Uploading data to MPContribs

Materials Project
1

Hello,
My group has predicted several new stable compounds for different crystal systems.
I want to contribute this data to MP. I wonder how the MP-ID is assigned to the structures and how detailed information is required to be uploaded.

Thank you,
Daviti

2

Hi @Davit_Gochitashvili, the MP-ID is assigned internally as part of our core data processing pipelines. Did you use VASP to calculate these structures? If so, and if your calculations are compatible with MP’s, we might be able to incorporate your structures into our core data (with appropriate credit of course).

3

Thank you for a quick reply.
Yes, we used VASP for calculations. So Composition and CIF file should be enough or more information is needed or the website will automatically parse out information form the CIF files?

4

Ok, great. Let’s check first whether your calculations happen to be compatible with MP. @Aaron_Kaplan would be able to point you to documentation on how to go about it.

1 Like
5

Hey @Davit_Gochitashvili, we check compatibility of the calculation settings, especially the pseudpotentials / POTCARs, used in a calculation. You can test our new validation software by running pip install pymatgen-io-validation, and then run

from pymatgen.io.validation import ValidationDoc

valid_doc = ValidationDoc.from_directory("path to your VASP calc")
print(valid_doc.valid)

Any reasons for validation failure will be contained in valid_doc.reasons, and any warnings in valid_doc.warnings

2 Likes
6

Thank you very much! I will check the runs and will, get back to you soon.

7

Hello,
I checked compatibility of my calculations using the code: it returned True.
What should be the next steps to upload data? I can created a project MPContribs portal.
https://contribs.materialsproject.org/projects/SnCompounds

Now I am going to upload CIF files of new stable Sn-based binary phases we predicted in our study. I will use arbitrary mp-IDs.

Please let me know if there is anything I should consider.

Best,
Daviti

8

Sounds good! Feel free to use a string for the identifier field in MPContribs that makes sense to you internally. How many structures are you planning to upload? Please let me know once you’re done with it. We’ll take a look at them and decide whether we’d also like to get your raw VASP directories to include them into our core dataset. Thanks!

9

Thank you for the feedback.
Okay, I will use more sensible names for the IDs.
There should be around 10 structures that are stable at ambient conditions (0 GPa and 0K).

I am planning to upload predicted structures from our other published studies as well, but I will register those as separate projects.

I need to prepare data to upload. It might take some time. I will let you know once I finish.

Thank you for your help!

1 Like
10

Hi @tschaume,

I noticed that all structures on MP have electronic structure information listed, like band gap. I wonder if I need to include that data to my contributions.

Thank you,
Daviti

11

You don’t need to but please feel free to add any additional data to the data component in MPContribs that you think might be interesting for general MP users to know. Once we take your original structures and add them to our core set of materials, their properties such as bandgaps will be calculated automatically (and will hopefully be the same as in your MPContribs project).

1 Like
12

Hello @Aaron_Kaplan,
I was trying to use pymatgen.io.validation for the layered structures calculated using optb86b-vdw. In reasons I received this error message:
['NO MATCHING MP INPUT SET --> no matching MP input set was found. If you believe this to be a mistake, please create a GitHub issue.']
Could you please give information what functionals are valid for the calculations including Van der Waals interactions?

Best,
Daviti

13

Hi @Davit_Gochitashvili, we don’t use vdW corrections in MP’s core calculations currently. It’s obviously needed for many materials, but would also require development of another “compatibility” scheme to align the thermodynamic convex hulls of a vdW-corrected and uncorrected functional

If the goal is to contribute calculations to MP, I would strongly recommend using either our workflow orchestration tool atomate2 - examples and explanation here - or the pre-defined VASP input sets from pymatgen:

  • MPRelaxSet: PBE or PBE+U (for transition-metal oxides)
  • MPScanRelaxSet: r2SCAN meta-GGA calculations

Alternatively, you can submit structures and we’ll run them when we get a chance!

1 Like