Lava Multi-Agent Optimization (LMAO)

Before installing LMAO, you must install the base Lava library. If you would like to run the satellite scheduling problem, you must also install Lava-Optimization.

With Lava installed, LMAO can be installed as an editable Python package with the following command:

python -m pip install .

There is a known issue with some versions of Scikit-Optimize where newer NumPy versions do not support np.int data structures. If you get this error, please CTRL+F within the library and replace all instances of np.int with np.int32.

---

Sample Application

We have included the satellite scheduling problem from Lava-Optimization as a baseline application here. The notebook has all of the necessary infrastructure to highlight how you can wrap your problem in a function wrapper and connect it with LMAO.

If you run into any issues, please reach out to the maintainer.

---

Authors

• Shay Snyder: [email protected] (maintainer) (1)
• Derek Gobin (1)
• Victoria Clerico (1)
• Sumedh R. Risbud (2)
• Maryam Parsa (1)

(1) George Mason University (2) Intel Labs

---

Acknowledgements

This work was supported by a generous gift the Intel corporation.

---

Citation

If you find LMAO useful in your work, we would appreciate a citation :)

@misc{snyder2024parallelizedmultiagentbayesianoptimization,
      title={Parallelized Multi-Agent Bayesian Optimization in Lava}, 
      author={Shay Snyder and Derek Gobin and Victoria Clerico and Sumedh R. Risbud and Maryam Parsa},
      year={2024},
      eprint={2405.04387},
      archivePrefix={arXiv},
      primaryClass={cs.DC},
      url={https://arxiv.org/abs/2405.04387}, 
}