Memristor Crossbar Array Solver

Code to produce and visualize the results of 'Memristor Crossbar Array Simulation for Deep Learning Applications' (DOI: 10.1109/TNANO.2024.3415382).

We also include some supplementary material, with an example and additional mathematics about the solver.

Prerequisites

1. Clone the repo & move into the new directory

git clone https://github.com/Wireless-Information-Networking/mca_solver.git
cd mca_solver

2. Create virtual environment & activate it

python -m venv mca_solver
source ./mca_solver/bin/activate

3. Install dependencies

pip install -r requirements.txt

Only torch, numpy, and scipy are needed to reproduce the results. The latex package needs texlive to run.

Usage

The results from the paper can be replicated by changing the global variable TESTING to False and running:

./experiment.sh

License

Distributed under the GPLv3 License. See COPYING for more information.

Contact

Elvis Diaz Machado - elvis.diaz@uab.cat

Project Link: Wireless-Information-Networking/mca_solver

Acknowledgments

• PhD Supervisor - Jose Lopez Vicario
• PhD Supervisor - Antoni Morell Perez