Publications

On the minimum number of radiation field parameters to specify gas cooling and heating functions

Published in OJA, 2025

In this paper, we modify our previous implementation of XGBoost to approximate atomic gas cooling and heating functions. We now use radiation field intensity averaged in various energy bins to describe the incident radiation field (instead of photoionization rates). We use feature importance tools to find the most important energy bins for predicting cooling and heating functions. We find that a sample of only 3 energy bins (or photoionization rates) are sufficient to accurately interpolate atomic gas cooling and heating functions at fixed metallicity.

Recommended citation: Robinson, David, Camille Avestruz, and Nickolay Y. Gnedin. “On the minimum number of radiation field parameters to specify gas cooling and heating functions.” OJA 8 (2025). https://astro.theoj.org/article/141235-on-the-minimum-number-of-radiation-field-parameters-to-specify-gas-cooling-and-heating-functions

Cosmic Reionization on Computers: Statistical Properties of the Distributions of Mean Opacities

Published in OJA (under review), 2025

In this paper, we explore the distribution of Lyman alpha optical depth along simulated quasar sightlines in the Cosmic Reionization on Computers (CROC) simulations. We compare the cumulative distribution function (CDF) of optical depths to observations, subsampling the simulated sightlines appropriately. We quantify the variance in the simulated CDFs due to this subsampling.

Recommended citation: Werre, Ella, David Robinson, Camille Avestruz, and Nickolay Y. Gnedin. "Cosmic Reionization on Computers: Statistical Properties of the Distributions of Mean Opacities". arXiv:2503.10768 (2025). https://arxiv.org/abs/2503.10768

The effects of different cooling and heating function models on a simulated analog of NGC300

Published in OJA (under review), 2024

In this paper, we incorporate our machine learning approximation for gas cooling and heating functions into a hydrodynamic isolated galaxy simulation. We compare the gas thermodynamics across two runs of the simulation: one using our machine learning models to compute the atomic gas cooling and heating functions, and one using an interpolation table. We find that our machine learning model predicts systematically hotter low-density gas.

Recommended citation: Robinson, David, Camille Avestruz, Nickolay Y. Gnedin, and Vadim A. Semenov. “The effects of different cooling and heating function models on a simulated analog of NGC300.” arXiv:2412.15324 (2024). https://arxiv.org/abs/2412.15324

Emergence of the temperature-density relation in the low-density intergalactic medium

Published in MNRAS, 2024

In this paper, we explore the evolution of the phase diagram of low-density gas in the intergalactic medium in four different simulated boxes from the Cosmic Reionization on Computers (CROC) project. We show that the evolution of the fraction of cold gas is approximately universal across the four different reionization histories, when parameterized by the hydrogen neutral fraction. We also explore the emergence of a tight temperature-density relationship through the decreasing width of the scatter around a power-law relationship. This decrease in width is also a nearly universal function of the neutral fraction.

Recommended citation: Wells, Alexandra, David Robinson, Camille Avestruz, and Nickolay Y Gnedin. "Emergence of the temperature–density relation in the low-density intergalactic medium." MNRAS 528 4 (2024). https://academic.oup.com/mnras/article/528/4/5845/7602414

Exploring the dependence of gas cooling and heating functions on the incident radiation field with Machine Learning

Published in MNRAS, 2024

In this paper, we use the machine learning algorithm XGBoost to approximate cooling and heating functions with a general incident radiation field calculated with CLOUDY. We are able to reduce the frequency of large prediction errors compared to interpolation table approaches. We also use feature importance techniques to explore what aspects of the incident radiation field most strongly affect the cooling and heating functions.

Recommended citation: Robinson, David, Camille Avestruz, and Nickolay Y. Gnedin. “Exploring the Dependence of Gas Cooling and Heating Functions on the Incident Radiation Field with Machine Learning.” MNRAS 528 1 (2024). https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/stad3880/7478000

Can Cooling and Heating Functions Be Modeled with Homogeneous Radiation Fields?

Published in ApJ, 2022

We explore the cooling and heating functions of Epoch of Reionization galaxies from the Cosmic Reionization on Computers (CROC) project. We find that the actual cooling and heating rates of CROC galaxies cannot be adequately described by assuming a spatially homogeneous radiation field.

Recommended citation: Robinson, David, Camille Avestruz, and Nickolay Y. Gnedin. “Can Cooling and Heating Functions be Modeled with Homogeneous Radiation Fields?” ApJ 936 50 (2022). https://iopscience.iop.org/article/10.3847/1538-4357/ac85e1

Probing spatial variation of the fine-structure constant using the CMB

Published in Phys. Rev. D., 2019

This paper includes results from a summer research project that Davy Qi and I worked on with Tristan Smith at Swarthmore College, where we examined the effects of spatial variation of the fine-structure constant on the CMB.

Recommended citation: Smith, Tristan, Dan Grin, David Robinson, and Davy Qi. “Probing Spatial Variation of the Fine-Structure Constant Using the CMB.” Physical Review D 99.4 (2019). https://journals.aps.org/prd/abstract/10.1103/PhysRevD.99.043531