David W Hogg, NYU Physics
Finding planets in very noisy data from stars.
In the last ten years, astronomers have discovered thousands of
planets around other stars. Planets are much fainter, smaller,
and lower mass than their host stars, so the signals they produce
are tiny perturbations on the stellar signals, and much smaller than
all important sources of noise. We use ideas from causal
inference, models of time-correlated noise, methods for fast linear
algebra, and some good software engineering to brute-force search
for planets in data from the NASA Kepler mission. I will show
some newly discovered planets and discuss improvements that might
make for further discoveries.
This work was done in collaboration with Dan Foreman-Mackey (UW)
with help from various people around Courant (Ambikasaran, O'Neil,
Goodman, and Greengard).