uno's Introduction

U-NO

This repository contains code accompanying the paper: U-NO: U-shaped Neural Operators

UNO_Tutorial.ipynb - A step-by-step tutorial for using and buidling U-NO. Link to Google colab

U-NO is now available on Neural Operator library. Quick Start

Requirements

PyTorch 1.11.0

Files

Files	Descriptions
integral_operators.py	Contains codes for Non-linear integral operators for 1D, 2D, and 3D functions.
UNO_Tutorial.ipynb	A tutorial on using the integral operators and U-NO.
Darcy Flow
darcy_flow_main.py	Script for loading data, training, and evaluating training UNO performing 2D spatial convolution for solving Darcy Flow equation.
darcy_flow_uno2d.py	UNO architectures for solving Darcy Flow equation.
train_darcy.py	Training routine for Darcy flow equations.
data_load_darcy.py	Function to load Darct-flow data.
Navier–Stokes
data_load_navier_stocks.py	Function to load Navier–Stokes data generated by data generator prodived
ns_uno2d_main.py	Script for loading data, training, and evaluating the UNO (2D) autoregressive in time for Navier–Stokes equation.
ns_train_2d.py	Training function for UNO(2D) in time for Navier–Stokes equation
navier_stokes_uno2d.py	UNO(2D) architecture in time for Navier–Stokes equation.
ns_uno3d_main.py	Script for loading data,training and evaluating the UNO(3D) performing 3D (spatio-temporal) convolution for Navier–Stokes equation.
navier_stokes_uno3d.py	UNO(3D) achitectures performing 3D convolution for Navier–Stokes equation.
ns_train_3d.py	Training function for UNO(3D) for Navier–Stokes equation.
Supporting Files
Data Generation	Folder contains scripts to generate data from Navier–Stokes equation and Darcy flow
utilities3.py	Contains supporting functions for data loading and error estimation.

Data

Link to two files containing 2000 simulations of Darcy Flow equation: Google Drive Link

The Data Generator folder contains script for generating simulation of Darcy Flow and Navier-Stocks equation.

uno's People

Contributors

Stargazers

Watchers

uno's Issues

Project License?

Would be useful to be able to use this in other projects. MIT License would be the simplest.

Thank you for the excellent work! I want to ask some questions about NS data generation, I run the code without any modification and get a file of a dozen GB, however I'm not sure how many samples are in the file. How to determine the number of samples in the file?

Recommend Projects