odinn-sciml / huginn.jl Goto Github PK
View Code? Open in Web Editor NEWFast and flexible glacier ice flow models
License: MIT License
Fast and flexible glacier ice flow models
License: MIT License
So far we are not including anything related to the mass balance in Huginn.jl. We need to make a release for Muninn.jl, add it as a dependency here, and add tests in PDE_UDE_solve.jl
including a temperature-index model.
Right now everything is implemented in-place in order to be memory efficient for the PDE solving. This is great for the forward PDE runs, but we need an out-of-place version to be run with AD, particularly for ReverseDiff.jl.
This implies to basically reproduce the same workflow as in place, changing all functions like SIA2D!
-> SIA2D
, to indicate that it's out-of-place. All the code is available in the current ODINN.jl repo.
The out-of-place workflow has been tested with the Halfar solution, but we still need an example with real glaciers. To be added in the PDE_UDE_solve.jl
file called in runtests.jl
.
We already have a test
flag for some workflows to have specific configurations during tests. However, this should become more systematic, especially following recent issues during tests.
For example, there seems to be an issue when removing workers in GH CI. If there is first a test which uses 4 workers and then comes another one which asks only for 1 worker, the code is made in such a way that 3 workers will be removed. This works fine in local, but seems to crash in GH CI. By using that flag in the parameters, we could avoid removing workers when it test. Such idea could be re-used for many future issues like this.
The test pde_solve_test
,
@testset "PDE solving integration tests" pde_solve_test(; rtol=0.01, atol=0.01, save_refs=false, MB=false, fast=true)
fails when fast=false
. This just runs the test with a larger list of glaciers, but there is some of them with missing data that make the test fail.
if fast
rgi_ids = ["RGI60-11.03638", "RGI60-11.01450", "RGI60-08.00213", "RGI60-04.04351", "RGI60-01.02170"]
else
rgi_ids = ["RGI60-11.03638", "RGI60-11.01450", "RGI60-08.00213", "RGI60-04.04351", "RGI60-01.02170",
"RGI60-02.05098", "RGI60-01.01104", "RGI60-01.09162", "RGI60-01.00570", "RGI60-04.07051",
"RGI60-07.00274", "RGI60-07.01323", "RGI60-01.17316"]
end
It will be good idea to omit those or fix the problem in some other way.
Right now we're using multiprocessing to parallelize the solve of the SIA for each glacier. This is effective in accelerating the simulations, but we still haven't explored the use of multithreading. Multithreading could be used inside each process for each glacier, in order to solve the SIA in parallel. I'm aware DifferentialEquation.jl enables this, so it might be fairly easy to implement.
Right now we are duplicating the environment.yml
file in each of the repos for the CI to have access to the Python environment. This is problematic, since any update to the Python environment will have to be extended to the 4 ODINN repositories.
For that, we should update the CI to download (e.g. using wget
) the environment.yml
file from Sleipnir.jl's GitHub repo.
Currently Huginn.jl
has implemented the same workflow that is working in Sleipnir.jl
to run the tests automatically after every commit to main. However, in Huginn this is not working, and the problem seems to be comming from PyCall
. Observing both stacktraces, we observe that in Huginn it seems that the conda environmnet is not properly set:
This error does't not show in Sleipnir. The problem is probably coming from the fact that here Sleipnir is being installed and imported as a separate module. Here is something in the PyCall documentation that may help find the solution.
This issue is used to trigger TagBot; feel free to unsubscribe.
If you haven't already, you should update your TagBot.yml
to include issue comment triggers.
Please see this post on Discourse for instructions and more details.
If you'd like for me to do this for you, comment TagBot fix
on this issue.
I'll open a PR within a few hours, please be patient!
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.