Hi all,
Today we have our guest Dr. Long giving a talk during our meeting. You can find a reminder on the content below. This will take ~30 minutes and afterwards we will continue with the usual slide roulette. So still send your two slides to haffert@strw.leidenuniv.nlmailto:haffert@strw.leidenuniv.nl and m.darcis@sron.nlmailto:m.darcis@sron.nl.
Looking forward to seeing you there.
Cheers, Michiel
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.
Michiel Darcis is inviting you to a scheduled Zoom meeting. Join Zoom Meeting https://universiteitleiden.zoom.us/j/66303224819?pwd=cqs8QU5ybpM7Wd78g13oUAB...
Meeting ID: 663 0322 4819 Passcode: nsiR4=A7
---
One tap mobile +496971049922,,66303224819#tel:+496971049922,66303224819,,,,*72170106# Germany
---
Dial by your location • +49 69 7104 9922tel:+496971049922 Germany
Meeting ID: 663 0322 4819 Passcode: 72170106
Find your local number: https://universiteitleiden.zoom.us/u/cbieNLXFga