ashiq24/Rayleigh_Benard_Convection

Rayleigh-Bénard Convection Dataset

This dataset contains data for Rayleigh-Bénard convection at different Rayleigh numbers. The data includes velocity fields and temperature distributions on a 128×128 grid.

Dataset Description

The dataset consists of .npz files containing:

• vx: x-component of velocity field
• vy: y-component of velocity field
• temp: temperature field
• time: time points

Sample Data Stats

Installation and Download

pip install -r requirements.txt

Download the dataset

from huggingface_hub import hf_hub_download
import os
os.makedirs("./rb_data", exist_ok=True)

# Download Ra=12,000 dataset
hf_hub_download(
    repo_id="your-username/rb-flow-visualization",
    filename="data/data_12e3.npz",
    local_dir="./rb_data",
    repo_type="dataset"
)

Dataset Loader Details

The dataset loader (dataloader.py) provides two main classes:

1. RBFlowDataset: A PyTorch Dataset class that handles:

   • Loading and preprocessing of .npz files
   • Single frame access
   • Sequence extraction
   • Statistical information

2. load_rb_flow_data: A utility function that creates a DataLoader with:

   • Batch processing
   • Data shuffling
   • Multi-worker support

Accessing Data Statistics

dataset = RBFlowDataset('data/data_12e3.npz')
stats = dataset.stats  # Dictionary containing field statistics
print(f"VX range: {stats['vx_min']} to {stats['vx_max']}")
print(f"Temperature mean: {stats['temp_mean']}")

Basic Usage

from dataloader import load_rb_flow_data

# Load the dataset
dataloader, dataset = load_rb_flow_data(
    data_path='data/data_12e3.npz',  # or data_20e3.npz
    batch_size=32,
    shuffle=True
)

# Iterate through batches
for batch in dataloader:
    vx = batch['vx']  # Shape: [batch_size, nx, ny]
    vy = batch['vy']
    temp = batch['temp']
    # Your processing here

Loading Sequences

from dataloader import RBFlowDataset

# Initialize dataset
dataset = RBFlowDataset('data/data_12e3.npz')

# Get a sequence of frames
sequence = dataset.get_sequence(start_idx=0, length=10)
vx_sequence = sequence['vx']  # Shape: [10, nx, ny]

Visualization Tools

The repository includes tools for creating various visualizations of the flow fields.

Creating Animations

from visualize import RBFlowVisualizer

# Initialize visualizer
viz = RBFlowVisualizer('data/data_12e3.npz')

# Create velocity field animation
viz.create_velocity_animation(
    output_path='velocity_animation.gif',
    fps=30,
    skip=3  # Arrow density (smaller = denser)
)

# Create temperature field animation
viz.create_animation('temp', 'temperature_animation.gif', fps=30)

Testing

The repository includes a comprehensive test suite (test_functionality.py) that verifies all functionality:

python test_functionality.py

The test suite checks:

1. Dataset loading and access

   • Basic loading functionality
   • Frame access
   • Sequence extraction
   • Data shapes and types

2. Data Processing

   • Normalization
   • Statistics computation
   • Batch processing
   • Sequence bounds checking

3. Visualization

   • Temperature field animations
   • Velocity field animations
   • File generation and saving

Running Individual Tests

import unittest
from test_functionality import TestRBFlowTools

# Run specific test
suite = unittest.TestLoader().loadTestsFromName('test_dataset_loading_12k', TestRBFlowTools)
unittest.TextTestRunner(verbosity=2).run(suite)

Citation

If you use this dataset or code in your research, please cite:

@article{rahman2024pretraining,
  title={Pretraining Codomain Attention Neural Operators for Solving Multiphysics PDEs},
  author={Rahman, Md Ashiqur and George, Robert Joseph and Elleithy, Mogab and Leibovici, Daniel and Li, Zongyi and Bonev, Boris and White, Colin and Berner, Julius and Yeh, Raymond A and Kossaifi, Jean and Azizzadenesheli, Kamyar and Anandkumar, Anima},
  journal={Advances in Neural Information Processing Systems},
  volume={37}
  year={2024}
}