SciML / SciML/NeuralOperators.jl
Implement DeepM&Mnet (Multi-physics Multi-scale Operator Network)
Nobody has claimed this yet.
- Dominant language
- Julia
- Stars
- 41
- Forks
- 15
- Avg merge
- 13h 14m
- Merged PRs (30d)
- 12
Description
Summary
Implement DeepM&Mnet for learning multi-physics, multi-scale operator mappings by decomposing complex systems into coupled sub-networks.
Reference
- Cai et al., "DeepM&Mnet: Inferring the electroconvection multiphysics fields based on operator approximation by neural networks," Journal of Computational Physics, 2021. DOI: 10.1016/j.jcp.2021.110296
Description
DeepM&Mnet decomposes multi-physics systems into a collection of DeepONet sub-networks, where each sub-network learns the operator for one physical field. The sub-networks are coupled through shared inputs/outputs, enabling the overall network to learn complex multi-physics interactions.
This could be implemented as a composition layer that connects multiple DeepONet instances, with configurable coupling between their inputs and outputs.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading the existing DeepONet implementation and the cited DeepM&Mnet paper to understand how the sub-networks and multiphysics couplings are represented. The work is complete when a composition layer can connect multiple DeepONet instances with configurable shared inputs and outputs for coupled physical fields.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- julia
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100