Joshua Yao-Yu Lin

Snapshot of the black hole image in M87. (image credit: EHT)

Machine Learning application for the Event Horizon Telescope

Joshua Yao-Yu Lin + George N. Wong, Ben S. Prather, Charles F. Gammie (ICML 2020 workshop)

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 black hole as well as the accretion rate and magnetic flux trapped on the black 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. 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 feature in particular. This work is accepted by ICML 2020 workshop. I am interested in the extension of building neural networks that could directly estimate parameters in radio interferometry data for EHT.

Snapshot of our neural network prediction on the location of dark matter subhalos in strong gravitational lensing simulation.

Hunting for Dark Matter Subhalos in Strong Gravitational Lensing with Neural Networks

Joshua Yao-Yu Lin + Hang Yu, Warren Morningstar, Jian Peng, Gilbert Holder

Dark matter substructures are interesting since they can reveal the properties of dark matter. Therefore, understanding the population and property of subhalos at cosmological scale would be an interesting test for cold dark matter. In recent years, it has become possible to detect individual dark matter subhalos near images of strongly lensed extended background galaxies. In this work, we use modified DenseNet to help us detect dark matter substructures in simulated strong lensing systems, and we show that deep neural networks could detect multiple dark matter subhalos in simulated data. This work is accepted by NeurIPS 2019 workshop. I am interested in the extension of building neural networks that could directly estimate parameters in ALMA’s strong lensing data since ALMA could provide higher resolution data.

AGN light curve

AGNet: Weighing Black Holes with Machine Learning

Joshua Yao-Yu Lin + Sneh Pandya, Devanshi Pratap, Xin Liu, Matias Carrasco Kind (submitted to NeurIPS 2020 workshop)

Supermassive black holes (SMBHs) are ubiquitously found at the centers of most galaxies. Measuring SMBH mass is important for understanding the origin and evolution of SMBHs. However, traditional methods require spectral data which is expensive to gather. In addition, the Vera Rubin Observatory (LSST) project will discover 1000,000,000 new SMBHs across most of the observable universe; it would take 20,000 years to weigh them with traditional methods. Therefore, a much more efficient approach is needed to maximize LSST science would be crucial and necessary to solve this problem, we present an algorithm that weighs SMBHs using time series quasar lightcurve, circumventing the need for expensive spectra. We train, validate, and test neural networks that directly learn from the Sloan Digital Sky Survey (SDSS) Stripe 82 data for a sample of 9038 spectroscopically confirmed quasars to map out the nonlinear encoding between black hole mass and multi-color optical light curves. We find a 1 sigma scatter of 0.35 dex between the predicted mass and the fiducial virial mass based on SDSS single-epoch spectra. Our results have direct implications for efficient applications with future observations from the Vera Rubin Observatory.

captions.

Learning Principle of Least Action with Reinforcement Learning

Zehao Jin* + Joshua Yao-Yu Lin* (*equal contribution), Siao-Fong Li

Nature provides a way to understand physics with reinforcement learning since nature favors the economical way for an object to propagate. In the case of classical mechanics, nature favors the object to move along the path according to the integral of the Lagrangian, called the action S. We consider setting the reward/penalty as a function of S, so the agent could learn the physical trajectory of particles in various kinds of environments with reinforcement learning. In this work, we verified the idea by using a Q-Learning based algorithm on learning how light propagates in materials with different refraction indices, and show that the agent could recover the minimal-time path equivalent to the solution obtained by Snell's law or Fermat's Principle. For future work, we would like to explore the possibility to relate this work with path integral.

Gravitational lensing could split up the neutrinos trajectory

Gravitational Lensing of the Cosmic Neutrino Background

Joshua Yao-Yu Lin, Gilbert Holder

We study gravitational lensing of the cosmic neutrino background. This signal is undetectable for the foreseeable future, but there is a rich trove of information available. At least some of the neutrinos from the early universe will be non-relativistic today, with a closer surface of last scattering (compared to the cosmic microwave background) and with larger angles of deflection. Lensing of massive neutrinos is strongly chromatic: both the amplitude of lensing and the cosmic time at which the potential is traversed depend on neutrino momentum, in principle giving access to our entire causal volume, not restricted to the light cone. As a concrete example, we focus on the case where the cosmic neutrino background would be strongly lensed when passing through halos of galaxy clusters and galaxies. We calculate the Einstein radius for cosmic neutrinos and investigate the impact of neutrino mass.