ComplexNDN: Complex Neural Dynamics Models

This repository includes the code for the paper Parallelizable Complex Neural Dynamics Models for Temperature Estimation with Hardware Acceleration.

Since PyTorch does not support parallel scanning, this repo is built by JAX in the Linux system. The parallel prefix sum scan algorithm is implemented by the jax.lax.associative_scan. The parallel algorithm reduces the time complexity from O(N) to O(logN) for serial calculations.

Quick start

Configuration

pip install -r requirements.txt

Run

Run the training program from the command line.

python trainer.py --scan

Star History

Cite as

@article{liao5033161parallelizable,
  title={Parallelizable Complex Neural Dynamics Models for Temperature Estimation with Hardware Acceleration},
  author={Liao, Xinyuan and Chen, Shaowei and Zhao, Shuai},
  journal={Available at SSRN 5033161},
  year={2024}
}