Fix gaussian basis set

[ladinesa]

A parsing error for gaussian has been reported for this upload . The basis set cannot be resolved so I simply set it to None, but @ndaelman-hu you should know more.

The method normalizer also fails so I put up a fix in nomad.

[coveralls]

Pull Request Test Coverage Report for Build 9285765274

Details

• 6 of 7 (85.71%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 93.039%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
electronicparsers/gaussian/parser.py	6	7	85.71%
<!--	Total:	6	7	85.71%	-->

Totals
Change from base Build 9252035458:	0.0%
Covered Lines:	35712
Relevant Lines:	38384

---

💛 - Coveralls

[JosePizarro3]

Hi there,

I cannot see this upload, but is this the last one by Akseli Mansikkamäki (https://nomad-lab.eu/prod/v1/gui/upload/id/MASTxxsySg-SkFffP9j2Zw)? Do you know details on the data? I think he is studying magnetism in charged molecules, but I'd like to make sure.

[ndaelman-hu]

Hi there,

I cannot see this upload, but is this the last one by Akseli Mansikkamäki (https://nomad-lab.eu/prod/v1/gui/upload/id/MASTxxsySg-SkFffP9j2Zw)? Do you know details on the data? I think he is studying magnetism in charged molecules, but I'd like to make sure.

Neither I can see the upload. @ladinesa Could you ask permissions to share the raw files or have us included?

[ndaelman-hu]

@ladinesa Okay, I got access to the failing upload. You made the right call: there simply isn't any basis set mentioned. Likely, this is due to:

1. this calculation starting from a previous SCF (not uploaded).
2. this is a frequency calculation, with a different emphasis.

Still, this is pretty bad practice on Gaussian's side, as SCF routines shouldn't tacitly assume the basis set. Fro the long-term, I'll see if I can hunt down their default, though Gaussian is likely to have some complex logic in assigning it.

[ndaelman-hu]

@JosePizarro3 Regarding the upload you linked, all the Gaussian files parsed well there. The issue seems to be with Orca, as we don't support CASSCF, yet. I have this on the planned list, but quite low.

We either add a schema there, or we harden the parsing not to fail for unrecognized methods.

[ndaelman-hu]

Digging further, the default is STO-3G. I can see to improve on the current correction then.

[ladinesa]

@ndaelman there is a problematic test. is it possible for you to look at it so we can merge it already? thanks

[ndaelman-hu]

So, some final comments here that will lead to new issues:

1. we need to add missing basis set specs to our tests
2. we will have to extend "GEN" (or "Gen") parsing, as this indicates a user-defined basis set