Speaker: Benjamin Pope
Affiliation: University of Queensland

Abstract

In order to directly image exoplanets, it is necessary to have both very well calibrated optical systems to suppress starlight (for example interferometers or coronagraphs), and equally sophisticated data analysis software to tease tiny planetary signals out of noise. The main limitation is optical aberrations, whether the distortion of telescope optics or the turbulence of the atmosphere, which produce speckles that can be difficult to distinguish from real planets. The technology underlying deep learning - automatic differentiation or 'autodiff' - offers a solution to both hardware and software design. Autodiff allows us to take derivatives by the chain rule of arbitrary numerical simulations, for example taking the gradient of any figure of merit with respect to the phase of incoming light. This can be used for phase retrieval in highly nonlinear regimes. We can optimize phase masks and other optical designs directly with respect to objective functions like astrometric precision or light suppression. By doing perturbation theory we can also see how noise propagates and construct linear self-calibrations like generalized closure phases. By rewriting the popular optical simulation package 'poppy' to do derivatives, we present 'morphine', a powerful new open source tool for optics and data analysis. While we have developed this for astronomy, the same ideas are applicable to many technologies involving spatial light modulators, phase masks, and related devices.

Slides for this talk are available at: https://benjaminpope.github.io/talks/uqphysics/uqphysics.html#/

About Physics Seminars

The weekly Physics Seminar series focuses on a broad range of physics research within SMP, along with frequent presentations from visiting researchers. Seminars are usually scheduled for 1.00pm on Tuesdays.

The talks are typically more specialised than a colloquium but are often attended by staff and PhD students across a broad range of areas. Speakers are thus encouraged to include introductory material in the talks.

All SMP researchers and HDR students are encouraged to speak. Please email Lewis Williamson to register your interest.

The seminars are open so there is no need to register your attendance.

Venue

Zoom address: https://uqz.zoom.us/j/94116861984