Los Alamos National Labs with logo 2021

Computational Physics Student Summer Workshop

Sponsored by the Los Alamos National Laboratory Advanced Scientific Computing (ASC) Program

Workshop Leads  

Workshop participants working

Lectures, Teamwork, and Mentoring: Integrated to help you learn, enhance your career, and build connections for the future

2022 Computational Physics Summer Workshop

June 13 - August 19, 2022

Apply Now

Applications are now open for the the Summer 2022 workshop and will be accepted on a rolling basis until January 21, 2022

Los Alamos National Laboratory's X Computational Physics Division, in cooperation with other related divisions including Theoretical Design and Computer, Computational, and Statistical Sciences is pleased to sponsor the 2022 Computational Physics Student Summer Workshop. The workshop seeks to bring to the Laboratory a diverse group of exceptional undergraduate and graduate students from within the United States for informative, enriching lectures and to work with its staff for 10 weeks on interesting and relevant projects that may culminate in articles or conference presentations.

2022 Workshop Flyer

Workshop Overview

Students are organized into teams of 2 working under the guidance of one or more mentors. Each participant is awarded a fellowship that typically ranges from $8,000 to $13,000 based on academic rank (junior, senior, 1st year graduate student, etc.). Lectures, Teamwork, and Mentoring are integrated to help you learn about computational physics and enhance your career. Shared technical goals help you build connections for the future. Generous Fellowships are awarded to support your educational and research efforts while in the summer workshop. Social Events and Tours enhance the team-building experience, creating lasting memories and professional relationships.

Note: The Summer 2022 internship is tentatively scheduled to take place in person. If the current health pandemic causes the workshop to be held remotely, student projects will still be offered along with a series of career-enrichment talks and virtual social hours.

Application Guidelines

The workshop is open to U.S. Citizens who have completed at least one year of college or university. Applications must be submitted online. As part of the application process, you will need to provide a cover letter and resume (your school may have a career office that can help with these), and transcripts (unofficial transcripts are fine), as well as the name and contact information for one person who will submit a letter of recommendation on your behalf.  If you are accepted, you will be required to submit official transcripts. An additional part of the application is to select your top three choices for your research project this summer. A description of the topics are listed below.

Applications are now open for the Summer 2022 workshop and will be accepted on a rolling basis until January 21, 2022.

Click here to apply now.

Fellowship Stipend

Participants will receive a fellowship stipend, the amount to be determined based on your current academic rank.  The stipend will be paid in three installments over the course of the summer.  You will be responsible to cover your own travel, food, and housing.  Housing is in short supply in Los Alamos during the summer, but we will do our best to provide resources to find housing.

Projects

For the 2022 workshop, there will be 12 different projects for students to choose from. Students will work in pairs and will be assigned various mentors that will oversee their project. The mentors are all established scientists at Los Alamos National Laboratory. For more detailed project information, please click on the topic names below.

A New Way to Measure Black Hole Spin Orientations

Mentor: Greg Salvesen (XCP-8)

Description:The anatomy of a black hole X-ray binary system consists of a black hole in a binary orbit with an inflated star, which feeds a disk of gas that emits X-ray light on its journey to the black hole. The system is said to have a spin-orbit misalignment if the black hole spin axis is not aligned with the binary orbital axis. Spin-orbit misalignments have major astrophysical implications, but are difficult to measure because the black hole spin axis is not directly accessible. Fortunately, theory predicts that the rotational axis of the disk aligns with the spin axis of the black hole. Encouragingly, our preliminary analysis of X-ray data from the Swift Observatory suggests that the disk orientation can be measured; thus, probing the black hole spin orientation. This is exciting, but our concern is that different data reduction and modeling choices might affect the results. The two-student team will refine this new technique to make a robust measurement of the disk orientation for a specific system called GRO J1655–40. This will require designing a computational pipeline to perform a systematic Bayesian analysis of X-ray spectra (written in Python and interfacing with XSPEC). If successful, the students will demonstrate an impactful new way to measure black hole spin orientations.

Uncertainty Quantification in High Explosive Products Equations of State Inferred from Experiments

Mentors: Jeff Leiding (T-1), Stephen Andrews (XCP-8), Chris Ticknor (T-1)

Description: Understanding how matter behaves in extreme conditions is of importance to many areas of study, such as earth science, astrophysics, and weapons physics. The behavior of energetic materials is particularly intriguing. In this project, students will learn how to perform research quality hydrodynamics simulations of High Explosives. We have developed UQ tools to infer equations of state from experimental data by performing iterative hydrodynamic simulations. The students will compare equations of state (and their uncertainties) derived from different types of experiments. Opportunities exist in this project to learn about high explosives, molecular physics, Bayesian techniques, uncertainty quantification, and hydrodynamic simulation of high explosives.

Understanding Warm Dense Matter beyond Density Functional Theory

Mentors: Charlie Starrett (XCP-5), Chris Fontes (XCP-5), Jerome Daligault (XCP-5)

Description: Warm dense matter refers to near-solid density plasmas at temperatures that occur in stars, the cores of giant planets, and in inertial fusion plasmas (tens of thousands to millions of Kelvin).  Understanding its physical parameters, like equation of state and optical properties, is key to accurate simulation of these physical systems.  In this project you will implement in computer-code, and apply, advanced models for predicting these properties.  In the past, our student’s work has led to publications in peer-reviewed journals, and we hope the same for this summer.

Modeling Aquarium Experiments: High Explosives Underwater

Mentors: Jessica Thrussell (XTD-PRI), Von Whitley (XTD-SS)

Description: The LANL "aquarium experiments" consisted of a cylinder of high explosive immersed in water or alcohol that was detonated from above. These experiments were diagnosed by an image intensifier camera, in order to measure detonation velocity, confinement effects, and other properties. Students will model PBX 9501 tests using the ALE code FLAG and develop simple analysis tools using PYTHON to compare results to data.

Taking Large-scale Nuclear Data Validation to the Next Level via Machine Learning Methods

Mentors: Denise Neudecker (XCP-5), Jesson Hutchinson (NEN-2), Mike Grosskopf (CCS-6)

Description: Nuclear data are the input for neutron-transport simulations of applications in the areas of, e.g., nuclear energy, criticality safety or astrophysics. Before a nuclear data library is released, these data are validated with respect to integral experiments that resemble, on a small scale, these application spaces to ensure that the library performs well for associated simulations. However, this validation is non-trivial as one integral experiment value is often simulated with 100 to 1000s of nuclear data. Compounding this issue, one validates a whole nuclear data library with more than a thousand integral experiments. Hence, one can be easily faced with the challenge of identifying those nuclear data, out of >10,000, which lead to biases in simulating >1000 integral experiments. The LDRD-DR project EUCLID tackled this high-dimensional problem by applying machine learning techniques. These efforts have uncovered so far clear, but previously unknown, issues in nuclear data, but also processes were developed for the first time to highlight unconstrained physics space in our nuclear data.

The project here will further explore these processes by studying the validation of nuclear data dependent on our choices of input data as well as the algorithm. Out of this study, we will generate a stable toolset that makes this scientific process available to wider use within the nuclear-data community at LANL and internationally. But also, additional open science questions will be explored. For instance, how to correctly include correlated data or how to bring in additional sets of heterogeneous data that are more uncertain. The students would learn about nuclear-data validation, validation experiments, python coding and machine learning algorithms. They would also be part of a larger team environment.

Improving DFEM Transport Performance on Cell Based AMR Meshes

Mentors: Pete Maginot (XCP-2), Andrew Till (CCS-2)

Description: Solving the thermal radiative transfer (TRT) equations spatially discretized with discontinuous finite elements (DFEM) on 2D, cell-wise, adaptively refined meshes is an expensive component of many LANL multiphysics simulations.  The most computationally intense component of solving DFEM TRT is referred to as a transport sweep, a direct solve for the full (angle, space, and energy dependent) photon intensity.  We wish to develop a set of tools that will allow for rapid analysis and optimization of this transport sweep.  To do this, we propose developing 1) a cell-based AMR hydrodynamics mesh emulator and 2) a transport proxy-app that solves the multi-group discrete ordinates equations using the piecewise-linear DFEM spatial discretization.  Work will be tailored to students’ interest, background, and abilities; examples of relevant topics include introductory C/C++ programming, on-node performance abstraction, numerical integration, and deterministic thermal radiation transport.

Center-of-Mass Multigroup Scattering in Monte Carlo Radiation Transport Codes

Mentors: Timothy Burke (XCP-3), Nathan Gibson (XCP-5)

Description: This project will investigate more accurate Multi Group (MG) scattering distributions for use in Monte Carlo (MC) codes that better preserve the physics of scattering off of light nuclei.  This involves generating MG scatter data in the Center-Of-Mass (COM) frame for use in MC codes rather than the lab frame and comparing to deterministic and MC simulations that use equivalent MG data in the lab frame.  Deterministic codes use MG scattering data generated in the lab frame but MC codes can perform scattering kinematics in the COM frame.  Data generated in the COM frame may be better represented via Legendre polynomial series expansions and may better represent the physics of neutrons scattering off of light elements like hydrogen.  Students will implement methods in a MG MC code that runs on GPUs and compare results with MCNP and Partisn.  If students finish this task early they will then investigate a maximum-entropy approach to generating moment-preserving methods for representing MG scatter distributions in MC codes that better represents the physics of neutrons scattering off of light nuclei.

Fully Kinetic Modeling of Plasma Flux Tube Transport

Mentors: Ari Le (XCP-6), Adam Stanier (T-5)

Description: In a curved magnetic geometry, a flux tube with high plasma pressure will rapidly drift. This process is related to edge transport in tokamak fusion devices, high-altitude nuclear explosion phenomena, and plasma transport after Comet Shoemaker-Levy 9 collided with Jupiter. Mass-loaded flux tube transport will be studied with the cylindrical coordinate version of LANL's fully kinetic particle-in-cell code VPIC.

High-order matrix-free three-dimensional Lagrangian hydrodynamics method

Mentors: Nathaniel Morgan (XCP-4), Svetlana Tokareva (T-5)

Description: For multiphysics problems involving large deformations, strong shocks and interactions of multiple materials, solving PDEs in a moving reference frame is one of the common approaches. In Lagrangian methods, the mesh moves with the fluid velocity which leads to accurate resolution of material interfaces. However, the Lagrangian mesh tends to tangle for strong deformations and vortical flows. High-order methods offer additional degrees of freedom per mesh element and allow us to achieve the required level of accuracy using much coarser meshes than low-order methods. A big drawback of the existing high-order finite element methods is the necessity to solve a linear system with a large sparse mass matrix at each time step. The goal of this project is to develop a novel matrix-free and high-order 3D Lagrangian Residual Distribution hydrodynamic (RDH) method that (1) can be used to simulate engineering problems, hydrodynamic experiments, and multi-physics problems with mesoscale material models and curvilinear meshes; and (2) scales well on modern parallel machines including Crossroads that uses CPUs and Sierra that relies on CPUs with GPUs. We will develop the high-order RDH method using curvilinear hexahedral cells and data structures in the ELEMENTS library that was created in the next generation code project (Ristra). Hexahedral cells readily support high-order Bernstein polynomials, which are positivity preserving and a key ingredient in the RDH framework. The work will leverage the matrix and array (MATAR) library within ELEMENTS for performance portability across all computer architectures (e.g., CPUs and GPUs). MATAR extends the Kokkos library to support the foundational data structures used in many multi-material multi-physics codes. Using ELEMENTS will accelerate the development timeline for the project.

Computing detonation waves in the blink of an eye

Mentors: Tariq Aslam (T-1), David Culp (XCP-2), Eduardo Lozano (T-1)

Description: The engineering design of high-explosive components requires fast and efficient computational methods. A new modeling framework is being developed to simulate the propagation of detonation waves in complex 3D geometries but only single-core calculations are currently possible. Working closely with the modeling team, the selected candidates will (1) develop the necessary tools to parallelize the numerical algorithms using LANL’s HPC machines, and (2) conduct the code verification and model validation against high-explosive test cases. The outcome of this project will be a fast wave propagation code capable of simulating 3D detonation waves on the fly.

Modeling and Simulation of Oceanic Turbulence

Mentors: Luke van Roekel (T-3), Filipe Pereira (T-3), Daniel Israel (XCP-4)

Description: The numerical simulation of oceanic turbulence is crucial for numerous engineering applications, such as global warming studies, coastal erosion, or national security problems. However, predicting these complex flows is challenging, and most available turbulence models are either excessively expensive or inaccurate. These limitations motivated our team to derive a novel scale-aware formulation to efficiently (cost vs. accuracy) predict turbulent oceanic flows. The model is currently being validated against reference results. We are interested in evaluating the impact of several parameters of the governing equations of the closure on the accuracy and robustness of the model. Opportunities exist in this project to learn about turbulence modeling, oceans, numerical simulation, verification, validation, and numerical uncertainty.

Uncertainty quantification of high explosives for binder and formulation differences

Mentors: Beth Lindquist (T-1), Ryan Jadrich (T-1)

Description: When using experimental and/or simulation data to develop a computational model, it is often the case that many models exist that are reasonably consistent with a given dataset. Furthermore, some parameters of the model might be tightly constrained by the data, while the data might not inform other parameters at all. Uncertainty quantification (UQ) provides an avenue to interrogate these issues and allows for the creation of an ensemble of models that are consistent with a given dataset. The goal of this project is to perform a UQ analysis for several high-explosives (HEs): PBX 9501, PBX 9404 and X-0298. PBX 9501 and PBX 9404 are well-established HEs with different behaviors (with the latter being much more sensitive) and the behavior of the newer X-0298 is intermediate between the two. We will seek to quantify the differences in the equation of state (EOS) ensembles in a statistically rigorous fashion. While there could be several avenues for the students to contribute to the project, one possibility would be to assist in performing the UQ for the relevant burn models for the EOSs. The exact project will be tailored to the students’ background and interests; in general, students can expect to develop skills in performing simulations, machine learning, statistical inference, and Python coding.