Joshua Rackers

Computational Multiscale

Author profile picture

Computational Multiscale

(505) 284-2187

Sandia National Laboratories, New Mexico
P.O. Box 5800
Albuquerque, NM 87185-1322


I am interested in predicting how molecules move and interact at the atomic scale. In particular, I research methods for producing quantum-accurate potential energy surfaces for biological molecules. The tools I build for this purpose draw on the fields of chemical physics, quantum chemistry, statistical physics, and machine learning.

Primary Areas of Research:

  • Simulating ion channels with the HIPPO (Hydrogen-like Intermolecular Polarizable POtential) and AMOEBA (Atomic Multipoles Optimized Energetics for Biomolecular Applications) polarizable force fields. Ion channels are proteins that are essential to life. They selectively permit ions to be shuttled across a cell membrane, allowing our hearts to beat and neurons to fire. However, the mechanism of this selectivity is still unknown. Polarizable force field simulations of ion channels may unlock the mystery of ion channel selectivity.
  • Massively parallel polarizable force field simulations on GPUs (Graphical Processing Units) with LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator) and OpenMM. Polarizable force fields like AMOEBA and HIPPO are currently unable to access today’s tremendously powerful supercomputing resources. Interfacing LAMMPS to OpenMM will solve this problem.
  • Using machine learning to predict electron densities. The electron density is the single most important property of a molecule. By combining what we know about the physics of atomic interactions with innovative machine learning techniques, we may be able to obtain the electron density directly. This would not only bypass computationally expensive ab initio calculations, but enable large-scale, quantum-accurate simulations of biomolecular systems.


  • 2006-2010: B.S. The Ohio State University (Physics and Political Science)
  • 2010-2013: Teach for America, Baltimore, MD
  • 2011-2012: Ms.Ed. Johns Hopkins University (Urban Education)
  • 2013-2019: Ph.D. Washington University in St. Louis (Biophysics)
  • 2019-current: Truman Fellow, Sandia National Laboratories


Alex Lee, William Bricker, Joshua Rackers, (2022). Predicting accurate ab initio DNA electron densities with equivariant neural networks Dna28 Document ID: 1595703

Shivesh Pathak, Joshua Rackers, Ignacio Ema Lpez, Rafael Lpez Fernndez, Alex J. Lee, William P. Bricker, Susi Lehtola, (2022). Accurate Hellmann-Feynman forces with optimized atom-centered Gaussian basis sets Computational Materials Science and EngineeringGordon Research ConferenceComparing Theories, Algorithms and Computation Protocols in Materials Sci Document ID: 1574072

Joshua Rackers, (2022). Beyond the Black Box: The potential and problems of equivariant electron densities Swiss Equivariant Learning Workshop Document ID: 1563084

Alex Lee, Joshua Rackers, William Bricker, (2022). Predicting quantum-accurate electron densities for DNA with equivariant neural networks American Chemical Society Meeting Document ID: 1482414

Joshua Rackers, (2021). Combining physics and AI for quantum-accurate simulations of biological molecules University of New Mexico Chemical and Biological Engineering Fall Seminar Document ID: 1392328

Joshua Rackers, Roseane Silva, Zhi Wang, Jay Ponder, (2021). A Polarizable Water Potential Derived from a Model Electron Density Journal of Chemical Theory and Computation Document ID: 1380949

Joshua Rackers, Alex Jemyung Lee, (2021). Machine learning to build quantum-accurate models for biological macromolecules NM Partnership Schools LDRD Poster Session Document ID: 1356114

Joshua Rackers, (2021). Uncharted Territory: Mapping the Quantum World with Machine Learning Bay Area Research SLAM Document ID: 1355899

Joshua Rackers, (2021). What can machine learning teach us about the limits of electron correlation? Non-Covalent Interactions in Large Molecules and Extended Materials Document ID: 1355842

Joshua Rackers, (2021). What can machine learning and the Hellmann-Feynman Theorem teach us about the limits of electron correlation? American Chemical Society Fall Meeting Document ID: 1355024

Joshua Rackers, (2021). What is Machine Learning good for anyway? Science Friday – Washington University in St. Louis Department of Biochemistry and Molecular Biophysics Document ID: 1344191

Joshua Rackers, (2021). What can you do with a polarizable force field? LAMMPS Workshop and Symposium Document ID: 1343393

Susan Rempe, Zhifeng Jing, Joshua Rackers, Lawrence Pratt, Chengwen Liu, Pengyu Ren, (2021). Thermodynamics of ion binding and occupancy in potassium channels Chemical Science Document ID: 1330370

Joshua Rackers, (2021). Combining physics and AI to understand the behavior of molecules Computational Research Leadership Council seminar series Document ID: 1317575

Joshua Rackers, (2020). The Hellmann-Feynman Theorem, Revisited PsiCon 2020 Document ID: 1243480

Joshua Rackers, (2020). Predicting the Behavior of Biomolecules with Quantum Physics UNM Biology Department Seminar Document ID: 1197106

Showing Results. Show More Publications