mo_sethet pt 2 - sethet

[jaelynlitz]

The bulk of the port is there - some resulting questions though:

• what is wetdep_map?
• indexes from get_spc_ndx and get_het_ndx - does that just correlate to GasId?
• what is necessary in the last for loop? Obviously need to set het_rates. Also check for Missing values? Don't need to do the write so don't need hetratestrg?
• I see Define gas_pcnst in aero_model and use that one everywhere else. #293 - does gas_pcnst == pcnst == gas_wetdep_cnt?

[codecov]

Codecov Report

Attention: Patch coverage is 89.65517% with 15 lines in your changes missing coverage. Please review.

Project coverage is 96.07%. Comparing base (636ab95) to head (4ddf27f).
Report is 29 commits behind head on main.

Files with missing lines	Patch %	Lines
src/mam4xx/mo_sethet.hpp	89.65%	15 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #295      +/-   ##
==========================================
- Coverage   96.16%   96.07%   -0.09%     
==========================================
  Files          43       44       +1     
  Lines        9486     9651     +165     
==========================================
+ Hits         9122     9272     +150     
- Misses        364      379      +15     

Flag	Coverage Δ
	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jaelynlitz]

This is finally ready to go! With help from @odiazib and @mjs271 yesterday - we tracked down the last of the indexing issues.

The outstanding discrepancy is around ktop. In the fortran, they compute the min ktop value across the 4 columns to optimize/group their operations. Because the C++ is a single column model, we just compute ktop for our column and operate from there. This results in a few values being computed in the C++ that the fortran skips. I have added the C++ computed values to the fortran validation data to allow for this difference while keeping a high tolerance on our tests.

[jeff-cohere]

Looks pretty good to me! I've left a couple of minor comments. Is there anything specific that you'd like a closer look at?

[jaelynlitz]

I removed the Kokkos for loop in gas_washout and that seemed to fix the precision issues I was seeing - let me know if we want to look into putting that back. Please forgive the embarrassingly long lead time on this - ready for final review @jeff-cohere @mjs271 @odiazib @singhbalwinder @overfelt

[singhbalwinder]

Very nice work! Thanks! I have some minor comments but overall looks good to me.

[odiazib]

We may need to change this for loop to a Kokkos::parallel_for.

[odiazib]

Let's use Kokkos::ThreadVectorRange.

[odiazib]

Kokkos::parallel_for here.

[odiazib]

Kokkos::parallel_for.

[odiazib]

Kokkos::parallel_for.

[jaelynlitz]

a few more parallel_fors to add and the passing tolerance for sethet is 1e-3 which seems a little too big

[jaelynlitz]

close in favor of #373