Rahul Manavalan

More Benchmarks

13

It would be nice to have more benchmarks as they discuss here - https://arxiv.org/pdf/2111.05512.pdf - https://arxiv.org/pdf/2203.13181.pdf

Mutli-dimensional input-grid for DeepONet

4

In https://github.com/SciML/NeuralOperators.jl/blob/master/src/DeepONet.jl#L122-L124, multi-dimensional input grid is handled using an Argument error. Is there a reason this was done this way? Also, if it is a matter of contracting AbstractArray(s), we...

Automating the SINDY workflow with Dataset Augmentation and Reduction

2

- Would ideas for augmenting the dataset (DelayEmbeddings.jl) and reducing the data, be welcome (in the scope of this package) for automating the model discovery workflow ?

High Dimensional Kernel Estimation - Tensor Train Estimation

1

I am curious if an implementation of the paper below is of interest to the package. https://arxiv.org/abs/2108.00089

I had significant errors trying to solve a 1D Poisson system Tridiagonal([1,-2,1]) with the implementation in the examples. While I realize that this is not the scope of the implementation,...

Julia API

2

A Julia API could be really useful considering the Julia Language has a lot of new users in Scientific Computing.

Second order time derivative - The Wave Equations

8

@xtalax Could you help me , implement this feature ? Any pointers on how to get started would be helpful

Equivalent Circuit Model of Electrochemical/Solid state Cells

2

Lithium Ion Battery Models using Equivalent Circuit Models could be a welcome addition to the features in this Standard Library. This introductory course from Boulder introduces the idea. http://mocha-java.uccs.edu/ECE5710/ECE5710-Notes02.pdf @rkurchin...

Mixed Space and Time Derivatives

1

Consider  Any unspecified symbol is a real constant. The time domain is not explicitly separable. I guess you could regard the time variable in the mixed derivatives as a...

Consider using fft to rfft

1

I ran into this issue a few months ago. So apologies for the late report. For the [FNO](https://github.com/SciML/NeuralOperators.jl/blob/main/src/Transform/fourier_transform.jl/#10), rfft is used. From https://github.com/FluxML/Zygote.jl/issues/1406 , see that Zygote's gradient is wrong....