12-787 Special Topics: Physics-Informed Machine Learning

12 units, Carnegie Mellon University

Many real-world engineering and Earth-science problems are challenging to model due to limited data, complex physical processes, computationally expensive numerical simulations, and significant sources of uncertainty. Physics-Informed Machine Learning provides a powerful way to address these challenges by combining machine learning with physical principles to create more efficient and physics consistent models. This course introduces key methodologies, including physics-informed neural networks, operator learning, generative modeling, knowledge discovery, and inverse modeling, and explains how they can enhance traditional modeling and simulation workflows. Students will learn both the theoretical foundations and practical implementation of these methods and will apply them to climate, hydrology, geoscience or other problems through hands-on projects aligned with their research interests.

Prerequisites: none listed

Offered: Spring 2027, Spring 2026

Courseweave needs JavaScript to show sections, workloads and planning tools.