SciML / SciML/NeuralOperators.jl

Implement U-shaped Neural Operator (U-NO)

Open
#110 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 U-NO, which adapts the U-Net encoder-decoder architecture to neural operators in function spaces.

Reference

  • Rahman et al., "U-NO: U-shaped Neural Operators," Transactions on Machine Learning Research, 2023. arXiv:2204.11127

Description

U-NO applies the U-Net multi-scale paradigm to neural operators, using integral operators that map between functions at different spatial scales. The encoder downsamples through spectral truncation, the decoder upsamples, and skip connections preserve high-frequency information. Reports up to 44% improvement on Navier-Stokes benchmarks over FNO.

This is architecturally related to the existing FNO implementation but requires multi-resolution spectral layers with downsampling/upsampling and skip connections between scales.

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

Start by locating the existing FNO implementation and its tests or entry points. Read the U-NO paper alongside the FNO code to define the multi-resolution spectral layers, downsampling and upsampling, and skip connections. Done means the U-NO architecture is implemented and its behavior is covered by appropriate tests, including the intended Navier–Stokes use case.

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
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.