A seminar which includes study and reports on current theoretical and experimental problems. Open only to students officially in the physics honors program or with consent of instructor.

This course is reserved for declared Physics majors. Physical Systems by and for AI explores the intersection of physics and artificial intelligence, offering students a unique opportunity to apply cutting-edge AI methods to designing and understanding physical systems and vice versa. As a pilot for a future Discovery-track course under the Berkeley Discovery Initiative, this class covers two emerging and interrelated topics: AI for Physical Design: Learn how AI tools, techniques, and concepts can be used to design, simulate, and understand novel physical systems for photonics, electronics, acoustics, and quantum devices. Explore techniques like inverse design, where deep-learning style optimization routines discover non-intuitive solutions, and neural-network-based surrogate models that accelerate physics simulations. Physical Systems for AI: Discover how physics can drive innovation in AI hardware, from photonic neural networks to neuromorphic systems inspired by the human brain. Gain insights into how physical phenomena beyond conventional digital electronics can be harnessed to build faster, more efficient AI accelerators and self-learning machines. Students will gain a broad overview of the field, learn foundational concepts, and gain hands-on experience in AI tools like PyTorch and physics tools like finite-difference simulations. They will apply their knowledge in collaborative final projects that aim to move beyond replicating existing studies. For these projects, students will develop a research proposal inspired by an existing paper, incorporating their own ideas, preliminary experiments, or simulations. They will present their proposals in class to receive critical feedback from the instructor and peers, and subsequently refine them into a final, concise NSF-style submission.

Requisites

• Physics Majors

Repeat Rules