[WIP] Migrate neural network-based indicators to new repository #1
Conversation
Co-authored-by: Hendrik Ranocha <[email protected]>
|
|
||
| [compat] | ||
| MuladdMacro = "0.2.4" | ||
| # Trixi = "0.6" |
There was a problem hiding this comment.
TODO @sloede: uncomment this line
examples/tree_1d_dgsem/elixir_euler_blast_wave_neuralnetwork_perssonperaire.jl
Outdated
Show resolved
Hide resolved
…ixi.jl and TrixiSmartShockFinder.jl
|
This looks ready from my side. Thus, a review would be great. Please do not merge yet though, as I'd like to check with the upcoming Trixi v0.6 first. |
| using TrixiSmartShockFinder | ||
| using Trixi | ||
|
|
||
| # TODO: Remove once Trixi.jl is released without these exports |
| # TODO: Remove once Trixi.jl is released without these exports | ||
| NeuralNetworkPerssonPeraire = TrixiSmartShockFinder.NeuralNetworkPerssonPeraire | ||
| NeuralNetworkRayHesthaven = TrixiSmartShockFinder.NeuralNetworkRayHesthaven | ||
| IndicatorNeuralNetwork = TrixiSmartShockFinder.IndicatorNeuralNetwork |
There was a problem hiding this comment.
I guess there are TODO notes in every elixir - I will skip them
| BSON = "0.3.3" | ||
| Flux = "0.13.15, 0.14" | ||
| OrdinaryDiffEq = "6.49.1" | ||
| # Trixi = "0.6" |
Co-authored-by: Hendrik Ranocha <[email protected]>
I'd say no. We will not register this as a package for now, and I don't plan on fixing issues that would arise from Dependabot-triggered updates. Since the package is so tiny, I also do not think setting up Documenter.jl is absolutely necessary |
|
Fine with me 👍 This is basically ready to be merged. Do you want to make the breaking release of Trixi.jl first? |
Thanks for reviewing and double-checking my thought process!
Yes, such that we can activate the currently commented Trixi.jl compat bounds for v0.6 and verify that they work. |
See also the Trixi.jl counterpart trixi-framework/Trixi.jl#1701.