Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Integrate LES functionality #309

Merged
merged 13 commits into from
Jul 11, 2019
Merged

Conversation

glwagner
Copy link
Member

@glwagner glwagner commented Jul 3, 2019

This PR starts the processes of integrating LES functionality from glwagner/pass-tests into Oceananigans master. It is very much WIP at the moment.

@glwagner
Copy link
Member Author

glwagner commented Jul 5, 2019

@ali-ramadhan this PR is ready for a first review.

Possibly we should also add docs in this PR. I could also envision a few more tests to ensure the correctness of the turbulent diffusivities in simple flow fields for Constant Smagorinsky and Anisotropic Minimum Dissipation. However this last is not necessary.

@ali-ramadhan
Copy link
Member

ali-ramadhan commented Jul 6, 2019

When we switch to finite volume operators (PR #283) it'll become much clearer where we are imposing no-flux and zero gradient conditions which should help along with comments like # no penetration and # no-gradient condition.

We also switch from to ϊ (\iota\ddot) for interpolation but we can take care of these later.

Would be good to have the turbulence closures using Oceananigans.Operators instead of redefining their own operators when I get back to working on #283 and #47.

@ali-ramadhan
Copy link
Member

Possibly we should also add docs in this PR.

How much of it is similar with dedaLES documentation of turbulence closures? Might be mostly copy paste?

@ali-ramadhan
Copy link
Member

Might be good to close #120 and release v0.9.0 once this PR is merged? Well, maybe release a new minor version once LES is verified.

@ali-ramadhan ali-ramadhan added the feature 🌟 Something new and shiny label Jul 6, 2019
@ali-ramadhan ali-ramadhan added this to the LES closures milestone Jul 6, 2019
Copy link
Member

@ali-ramadhan ali-ramadhan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me so far. With a minor fix all the tests pass (+ GPU tests on Engaging).

Aim of this PR should be to integrate LES closures ASAP and then we can worry about verifying LES and cleaning things up in future PRs.

@ali-ramadhan
Copy link
Member

I could also envision a few more tests to ensure the correctness of the turbulent diffusivities in simple flow fields for Constant Smagorinsky and Anisotropic Minimum Dissipation. However this last is not necessary.

Hmmm, wonder if it's worth doing this if we're going to do some more rigorous LES verification anyways? The rigorous tests probably won't pass initially which will lead us to do these simpler tests first maybe.

@codecov
Copy link

codecov bot commented Jul 6, 2019

Codecov Report

Merging #309 into master will decrease coverage by 6.39%.
The diff coverage is 63.63%.

Impacted file tree graph

@@            Coverage Diff            @@
##           master     #309     +/-   ##
=========================================
- Coverage   72.57%   66.17%   -6.4%     
=========================================
  Files          24       25      +1     
  Lines        1032     1159    +127     
=========================================
+ Hits          749      767     +18     
- Misses        283      392    +109
Impacted Files Coverage Δ
src/Oceananigans.jl 62.5% <ø> (-37.5%) ⬇️
src/time_steppers.jl 74.17% <100%> (-7.14%) ⬇️
src/equation_of_state.jl 100% <100%> (ø) ⬆️
src/closures/constant_diffusivity_closures.jl 65.62% <100%> (+2.29%) ⬆️
src/closures/turbulence_closures.jl 76.66% <22.22%> (-18.79%) ⬇️
src/closures/closure_operators.jl 51.63% <40.54%> (-6.81%) ⬇️
src/closures/constant_smagorinsky.jl 52.77% <51.61%> (+34.59%) ⬆️
src/closures/anisotropic_minimum_dissipation.jl 96.96% <96.96%> (ø)
src/poisson_solvers.jl 39.16% <0%> (-60.01%) ⬇️
... and 10 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8fdff75...11e3043. Read the comment docs.

1 similar comment
@codecov
Copy link

codecov bot commented Jul 6, 2019

Codecov Report

Merging #309 into master will decrease coverage by 6.39%.
The diff coverage is 63.63%.

Impacted file tree graph

@@            Coverage Diff            @@
##           master     #309     +/-   ##
=========================================
- Coverage   72.57%   66.17%   -6.4%     
=========================================
  Files          24       25      +1     
  Lines        1032     1159    +127     
=========================================
+ Hits          749      767     +18     
- Misses        283      392    +109
Impacted Files Coverage Δ
src/Oceananigans.jl 62.5% <ø> (-37.5%) ⬇️
src/time_steppers.jl 74.17% <100%> (-7.14%) ⬇️
src/equation_of_state.jl 100% <100%> (ø) ⬆️
src/closures/constant_diffusivity_closures.jl 65.62% <100%> (+2.29%) ⬆️
src/closures/turbulence_closures.jl 76.66% <22.22%> (-18.79%) ⬇️
src/closures/closure_operators.jl 51.63% <40.54%> (-6.81%) ⬇️
src/closures/constant_smagorinsky.jl 52.77% <51.61%> (+34.59%) ⬆️
src/closures/anisotropic_minimum_dissipation.jl 96.96% <96.96%> (ø)
src/poisson_solvers.jl 39.16% <0%> (-60.01%) ⬇️
... and 10 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8fdff75...11e3043. Read the comment docs.

@christophernhill
Copy link
Member

@ali-ramadhan and @glwagner is the incmod(), decmod() stuff the right thing to do in this. I am trying to think how this will interact with Halo() code elsewhere? Would we be better off filling halos at various intermediate stages for now. We can then tidy up and use wiser halos later? Any thoughts?

@ali-ramadhan
Copy link
Member

ali-ramadhan commented Jul 8, 2019

@ali-ramadhan and @glwagner is the incmod(), decmod() stuff the right thing to do in this. I am trying to think how this will interact with Halo() code elsewhere? Would we be better off filling halos at various intermediate stages for now. We can then tidy up and use wiser halos later? Any thoughts?

Not sure if this is the final form of this PR, but I suggested we just merge the closures as is, with operators that use incmod1 and decmod1, mostly to integrate the closures ASAP. It's definitely inconsistent with the rest of the fields that actually use halos.

We could focus on cleanup of this kind once we verify that the LES closures work, unless this is a pretty urgent issue.

@christophernhill
Copy link
Member

@ali-ramadhan OK. I guess I am missing whether the LES can work with incmod at same time as structures are haloed? e.g in below, unless u is some sort of view (?), it looks slightly wrong to me if u[1] (for example) is in a halo?

@inline function ∂x_caa(i, j, k, grid, u::AbstractArray)
    i⁺ = incmod1(i, grid.Nx)
    return @inbounds (u[i⁺, j, k] - u[i, j, k]) / grid.Δx
end

@ali-ramadhan
Copy link
Member

I think they should as the incmod1 operators should do the exact same job as the halos (assuming periodic boundary conditions), but might also be missing something.

@glwagner
Copy link
Member Author

glwagner commented Jul 9, 2019

How much of it is similar with dedaLES documentation of turbulence closures? Might be mostly copy paste?

Math is the same, but docs syntax is different.

@glwagner
Copy link
Member Author

glwagner commented Jul 9, 2019

@christophernhill @ali-ramadhan using incmod1 will probably work for periodic domains, as it is equivalent to ignoring the existence of halos.

Using incmod1 does defeat the work done to implement side wall boundary conditions.

@christophernhill
Copy link
Member

christophernhill commented Jul 9, 2019 via email

@ali-ramadhan
Copy link
Member

ali-ramadhan commented Jul 10, 2019

@glwagner From yesterday's meeting it sounds like this is ready to be merged?

@glwagner
Copy link
Member Author

@ali-ramadhan I think so! Docs will come in a subsequent PR.

@ali-ramadhan ali-ramadhan merged commit c961d39 into CliMA:master Jul 11, 2019
@glwagner glwagner deleted the integrate-les branch July 24, 2019 19:57
@ali-ramadhan ali-ramadhan changed the title [WIP] Integrate LES functionality Integrate LES functionality Aug 1, 2019
arcavaliere pushed a commit to arcavaliere/Oceananigans.jl that referenced this pull request Nov 6, 2019
[WIP] Integrate LES functionality

Former-commit-id: c961d39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feature 🌟 Something new and shiny
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants