Research

Searching for White Dwarf Main Sequence (WDMS) Binary Systems with Neural Networks

• Search through millions of low resolution stellar spectra from Gaia Data Release 3
• ~700 confirmed WDMS systems with both high and low resolution spectra as training sample
• Data augmentation with synthetic training data
• Developing a two-headed neural network to classify WDMS systems and fit stellar parameters
• Interpreting the decision making process of the trained model through visualization methods

Accelerating Spin Dynamics Numerical Simulation with Physics-Aware Neural Network

• Accelerating quantum mechanics simulation with neural network
• Developing a neural network for inference of force that would preserve SO(3) rotational symmetry in the underlying physics system
• Developing a physics-aware neural network that could replace the traditional FEM numerical solver for the spin dynamics problem

Kinematical Analysis of Substructure in the Large Magellanic Cloud

• Millions of stars from Gaia Early Data Release 3
• Examined the kinematical structures of the Large Magellanic Cloud
• Collaboration with research group from NMSU, STScI and NOAO
• Paper published in ApJ

Density Map of the Milky Way Galaxy

• Millions of stars from Gaia Data Release 3
• Computed intrinsic velocity dispersion by removing the contribution from uncertainty of individual stars
• Combined Jeans Equation and Poisson Equation to measure the surface density of any given point in our Galaxy
• Compared to models of visible matter to extract the distribution of dark matter

Exploring asymmetries in the kinematics of the Galactic disk with Gaia and APOGEE

• SDSS-IV Project 0722
• Converted observables to phase-space information
• Compared observation results to existing numerical simulation
• Built a simple warp model with Jeans equation
• Paper published in ApJ

Galactic kinematics with OB stars from LAMOST-Gaia dataset

• Coded adaptive kernel density estimation (KDE) in Python
• Extracted kinematics structure from dataset with various methods
• Determined most possible theoretical explanation for observed ripples in radial velocity
• Paper published in ApJ Letter

Confirmation of strong lensing candidates using CFHT Megacam

• Refined the candidate list from previous research experience during June 2017 - September 2017
• Wrote observation proposal and designed details (exposure time, sequence of observation, etc.) of observation
• Applied and approved for CFHT Megacam observation (18BS06) in September 2018 as the Principle Investigator
• Processed Megacam imaging data (coadding, psf, photometry and foreground removal)
• Paper published in RAA

Searching for galaxy-galaxy strong lensing candidates in SDSS-III BOSS

• Improved previous spectroscopic searching method
• Data-processing code exceeds 10 thousand lines of Python
• Searched through the entire database (~1.5 million galaxies) within 12 hours
• Compiled a list of most possible strong lensing candidates

Searching for super-Eddington accreting black holes candidates in SDSS-III Reverberation Mapping campaign

• Built spectra decomposition programs from scratch in Python
• Extensive literature reading
• Compared Radius-Luminosity relationship with results from other researchers