SciML / SciML/NeuralOperators.jl

Implement Mamba Neural Operator (Alias-Free MNO)

Open
#123 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
41
Forks
15
Avg merge
13h 14m
Merged PRs (30d)
12

Description

Summary

Implement the Mamba Neural Operator, which applies selective state-space models (Mamba) to PDE solving with global receptive fields and linear complexity.

Reference

  • "Alias-Free Mamba Neural Operator," NeurIPS 2024. Paper

Description

The Mamba Neural Operator applies the Mamba selective state-space model architecture to operator learning. It achieves global receptive fields with linear complexity (vs quadratic for transformers), uses adaptive state-space matrices, and includes an alias-free design to prevent spectral aliasing. Reports up to ~90% error reduction over transformer baselines and greatly improved long-time stability for autoregressive rollouts.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No files, tests, or entry points are named. Start by reading the linked NeurIPS 2024 paper and inspecting the repository's existing neural-operator implementations to determine the integration point. Done means the alias-free Mamba Neural Operator is implemented with selective state-space modeling, global receptive fields, linear complexity, and support for PDE solving.

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

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.