Benjamin Moyer

University of Maryland · Department of Geology

Benjamin
Moyer

Computational geophysicist. I work on Bayesian inverse problems, uncertainty quantification, and the scientific software that makes them run on large machines.

P0 s180 s

II.KAPI · Ende, Indonesia, Mw 7.8. Recovering the size of a source from a single station like this — and reporting honestly how uncertain that estimate is — is the subject of current work.

Research

My dissertation, The Delicate Balance of Error, is about which parts of a geophysical model the data actually constrain and which parts are wishful thinking. Most of it is trans-dimensional Bayesian inversion of gravity and magnetic data, where the number of parameters is itself unknown and has to be inferred alongside everything else.

That work runs on GPUs from a single source tree. The kernels are written once in Julia and execute unchanged on NVIDIA, AMD, and Intel hardware — billion-parameter inference on one consumer card, verified to produce identical results across all three vendors.

Separately, at Lawrence Livermore National Laboratory, I build machine learning pipelines for seismic source characterization: estimating how large an earthquake or explosion was, and how confident that estimate deserves to be, from a single seismic station.

Recent notes

Selected presentations

  • 2026

    Gigavoxels and How to Afford Them: Verified-Forward Trans-Dimensional MCMC Gravity & Magnetics on One Cheap GPU

    DOE CSGF Annual Program Review
  • 2026

    On the Feasibility of Single-Station Scalar Moment Estimation for Earthquake and Non-Earthquake Sources

    SSA Annual Meeting, Pasadena
  • 2025

    From Waves to Yields: AI-Powered Insights into Explosion Source Parameters

    AGU Fall Meeting, S43A-06
  • 2024

    The Long or the Short of It: Optimizing Chain Length and Number in a Probabilistic Joint Inversion Framework

    AGU Fall Meeting, NS21A-03
  • 2023

    Bayesian Inversion of Magnetic Data with Improvements from Constrained Low-Rank Approximation and Quasi-Random Sampling

    AGU Fall Meeting, GP33D
  • 2023

    Joint Inversion of Gravity and Magnetic Data for Lava Tube Detection and Characterization

    NASA Exploration Science Forum

Background

  • 2022–

    Ph.D. in Geology, University of Maryland

    Advisors: Vedran Lekić and Nicholas C. Schmerr
  • 2023–

    DOE Computational Science Graduate Fellowship

    Practicums at Lawrence Livermore National Laboratory
  • 2019–21

    M.Sc. in Geology, Wayne State University

    Trans-dimensional Bayesian inversion of gravitational and structural data
  • 2014–19

    B.A. in Physics, Oakland University

    Minor in political science