Postdoc Colloquium

astronomy logo

MOD_JSE_MEGAMENU_TOGGLE_MENU

Calendar

Postdoc Colloquium

Tuesday, September 17, 2024, 03:30pm - 04:30pm

This repeat is an exception to the normal repeat pattern

Previous Repeat

Next Repeat

Yinan Zhao, The University of Texas at Austin
Deep Learning for Earth-Like Planet Detection in Presence of Stellar Activity

Abstract

The presence of stellar activity in radial velocity (RV) measurements is the current major limitation to detect earth like planets around solar type stars. More methods are moving into the direction of disentangling stellar activity at the spectral level. In this talk, I will introduce an algorithm designed to enhance our comprehension and mitigation of stellar activity: The deep learning-based
algorithm that allows an efficient modelling of stellar activity at the "spectral shell domain" enhancing the detection of Earth-like planets. This deep learning algorithm has been tested on 3 intensively observed stars: Alpha Centauri B, Tau ceti and the Sun. By injecting simulated planetary signals at the spectral level, we demonstrate that our machine learning algorithm can achieve, for Alpha Centauri B, Tau ceti, a detection threshold of 0.5 m/s in semi-amplitude for planets with periods ranging from 10 to 300 days. On the HARPS-N solar dataset, our algorithm is even more efficient at mitigating stellar activity signals and can reach a threshold of 0.2 m/s,
which would correspond to 2.2 earth-mass planet on the orbit of the Earth. It is the first time that such low detection thresholds are reported for the Sun, but also for other stars, and therefore highlight the efficiency of our deep learning-based algorithm at mitigating stellar activity in RV measurements.

Matthew De Furio, The University of Texas at Austin
Identification of a Turnover in the Initial Mass Function of a Young Stellar Cluster Down to 0.5 MJ

Abstract

Star and brown dwarf formation theory predicts a mass limit to the turbulent fragmentation of molecular clouds, thought to be 2-10 MJ, providing a natural turnover in the mass function down to planetary masses. Previous studies in the solar neighborhood and nearby star-forming regions identified an increasing mass function from the hydrogen-burning limit down to about 10 MJ, but were not sensitive to lower mass objects. In this talk, I will present results exploring the free-floating planetary mass object population in NGC 2024, a young (< 1 Myr) embedded star-forming region, and the characterization of the initial mass function down to sub-Jupiter masses using JWST/NIRCam. In GTO-1190, we obtained deep imaging of the central core of NGC 2024, covering 0.7 - 5 μm and identified 48 likely cluster members. We find that a broken power-law model of the mass function derived from atmospheric models best represents the data, increasing from 60 MJ to roughly 12 MJ then followed by a decrease down to 0.5 MJ. This is the first evidence for a decrease in the mass function below 10 MJ in a stellar population to date. Our program probes the lowest masses theoretically produced through star formation processes, finding no likely cluster members below about 3 MJ, potentially the fundamental limit of the star and brown dwarf formation process.

Location: PMA 15.216B and online