CCAPP Seminar by Eric Huff from JPL

Image
ccapp
October 8, 2019
11:30AM - 12:30PM
Location
PRB 4138

Date Range
Add to Calendar 2019-10-08 11:30:00 2019-10-08 12:30:00 CCAPP Seminar by Eric Huff from JPL Eric Huff - JPLThe next decade will see a dramatic increase in the quantity of astronomical surveydata, with the advent of ambitious wide-field projects like WFIRST, Euclid, LSST,SPHEREx, and DESI. These surveys will make dramatic gains in cosmology andastrophysics, representing multiple orders-of-magnitude increases in the volume of universe mapped, and the number of galaxies with photometric and spectroscopicmeasurements. With this great increase in data volume and quantity comes a number of newchallenges. Each improvement in statistical power exposes new sources ofsystematic error, and it is not clear that traditional analysis methods will scale up tothe new data sets. JPL's 'Dark Sector' cosmology group is heavily involved in these efforts, and we areworking to make this future a bit more tractable. I will describe recent progress byDark Sector staff and postdocs on the use of machine learning methods to detect andcharacterize astrophysical anomalies, on emulating future space mission data in thelab, and on the more general problems that arise when systematic biases couple tocosmological signals. PRB 4138 Center for Cosmology and AstroParticle Physics (CCAPP) ccapp@osu.edu America/New_York public
Description

Eric Huff - JPLThe next decade will see a dramatic increase in the quantity of astronomical surveydata, with the advent of ambitious wide-field projects like WFIRST, Euclid, LSST,SPHEREx, and DESI. These surveys will make dramatic gains in cosmology andastrophysics, representing multiple orders-of-magnitude increases in the volume of universe mapped, and the number of galaxies with photometric and spectroscopicmeasurements.

With this great increase in data volume and quantity comes a number of newchallenges. Each improvement in statistical power exposes new sources ofsystematic error, and it is not clear that traditional analysis methods will scale up tothe new data sets.

JPL's 'Dark Sector' cosmology group is heavily involved in these efforts, and we areworking to make this future a bit more tractable. I will describe recent progress byDark Sector staff and postdocs on the use of machine learning methods to detect andcharacterize astrophysical anomalies, on emulating future space mission data in thelab, and on the more general problems that arise when systematic biases couple tocosmological signals.

Events Filters: