Remove unused fieldvector

[charleskawczynski]

Followup to #2130

[charleskawczynski]

bors r+

[bors]

Build failed:

• buildkite/climaatmos-ci

[charleskawczynski]

Why is

        Fields.bycolumn(axes(Y.c)) do colidx
            precipitation_tendency!(p.Yₜ, Y, p, t, colidx, p.precip_model)
        end # TODO: Set the diagnostics without computing the tendency.

used in the diagnostics? cc @trontrytel . Can we split this into precompute and add a function that adds fluxes so that we can remove p.Yₜ?

[trontrytel]

Not sure why? Not wanting to cache more things and or duplicate the code?

I don't have the bandwidth to fix this this week. Feel free to open an issue and tag me on it. I can try getting it done before the next sprint

used in the diagnostics? cc @trontrytel . Can we split this into precompute and add a function that adds fluxes so that we can remove p.Yₜ?

[szy21]

I think we might use it to make sure the precipitation tendency is consistent with the state, as we call set_precomputed_quantities! in the diagnostics which updates the thermo state. It might be ok to remove it (i.e. I think it's ok if the diagnostics we save are slightly inconsistent, but maybe other people have different opinions.)

[charleskawczynski]

I think we might use it to make sure the precipitation tendency is consistent with the state, as we call set_precomputed_quantities! in the diagnostics which updates the thermo state. It might be ok to remove it (i.e. I think it's ok if the diagnostics we save are slightly inconsistent, but maybe other people have different opinions.)

I think it's fine to precompute things needed for precipitation tendencies, but the precip tendendencies themselves don't seem to be used in the diagnostics-- Also if they are, then I think we can get them from the integrator?

[charleskawczynski]

@trontrytel, for now, I've just wrapped the tendency updates in if-blocks (if !isnothing(Yₜ); Yₜ.var += ...; end), so that we can pass in nothing from diagnostics-- all other quantities are computed (so p should be appropriately updated). Is this okay for now?

[szy21]

I think we might use it to make sure the precipitation tendency is consistent with the state, as we call set_precomputed_quantities! in the diagnostics which updates the thermo state. It might be ok to remove it (i.e. I think it's ok if the diagnostics we save are slightly inconsistent, but maybe other people have different opinions.)

I think it's fine to precompute things needed for precipitation tendencies, but the precip tendendencies themselves don't seem to be used in the diagnostics-- Also if they are, then I think we can get them from the integrator?

We calculate col_integrated_rain and col_integrated_snow in precipitation_tendency!, that's probably why we call it here. We can move them somewhere else though - we don't need them in the tendency.

[trontrytel]

@trontrytel, for now, I've just wrapped the tendency updates in if-blocks (if !isnothing(Yₜ); Yₜ.var += ...; end), so that we can pass in nothing from diagnostics-- all other quantities are computed (so p should be appropriately updated). Is this okay for now?

Sounds good. Apologies it's such a mess over there. It would be great to allocate some time and straighten things up (looking at myself pointedly).

The logic we want would be to put compute_precipitation_cache somewhere in precomputed_quantities, instead of calling it from the tendency? And then use it from both the tendency and diagnostics?

[charleskawczynski]

Sounds good. Apologies it's such a mess over there. It would be great to allocate some time and straighten things up (looking at myself pointedly).

The logic we want would be to put compute_precipitation_cache somewhere in precomputed_quantities, instead of calling it from the tendency? And then use it from both the tendency and diagnostics?

No problem. I'm just trying to incrementally improve on CliMA/ClimaCore.jl#1467 for now. It's not clear to me how to best structure things, but there's definitely room for improvement.

[charleskawczynski]

bors r+

[bors]

Build succeeded!

The publicly hosted instance of bors-ng is deprecated and will go away soon.

If you want to self-host your own instance, instructions are here.
For more help, visit the forum.

If you want to switch to GitHub's built-in merge queue, visit their help page.

• buildkite/climaatmos-ci
• docbuild
• format