A machine-learning view of our Milky Way

Event Date:
2018-11-05T15:00:00
2018-11-05T16:00:00
Event Location:
Hennings 318
Speaker:
Nina Hernitschek (Caltech)
Related Upcoming Events:
Intended Audience:
Undergraduate
Event Information:

Every night, telescopes around the world obtain a flood of new data as parts of deep and wide surveys. This amount of data will steeply rise once upcoming sureys such as the Large Synoptic Survey Telescope (LSST), which will image the entire visible sky every few nights, start their operation. To investigate this huge amount of data, machine-learning algorithms are absolutely necessary for image analysis, classification of sources, time-series analysis and also structure finding.

Using the example of the Pan-STARRS1 3pi survey as a deep panoptic high-latitude survey in the time domain, in this talk it will be shown how to explore the capabilities of such surveys for carrying out time-domain science in a variety of applications. We use structure-function fitting, period fitting and subsequent machine-learning classification to search for and classify high-latitude as well as low-latitude variable sources, in particular RR Lyraes, Cepheids and QSOs. Our further analysis reveals the extent of the Sagittarius stream as well as other overdensities in the Milky Way's outer halo. In addition, it will be shown how carefully chosen follow-up observations can support large surveys, such as Pan-STARRS1 3pi, and our goal of investigating the structure of the Milky Way's halo.

Please join us before the Colloquium in Hennings 318 for coffee, tea and snacks at 2:45 pm

Add to Calendar 2018-11-05T15:00:00 2018-11-05T16:00:00 A machine-learning view of our Milky Way Event Information: Every night, telescopes around the world obtain a flood of new data as parts of deep and wide surveys. This amount of data will steeply rise once upcoming sureys such as the Large Synoptic Survey Telescope (LSST), which will image the entire visible sky every few nights, start their operation. To investigate this huge amount of data, machine-learning algorithms are absolutely necessary for image analysis, classification of sources, time-series analysis and also structure finding. Using the example of the Pan-STARRS1 3pi survey as a deep panoptic high-latitude survey in the time domain, in this talk it will be shown how to explore the capabilities of such surveys for carrying out time-domain science in a variety of applications. We use structure-function fitting, period fitting and subsequent machine-learning classification to search for and classify high-latitude as well as low-latitude variable sources, in particular RR Lyraes, Cepheids and QSOs. Our further analysis reveals the extent of the Sagittarius stream as well as other overdensities in the Milky Way's outer halo. In addition, it will be shown how carefully chosen follow-up observations can support large surveys, such as Pan-STARRS1 3pi, and our goal of investigating the structure of the Milky Way's halo. Please join us before the Colloquium in Hennings 318 for coffee, tea and snacks at 2:45 pm Event Location: Hennings 318