:38Z

Summary

Include a summary of major changes in bullet points:

Workflow for calculating harmonic & anharmonic properties of phonons using hiPhive.

Additional dependencies introduced (if any)

hiPhive -- note that the renormalization & fixed_rattling features are currently not available in the [master branch of hiphive] . These features are currently in @jsyony37's fork under the branch named personal. The original hiPhive code as well as @jsyony37's fork are hosted on gitlab
ShengBTE

TODO (if any)

migrate the workflow to pydanticV2
write unit tests

Checklist

Before a pull request can be merged, the following items must be checked:

Code is in the standard Python style.
The easiest way to handle this is to run the following in the correct sequence on
your local machine. Start with running black on your new code. This will
automatically reformat your code to PEP8 conventions and removes most issues. Then run
ruff.
Docstrings have been added in the Numpy docstring format.
Run ruff on your code.
Type annotations are highly encouraged. Run mypy to
type check your code.
Tests have been added for any new functionality or bug fixes.
All linting and tests pass.

Note that the CI system will run all the above checks. But it will be much more
efficient if you already fix most errors prior to submitting the PR. It is highly
recommended that you use the pre-commit hook provided in the repository. Simply run
pre-commit install and a check will be run prior to allowing commits.

…to hiphive

codecov · 2023-10-10T00:27:45Z

Codecov Report

Attention: Patch coverage is 0.34682% with 862 lines in your changes missing coverage. Please review.

Project coverage is 66.81%. Comparing base (29a5731) to head (1421e04).
Report is 36 commits behind head on main.

❗ Current head 1421e04 differs from pull request most recent head bbebbd4

Please upload reports for the commit bbebbd4 to get more accurate results.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #559      +/-   ##
==========================================
- Coverage   74.94%   66.81%   -8.14%     
==========================================
  Files         136       79      -57     
  Lines       10513     7539    -2974     
  Branches     1643     1101     -542     
==========================================
- Hits         7879     5037    -2842     
- Misses       2143     2197      +54     
+ Partials      491      305     -186

Files	Coverage Δ
src/atomate2/vasp/files.py	`70.27% <16.66%> (-15.70%)`	⬇️
src/atomate2/vasp/flows/hiphive.py	`0.00% <0.00%> (ø)`
src/atomate2/vasp/jobs/hiphive.py	`0.00% <0.00%> (ø)`

... and 119 files with indirect coverage changes

JaGeo

I have seen this by chance. As I have worked on the phonon workflow, I wanted to leave some comments. And also ask about how easily it could be extended towards other codes than VASP and/or potentially "forcefields".

Thanks!

JaGeo · 2023-10-10T11:52:56Z

src/atomate2/vasp/flows/hiphive.py

+
+        # 2. Generates supercell, performs fixed displ rattling and
+        # runs static calculations
+        vasp_static_calcs = run_static_calculations(


Might be interesting to converge this implementation and the one in the phonon workflow with Phonopy at some point.

In this workflow, we are perturbing all the atoms of the supercell with a Gaussian distribution with a user defined mean and a fixed standard deviation of 1. The implementation can be found here.

Might be interesting to converge this implementation and the one in the phonon workflow with Phonopy at some point.

As far as I know, Phonopy's generate_displacement function doesn't perturb all the atoms in the supercell, and hence we had to implement our generate_displaced_structures function.
Maybe, I'm missing out on something here?

I was referring to this part but maybe not worth spending the time unifying it

atomate2/src/atomate2/common/jobs/phonons.py

Line 251 in 4c80764

def run_phonon_displacements(

Ohh I see. Yes, unifying hiphive workflow with run_phonon_displacements does make sense!
Thanks, I'll try to unify my workflow with this!

atomate2/src/atomate2/common/jobs/phonons.py

Line 251 in 4c80764

def run_phonon_displacements(

The only drawback that I see: you probably need to put the structures in a datastore. Thus, having it in one job might be better. In any case, we could at least think about unifying it

@hrushikesh-s I think our most recent discussion is using a fixed set of displacement rather than MCRattle, right? Check PerturbStructureTransformation in pymatgen (https://github.com/materialsproject/pymatgen/blob/1a277687f31d63d1011d59a8a529f194486265ce/pymatgen/transformations/standard_transformations.py#L743), we have deprecated the tag of disp_cut after previous discussion, right?

JaGeo · 2023-10-10T12:01:19Z

src/atomate2/vasp/flows/hiphive.py

+        default is "srun -n 32 ./ShengBTE 2>BTE.err >BTE.out".
+    PHONO3PY_CMD (str):
+        Command for executing Phono3py, default is
+        "phono3py --mesh 19 19 19 --fc3 --fc2 --br --dim 5 5 5".


Is it possible to use the API of phono3py here instead? I feel one cannot easily switch to another code otherwise.

I'm not using phono3py for calculating lattice thermal conductivity. Thanks for catching up this error, I'll remove the PHONO3PY_CMD argument.

Yeah, using phono3py api instead, we need to provide thermal conductivity calculation option using phono3py because there might be some bugs for the FORCE_CONSTANTS_3RD generation step in HiPhive and it is not solved in their side yet.

JaGeo · 2023-10-10T12:03:43Z

src/atomate2/vasp/flows/hiphive.py

+                )
+
+        # ##### 8. Lattice thermal conductivity calculation using phono3py
+        # if calculate_lattice_thermal_conductivity:


Ah, okay, it's not included. Then, maybe, the command should be removed. Otherwise, I would recommend using the API of phono3py.

At the moment the unit tests are very buggy! Pls, forgive it for now. To be fixed in the iterations to come.

…to hiphive

JaGeo · 2023-10-27T06:37:39Z

src/atomate2/common/jobs/phonons.py

@@ -246,7 +246,8 @@ def generate_frequencies_eigenvectors(
    )


-@job(data=["forces", "displaced_structures"])
+# @job(data=["forces", "displaced_structures"])


Just a reminder: this is really important for large structures. Just saw this change by chance.

Bumps [pydantic-settings](https://github.com/pydantic/pydantic-settings) from 2.3.1 to 2.3.3. - [Release notes](https://github.com/pydantic/pydantic-settings/releases) - [Commits](pydantic/pydantic-settings@v2.3.1...v2.3.3) --- updated-dependencies: - dependency-name: pydantic-settings dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [python-ulid](https://github.com/mdomke/python-ulid) from 2.6.0 to 2.7.0. - [Release notes](https://github.com/mdomke/python-ulid/releases) - [Changelog](https://github.com/mdomke/python-ulid/blob/main/CHANGELOG.rst) - [Commits](mdomke/python-ulid@2.6.0...2.7.0) --- updated-dependencies: - dependency-name: python-ulid dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [phonopy](https://phonopy.github.io/phonopy/) from 2.24.2 to 2.24.3. --- updated-dependencies: - dependency-name: phonopy dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

temporarily commenting out quality control job

1. using replace instead of add in hiphive_static_calcs job 2. ensuring that the given structure. in primitive, if not then converting to prim structure. 3. updated logic to write the FC_3rd order file 4. ruff fixes 5. added support of amabte, shengbte, and phono3py in LTC calculation 6. updated logic to only use lambda=0 for TI, when imag_modes_bool = False

1. ruff fixes 2. updated class attributes 3. added bulk modulus to the metadata 4. #TODO: Implement Quality Control Job 5. updated fireworks config for each job 6. updated definition to extract phonon dispersion 7. updated definition to extract LTC, depending upon whether or not renormalization was performed. 8. updated name of the flow

using tags in accordance with atomate LD workflow

1. supports shengBTE, almaBTE & phono3py as of now. 2. before running any of these calculators, make sure to successfully compile them as explained on their respective websites. 3.Also, remember to put the execution command for each of these calculators in atomate2.yaml file

the decision is now made based on whether n_imaginary > 0 from the fitting_data.json @ 0K

1. largest displacement is based on the avg row (periodic table) number of the compound. 2. displacements are skewed towards the large displacement 3. # of displaced structures to generate depends on the n_structures, which depends upon the symmetry of the primitive structure 4. smallest displacement is fixed at 0.01 Å

1. added a job that creates multiples sub jobs for each hiphive fitting and then collects the fitting data & the fcs. fcp, cluster space and other properties corresponding to the best cutoff! 2. setting the min # of displ to 4 & max # of displ to 60

Trying different strategies for perturbing the atoms in the supercell -- 1. Fixed displacement 2. MC Rattle 3. MD informed displacements Adding a penalty to the displacements that are generated out of the distribution.

1. selects cutoffs based on target DOFs for 2nd, 3rd and 4th order.

2. all the functions have doctrings. 3. modified the function that generates perturbed structures 4. updated def get_cutoffs -- generates a grid of cutoffs that satisfy the given # of required DOFs for 2nd, 3rd and 4th order interaction. 5. sequentially performs the linear regression for each cutoff. 6. modified the files that are saved during the execution of workflow

2. removed files that are not required to be saved during the workflow execution.

fixes

2. adding the use of MACE's RelaxMaker & StaticMaker

hrushikesh-s added 4 commits September 13, 2023 22:03

adding a function to copy hiphive outputs

43dae04

Merge branch 'hiphive' of https://github.com/hrushikesh-s/atomate2 in…

c163fac

…to hiphive

hiphive flow and job

4424f40

bug fix

8a08ab9

JaGeo reviewed Oct 10, 2023

View reviewed changes

hrushikesh-s and others added 9 commits October 11, 2023 16:08

Merge branch 'materialsproject:main' into hiphive

1421e04

first attempt at adding tests

d15921b

At the moment the unit tests are very buggy! Pls, forgive it for now. To be fixed in the iterations to come.

Merge branch 'hiphive' of https://github.com/hrushikesh-s/atomate2 in…

dcfda62

…to hiphive

Merge branch 'materialsproject:main' into hiphive

d5ad270

updates

b635753

debugging error in generate_hiphive_displacements

c16e280

Update hiphive.py

3df39f5

bug fixes

065d908

uncommenting remaining jobs

968f359

janosh force-pushed the main branch from 6ceb1ec to 3764841 Compare October 27, 2023 00:12

hrushikesh-s added 8 commits October 26, 2023 19:29

updates

4ec85ce

bug fix

ef0638e

bug fix

ca44a18

updates

51a790f

updates

c8e1a51

Update phonons.py

627654a

Update hiphive.py

f30d29c

bug fixes

365775f

JaGeo reviewed Oct 27, 2023

View reviewed changes

hrushikesh-s added 5 commits October 26, 2023 23:55

Update hiphive.py

113e970

using the decorator @job(data=["forces", "displaced_structures"])

b055aad

Update hiphive.py

adfd241

Update hiphive.py

9f61b69

Update hiphive.py

2a4c89c

dependabot bot and others added 30 commits June 27, 2024 18:25

commenting out quality control

1f638c6

temporarily commenting out quality control job

adding renormalization changes from Junsoo

a0d76b6

customized INCAR tags

319cfc6

using tags in accordance with atomate LD workflow

added anharmonicphonons block in pyproject.yaml

5515d1a

update the logic to use lambda=0 for TI

e957c61

the decision is now made based on whether n_imaginary > 0 from the fitting_data.json @ 0K

adding fw configs to individual jobs

b1f93de

new displacement strategy

4bcd97e

Trying different strategies for perturbing the atoms in the supercell -- 1. Fixed displacement 2. MC Rattle 3. MD informed displacements Adding a penalty to the displacements that are generated out of the distribution.

added CHGNet MD step to generate displacement snapshots

a41d31e

updateing arguments of the MD run

a0d2050

minor bug fixes to MD flow

f1217b6

updated MD args for running mp-23339

2790070

new cutoffs sampling startegy

58aedd1

1. selects cutoffs based on target DOFs for 2nd, 3rd and 4th order.

1. refactored the hiphive flow

7befb93

2. removed files that are not required to be saved during the workflow execution.

Merge branch 'hiphive_new_sampling_startegy' into hiphive

06b2a3a

fixes

c61dcf7

Merge pull request #11 from hrushikesh-s/hiphive_new_sampling_startegy

dbbfa40

fixes

ruff fixes

af590ba

mypy fixes to hiphive flow

20ad5c3

updated logic to remove cutoffs that result in 0 constraints!

bbebbd4

1. no major code logic updates

5f977f0

2. adding the use of MACE's RelaxMaker & StaticMaker

reduced the user_kpoints_settings -- "reciprocal_density" to 100

02ff2cc

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[WIP] Multi step lattice dynamics workflow using hiPhive #559

[WIP] Multi step lattice dynamics workflow using hiPhive #559

hrushikesh-s commented Oct 10, 2023

codecov bot commented Oct 10, 2023 •

edited

Loading

JaGeo left a comment

JaGeo Oct 10, 2023

hrushikesh-s Oct 10, 2023

JaGeo Oct 10, 2023

hrushikesh-s Oct 10, 2023

JaGeo Oct 10, 2023

Zhuoying Oct 10, 2023

JaGeo Oct 10, 2023

hrushikesh-s Oct 10, 2023

Zhuoying Oct 10, 2023

JaGeo Oct 10, 2023

JaGeo Oct 27, 2023

[WIP] Multi step lattice dynamics workflow using hiPhive #559

Are you sure you want to change the base?

[WIP] Multi step lattice dynamics workflow using hiPhive #559

Conversation

hrushikesh-s commented Oct 10, 2023

Summary

Additional dependencies introduced (if any)

TODO (if any)

Checklist

codecov bot commented Oct 10, 2023 • edited Loading

Codecov Report

JaGeo 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

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

codecov bot commented Oct 10, 2023 •

edited

Loading