NeuralPDE.jl
NeuralPDE.jl copied to clipboard
SciML
Physics-Informed Neural Networks (PINN) Solvers of (Partial) Differential Equations for Scientific Machine Learning (SciML) accelerated simulation
When training a neural network is helpful to know the statistics of the loss function. In Tensorflow or deepXDE, for example, usually one can plot the loss function vs the...
Hi everyone, I have been working though the tutorial "Optimising Parameters (Solving Inverse Problems) with Physics-Informed Neural Networks (PINNs)". I am getting an error on the sixth box when defining...
Couple of strategy/scenarios 1) train-to-train model - train a sub-class of PDEs with NeuralOperator - transfer learning pre-trained prediction from NeuralOperator problem to PINNs - PINNs training with pre-trained model...
This is a script that solves a linear SDE using the KKL expansion for the Wiener process. A couple things -- 1. This script creates individual variables for the modes....
@hpieper14 @ChrisRackauckas @frankschae @KirillZubov Design discussions here so that they don't get slurped up by slack.
When trying to setup CUDA with NNODE it seems I hit some Zygote issues with the adjoints of `Array`. @DhairyaLGandhi could I get some help on this? ```julia using NeuralPDE,...
It was disabled because it was done improperly. It defined the RODE as having a fixed Wiener process, which means it would not generalize to giving solutions from the full...
This should error @Vaibhavdixit02 because both AutoZygote and BFGS don't do constraints, so...
We should get some examples showing bigger neural networks training things like Lotka-Volterra. Might need GPUs.
This is due to some of the Integrals.jl derivative issues which are getting fixed up by @sharanry, we should test after that's done.