Matt Ho

Matthew Ho

Carnegie Mellon Physics Ph.D. Student

Academic Curriculum Vitae

About Me

I'm a fifth-year CMU Physics PhD candidate working at the McWilliams Center for Cosmology. I'm interested in applying various methods of statistics and machine learning to advance studies in computational and observational cosmology. In addition to my thesis work, I am an active member in the LSST Dark Energy Science Collaboration, where I develop analysis pipelines for cluster mass measurements and weak lensing maps. I also organize the CMU Physics Industry Speaker Series and enjoy cycling, soccer, and playing chess. My expected graduation date is May 2022.

For comprehensive information about my publications and academic background, check out my curriculum vitae. For a more up-to-date, but less polished, list of my current projects, check out my Github.

Research Highlights


The Dynamical Mass of the Coma Cluster from Deep Learning (2021, Submitted to Nature Astronomy)

In this analysis, we apply our neural network methodology to recover a dynamical mass estimate of the famous Coma cluster. We utilize our approximate Bayesian uncertainty methods to predict Coma's mass to be 1014.92 ± 0.12 h-1M within 1.55 ± 0.02 h-1 Mpc of its center. We find that these predictions are statistically consistent with historical estimates of Coma's virial mass, with strong consistency among estimates of the past two decades.


Approximate Bayesian Uncertainties on Deep Learning Dynamical Mass Estimates of Galaxy Clusters (2021 ApJ, 908, 204)

We study methods for reconstructing Bayesian uncertainties on dynamical mass estimates of galaxy clusters using Convolutional Neural Networks (CNNs). We show that Bayesian CNNs produce highly accurate dynamical cluster mass posteriors. These model posteriors are log-normal in cluster mass and recover 68% and 90% confidence intervals to within 1% of their measured value. We note how this rigorous modelling of dynamical mass posteriors is necessary for using cluster abundance measurements to constrain cosmological parameters.


A Robust and Efficient Deep Learning Method for Dynamical Mass Measurements of Galaxy Clusters (2019 ApJ, 887, 1)

We demonstrate the ability of Convolutional Neural Networks (CNNs) to mitigate systematics in the virial scaling relation and produce dynamical mass estimates of galaxy clusters with remarkably low bias and scatter. We present two models, CNN1D and CNN2D, which leverage this deep learning tool to infer cluster masses from distributions of member galaxy dynamics.


Materials Search

Materials Search is a web-scraping and data-mining tool to aid researchers in finding new candidates for superconductivity. The tool searches crystal databases and paper records for information regarding the properties of possible crystal configurations. It processes this information using statistical analysis to provide useful data at a glance.

Materials Search is useful for consolidating information in order to draw inferences on a particular material’s magnetic and electronic properties.

Course Notes

I've written and published detailed notes for each of my core physics courses throughout my graduate studies. The full list can be viewed here.