MKI BBL 10/19/2020 — 3 Talks (speakers: Jay Wadekar, NYU; Gus Beane, CfA; Joshua Yao-Yu Lin, University Of Illinois At Urbana-Champaign)
Generating neutral hydrogen from dark matter with neural networks
Abstract: Hydrodynamic simulations have a huge computational cost (~10 million CPU hours for 0.001 Gpc^3 volume) and cannot therefore be directly used in predictions for upcoming surveys probing ~100 Gpc^3 volumes. Focusing on neutral hydrogen (HI), I will show that neural networks can be trained on hydro simulations to quickly generate accurate HI maps from gravity-only dark matter simulations. I will also show that the environment of a dark matter halo has a crucial effect on its HI mass; I will present a novel symbolic expression which encodes this environmental effect and was obtained using symbolic regression.
Bio: I am a final year PhD student at New York University. My work has been on observational cosmology using galaxy surveys and DM phenomenology but recently I’ve been working on neural networks and hydrodynamic simulations with the group at CCA.
Webpage: https://jaywadekar.github.io/research.html
Email: jay.wadekar@nyu.edu
Host: Vogelsberger group
*************************************************
Dynamically Dating StarsAbstract: In the conventional Galactic picture, stars are thought to be born on circular and planar orbits. They then evolve into more non-circular and non-planar orbits through interactions with various Galactic components, such as the bar, spiral arms, and giant molecular clouds. Using two samples of stars with well-determined ages through different methods, we explore whether the type of orbit a star is on can be used to accurately determine the age of that star. We discuss a novel application of dynamical age dating to the difficult problem of calibrating gyrochronology, a particularly attractive prospect in light of the rotation periods to be measured with TESS.
Bio: Gus is a second year graduate student at the CfA working with Lars Hernquist on hydrodynamical simulations of the Milky Way. He is discussing work from his undergraduate studies at the University of Pennsylvania and as an intern at the Flatiron Institute.
Email: angus.beane@cfa.harvard.edu
Host: Vogelsberger Group
*************************************************
Feature Extraction on Synthetic Black Hole Images
Abstract: The Event Horizon Telescope (EHT) recently released the first horizon-scale images of the black hole in M87. Combined with other astronomical data, these images constrain the mass and spin of the hole as well as the accretion rate and magnetic flux trapped on the hole. An important question for EHT is how well key parameters such as spin and trapped magnetic flux can be extracted from present and future EHT data alone. Here we explore parameter extraction using a neural network trained on high resolution synthetic images drawn from state-of-the-art simulations. We find that the neural network is able to recover spin and flux with high accuracy. We are particularly interested in interpreting the neural network output and understanding which features are used to identify, e.g., black hole spin. Using feature maps, we find that the network keys on low surface brightness features in particular.
Link to the paper: https://arxiv.org/abs/2007.00794
Bio: Joshua is a fifth-year physics Ph.D. student at the University of Illinois at Urbana-Champaign. He is interested in Machine Learning applications in astrophysics. He mainly works on dark matter substructures in strong lensing with Gil Holder and supermassive black holes with Charles Gammie. He is also interested in machine learning applications in Radio Interferometry.
Email: joshualin24@gmail.com
Host: Hsin-Yu Chen
Speakers
- Jay Wadekar, NYU
- Gus Beane, Harvard CfA
- Joshua Yao-Yu Lin, University of Illinois at Urbana-Champaign