Hi all,

This week Dr. Joseph Long will be visiting Leiden Observatory to work with us on the development of new High-Contrast Imaging post-processing techniques. I am planning to take him out to dinner on Tuesday evening (tomorrow). Please let me know if you want to join us by the end of today. Then I can reserve a table with enough places.

Also, Dr. Long will also give a talk at the instrumentation group meeting this week. Title and abstract are below.

Title: More data than you want, fewer data than you need: machine learning approaches to starlight subtraction with MagAO-X

Abstract: High-contrast imaging relies on post-processing to remove residual starlight from the host star to reveal planets and disks. Most observers do this with principal components analysis (i.e. KLIP) using modes computed from the science images themselves. These modes may not be orthogonal to planet and disk signals, leading to over-subtraction. The wavefront sensor data recorded during the observation provide an independent signal with which to predict the instrument point-spread function. MagAO-X is an extreme adaptive optics system for the 6.5-meter Magellan Clay telescope. MagAO-X is designed to save all sensor information, including millions of kHz-speed wavefront measurements per observation. My approach combines a GPU-accelerated physical optical model and a machine learning model within the same automatic differentiation framework (specifically JAX by Google) to "learn" a mapping from wavefront sensor frames to the inputs of a physical optics model. This preserves the interpretability of the model while also enforcing physical constraints. I will present progress so far in constructing and training such a model, as well as some future directions I plan to explore.

Met vriendelijke groet / With kind regards,

Dr. Sebastiaan Haffert

Assistant professor

Leiden Observatory, Leiden University

Einsteinweg 55, 2333 CC, Leiden, The Netherlands