Peter Harrington

Biographical Sketch

Peter is a Machine Learning Engineer in the Data and Analytics Services group at NERSC. There, he works with domain scientists on ML applications in a variety of scientific fields. He also helps maintain and benchmark the ML software stack on NERSC's supercomputers, assists users with their ML workflows, and engages the scientific ML community with training events. He has a background in physics and computing, with a BS in Astrophysics and a MS in Scientific Computing & Applied Mathematics from the University of California Santa Cruz. Previously, Peter worked at Berkeley Lab as an applied machine learning researcher in the Computational Research Department.

Recent Papers

Most recent list on Google Scholar profile.

• Subramanian, Shashank, et al. "Towards foundation models for scientific machine learning: Characterizing scaling and transfer behavior." Advances in Neural Information Processing Systems 36 (2024). 
• Pathak, Jaideep, et al. "Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators."  arXiv:2202.11214 (2022).
• Hayat, M., Stein, G., Harrington, P., Lukić, Z., Mustafa, M. "Self-supervised Representation Learning for Astronomical Images", The Astrophysical Journal Letters, 2021. doi:10.3847/2041-8213/abf2c7
• Harrington, P., Mustafa, M., Dornfest, M., Horowitz, B., Lukić, Z. "Fast, high-fidelity Lyman α forests with convolutional neural networks", arxiv preprint, 2021. arxiv:2106.12662
• Horowitz, B., Dornfest, M., Lukić, Z., Harrington, P. "HyPhy: Deep Generative Conditional Posterior Mapping of Hydrodynamical Physics", arxiv preprint, 2021. arxiv:2106.12675