8000 GitHub - M3RG-IITD/LaMO: Latent Mamba Operator for Partial Differential Equations
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

M3RG-IITD/LaMO

Repository files navigation

Latent Mamba Operator for Partial Differential Equations

License: MIT ArXiv

This repository contains the official implementation of our Latent Mamba Operator for Partial Differential Equations.

Latent Mamba Operator for Partial Differential Equations
Karn Tiwari, Niladri Dutta, N M Anoop Krishnan, Prathosh A P



Figure 1. Overview of LaMO.

Codebase for Reproducibility: Getting Started

  1. Install Python 3.12.3. For convenience, please go ahead and execute the following command.
pip install -r requirements.txt
  1. Prepare Data. You can obtain experimental datasets from the following links (Download).
Dataset Task Geometry Link
Elasticity Estimate material inner stress Point Cloud [Google Cloud]
Plasticity Estimate material deformation over time Structured Mesh [Google Cloud]
Navier-Stokes Predict future fluid velocity Regular Grid [Google Cloud]
Darcy Estimate fluid pressure through medium Regular Grid [Google Cloud]
AirFoil Estimate airflow velocity around airfoil Structured Mesh [Google Cloud]
Pipe Estimate fluid velocity in a pipe Structured Mesh [Google Cloud]
  1. Train and evaluate the model. We provide the experiment scripts of all benchmarks under the folder ./scripts/. You can reproduce the experiment results as the following examples:
bash scripts/LaMO_Elas.sh # for Elasticity
bash scripts/LaMO_Plas.sh # for Plasticity
bash scripts/LaMO_NS.sh # for Navier-Stokes
bash scripts/LaMO_Darcy.sh # for Darcy
bash scripts/LaMO_Airfoil.sh # for Airfoil
bash scripts/LaMO_Pipe.sh # for Pipe

Note: You must change the argument --data-path in the above script files to your dataset path.

  1. Develop your own model. Here are the instructions:

    • Add the model file under the folder ./models/.
    • Add the model name into ./model_dict.py.
    • Add a script file under folder ./scripts/ and change the argument --model.

Main Results



Figure 2. Main Results Compared with SOTA Operator. L2 Error is reported.

LaMO achieves 32.3% averaged relative increment over the previous second-best model i.e, Transolver.

Showcases



Figure 3. Showcases. Error heatmap is plotted for Transolver and LaMO for Darcy, Plasticity, and Navier-Stokes.

Citation

Please consider citing our paper if you find it helpful. Thank you!

@misc{tiwari2025latentmambaoperatorpartial,
      title={Latent Mamba Operator for Partial Differential Equations}, 
      author={Karn Tiwari and Niladri Dutta and N M Anoop Krishnan and Prathosh A P},
      year={2025},
      eprint={2505.19105},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2505.19105}, 
}

Contact

If you have any questions, please feel free to contact me at:

Karn Tiwari: karntiwari@iisc.ac.in

Acknowledgement

We appreciate the following GitHub repos a lot for their valuable code base or datasets on which we have built our code:

  1. https://github.com/neuraloperator/neuraloperator

  2. https://github.com/thuml/Transolver/tree/main

  3. https://github.com/state-spaces/mamba/tree/main

  4. https://github.com/goombalab/hydra/tree/main

  5. https://github.com/MzeroMiko/VMamba

About

Latent Mamba Operator for Partial Differential Equations

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published
0