TGLF interface

[aaronkho]

• Added TGLF transport model interface, structure based on QuaLiKiz transport model
• Added extra physics quantities to TGLF-based interface
• Added collisionality multiplier to TGLF-based interface

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[aaronkho]

@theo-brown Please reassign if you think someone else should also review this.

[theo-brown]

Great stuff! Excited to see this come in. There might be a couple of further rounds of review if others from the GDM team want to chip in, but we should be able to get the ball rolling.

There are few linting/style errors that should be caught by running the linter/formatter. Have you got that set up?

Also please could you investigate the failing tests and make fixes where possible?

Thanks!

[aaronkho]

Ok I fixed the majority of the CI tests that I could. It seems the current failing test is just due to the safe_divide changing some parameters by a marginal amount. The relative tolerance for matching is 1.0e-10, but many values are in the 1.0e-9 range which is probably small enough to not cause concern.

This will either take someone else to re-run the ITER hybrid TGLFNN test case and replace the appropriate test netcdf file or I could try to remove the offending safe_divide (not sure which one is the major contributor though).

[aaronkho]

@jcitrin As mentioned above, I added safe_divide to some quantities in the TGLF-based calculations. This causes the ITER-hybrid TGLFNN test case to fail. As far as I can tell, all other tests pass.
The runner logs show that the differences are extremely marginal but still enough to fail the test. Would someone be able to update that test case based off of this PR? Also, I've addressed the majority of the review comments from @theo-brown, what are the next steps?

[theo-brown]

I'll start the import process, and I can rerun the test for you.

[theo-brown]

What happens next is:

1. We import it into the internal version of the repo (done)
2. Internal review, potentially making some changes (in progress)
3. Merge with attribution preserved

[theo-brown]

Have you got any sim results for e.g. the iterhybrid case?

[theo-brown]

This is really awesome, btw. Works super seamlessly.

Below is TGLF (dashed) vs TGLFNN-ukaea (solid).

[aaronkho]

Great to see! And glad to hear that you're happy with it thus far. :)

[theo-brown]

Sorry for the delay in bringing it in, getting TGLF to compile within Google is a bit complicated.

[theo-brown]

Ok, TGLF now working internally. A few minor changes have come up in internal review, so the final merged version might have some differences. This PR will be merged soon!

[aaronkho]

Awesome! We also noticed a minor issue in the default values set in tglf_transport_model.py, specifically the defaults should be use_bpar = False; use_bper = True, but we accidentally coded it in the other way around (use_bpar = True; use_bper = False). Can you correct this on your internal version?