Code

Experience

I mostly write code in Python, occasionally C. In the past few years I have become very familiar with Jax, a Python library for array-oriented numerical computation, with automatic differentiation and JIT compilation to enable high-performance on GPUs. I have substantial experience working with various sampling and optimization methods in high-dimensional parameter space, and I have some machine-learning experience (mostly using Python’s scikit-learn library).

I am adept at working with big data sets and running applications in parallel on large super computers. I have significant experience running cosmological N-body simulations using the GADGET-4 simulation code, as well as generating linear power spectra with CAMB, creating initial conditions with 2nd order Lagrangian Perturbation Theory (2LPTic), and identifying halos with ROCKSTAR.

I served as the Allocation Manager for the Large Suite of Dark Matter Simulations (LasDamas) Project on XSEDE from 2017-2022.

Here are some open-source codes that I have worked on:

diffsky

I am helping to build a fully differentiable forward modeling pipeline for making predictions of galaxy properties. The diffsky pipeline contains population-level models for predicting galaxy photometry, taking into account dust attenuation, bursty star formation, smooth in-situ star formation, and galaxy merging. The code is written in Jax. The diffsky pipeline brings together several other public codes: dsps, a differentiable stellar population synthesis code; diffstar, a differentiable model for smooth in-situ star formation; and diffmah, a differentiable model for smooth halo mass assembly histories. A publication describing the diffsky pipeline is currently in preparation.

GitHub

Halo Mass Corrections

In Beltz-Mohrmann et al. (2021) I examined the discrepancies in halo masses between dark matter only and hydrodynamic simulations with different baryonic feedback prescriptions. For this work I made use of results from the Illustris, IllustrisTNG, and Eagle simulations. I determined a correction factor as a function of halo mass that could be used to adjust the halo masses in a given dark matter only simulation to resemble the halos in a given hydrodynamic simulation. I developed a Python module for conveniently implementing these halo mass corrections.

GitHub

Contributions

I have also contributed to the following public codes:

• Corrfunc: a suite of codes to calculate correlation functions and other clustering statistics for simulated and observed galaxies
• diffmah: a differentiable model for the mass assembly history of individual and populations of dark matter halos