Skip to content

Mathmode/ReCoNNs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 

Repository files navigation

This project contains the code for the use of Regularity Conforming Neural Network (ReCoNNs) architectures for solving PDEs. For a detailed explaination of the methodology behind the code, the preprint of the corresponding article is available at arXiv:2405.14110

The code covers four test examples: A 1D problem with a jump in the derivative, a 2D problem with jumps in the gradient across an interface, the L-shape problem that admits a singularity at the re-entrant corner, and a case of interior material vertices, showing both of the last two types of singularities.

Each example admits its own "Main_X.py" file (Main_1D, Main_jump, Main_L, Main_4_Materials, respectively). The relevant architectures are found in the SRC.Architectures_X files, and the corresponding loss functions in SRC.Loss_X. SRC.Postprocessing is used to plot the 2D solutions once obtained.

Note that, depending on the version of TensorFlow used, the second argument in the callback measure_exponent, corresponding to the index of the singular layer, may need to be changed, as the order of construction varies between versions.

About

No description or website provided.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages