-
-
Notifications
You must be signed in to change notification settings - Fork 200
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
Docs modifications to use QuadratureTraining instead of GridTraining #729
Conversation
@@ -113,7 +113,7 @@ Then finally defining and optimizing using the `PhysicsInformedNN` interface. | |||
|
|||
```@example param_estim | |||
discretization = NeuralPDE.PhysicsInformedNN([chain1, chain2, chain3], | |||
NeuralPDE.GridTraining(dt), param_estim = true, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
dt is still defined?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It seems to be used in other places in this file, thats why I didnt remove it
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes, dt
is used for evenly spaced time grid for the data
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
but it's not required when changing to QuadratureTraining.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes, it can be removed
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ok I think I have fixed this one
@@ -65,7 +65,7 @@ function norm_loss_function(phi, θ, p) | |||
end | |||
|
|||
discretization = PhysicsInformedNN(chain, | |||
GridTraining(dx), | |||
QuadratureTraining(), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
dx is still defined.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
same as dt from the other file, it seems to be used in other places in this file, thats why I didnt remove it
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
But none of this file should be using it anymore?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok I removed the instances of dx but it seems like the program is still creating a grid of some kind (see line 61 and below), maybe only for evaluation. I am assuming I don't want to delete all that stuff?
Docs is still erroring at https://github.com/SciML/NeuralPDE.jl/actions/runs/6185760782/job/16791965357?pr=729#step:5:86 |
It at least wouldn't need a bump. What was the last version it was using wherei t was passing? |
This is passing for me locally without bringing up the ComponentArrays issue. @ChrisRackauckas Can we rebuild the CI or something? The the changes I made should not affect ComponentArrays |
b41c094
to
aa87003
Compare
I rebased onto master. Let's see how that goes. |
Just please set the ComponentArrays bound to what it needs to be so we can merge. |
I am not sure whats happening but I tried building the docs locally and it fails with the same error. But if I tried reducing the maxiters to 1200 from 2000 in https://github.com/SciML/NeuralPDE.jl/blob/master/docs/src/tutorials/constraints.md?plain=1#L91, it surprisingly passes. |
The docs build is now failing because of
|
@ChrisRackauckas any idea why the doc build is erroring with https://github.com/SciML/NeuralPDE.jl/actions/runs/6323742800/job/17171915807?pr=729#step:5:43 I traced it to https://github.com/SciML/Integrals.jl/blob/master/lib/IntegralsCubature/src/IntegralsCubature.jl#L171 locally but I am not what is the exact reason. |
@sdesai1287 can you rebase your branch on master? Compat bounds were introduced for |
…ce for special likelihood term
Need to update docs for diff training strategies,etc
…ion 1, (keep existing compat)
The docs successfully built on master, so I put this on top of a successful master. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@ChrisRackauckas docs build is passing. I think we can merge this.
🎉 |
solved #634