Computer Science Speaking Skills Talk
— 3:00pm
Location:
In Person
-
Reddy Conference Room, Gates Hillman 4405
Speaker:
ARJUN TEH
,
Ph.D. Student, Computer Science Department, Carnegie Mellon University
https://www.cs.cmu.edu/~ateh/
Computational Lens Design
Optical devices are used everywhere from microscopes to smartphones to telescopes in space. These devices are made of many non-trivial elements that each have unique optical properties, all which help to contribute to a performant lens. Behind each of these designs, there are experts that use a combination of heuristics and brute force search algorithms to find good designs within this high dimensional design space. We help to accelerate this search process by providing a differentiable simulator that can optimize for design objectives efficiently. By leveraging both geometric ray theory and modern inverse rendering techniques, we present a differentiable ray tracer that can efficiently calculate gradients with respect to lens shape and material. In this talk, I will cover the physical model and theory behind our differentiable ray tracer, and how we optimize for different lens types such as gradient index lenses and mirrors. I will also present our current work on using Markov chain Monte Carlo methods to assist with exploration of the design space of lenses. Presented in Partial Fulfillment of the CSD Speaking Skills Requirement.
Event Website:
https://csd.cmu.edu/calendar/speaking-skills-TEH-414