Careers in numerical computation#

Today#

Research
- National Labs
- Industrial R&D
- Academia
Production
- Cloud and hardware
- Independent Software Vendors (ISV)
- Open source/services companies
- Engineering companies

Department of Energy
- Office of Science (Berkeley, Argonne, Pacific Northwest, …)
- National Nuclear Security Administration (Livermore, Los Alamos, Sandia)
- Office of Renewable Energy (NREL)
National Oceanic and Atmospheric Administration (NOAA)
- Weather and climate, operational numerical weather prediction and research
National Center for Atmospheric Research (NCAR)
- Comprehensive climate focus (includes ocean, sea ice, land ice, land)
National Institute of Standards and Technology
- Fundamental materials (e.g., gas equations of state), metrology
National Aeronautics and Space Administration (NASA)
- Earth research, especially remote sensing (satellite and aerial)
- Fluid and structural mechanics for safe, reliable, efficient flight (includes partnerships with FAA)
- Materials science and engineering (overlapping scope with industry)

HPC Middleware (MPI implementations, resource managers)
Numerical Libraries (PETSc, hypre, Sundials, Trilinos, MFEM)
Open source simulation applications (MOOSE library, LAMMPS, Nek, Albany)
Internal/export controlled simulation (BLAST, FUN3D, Sierra, MCNP, MOOSE apps)
- Nuclear reactor safety, stockpile stewardship, proprietary designs
- Goodyear-Sandia partnership on tire simulation
Community Earth System Model (NCAR), Energy Exascale Earth System Model (E3SM)
Some orgs publicly embrace open source LLNL Open Source

Hardware vendors and cloud providers: CUDA, HIP, SYCL, OpenMP, MPI
- “Science/engineering should use our programming models/run fast on our hardware”
- “Data science” (statistical computing) should also be fast
High-stakes methods development
- Example: High-Lift Prediction Workshop (participants)
Research papers in specific venues are valuable
Windows into industrial research culture
- Who participates in comparison studies, standards organizations, etc.?
- Who attends technical conferences (not trade shows) and holds elected positions in professional societies?

Tenure-track
- Teaching
- External funding is mainly to support students and research staff
- Eternally swimming in hats
Research faculty/research software engineer
- Usually reliant on external grants, often a “center”
- There are some “research software engineering” facility/service groups (e.g., Princeton)
- CU Institutes (CIRES, INSTAAR, LASP, JILA, BioFrontiers, …)

Competitive government funding
- Grant review panel, process to avoid conflicts of interest
- Applicants receive reviews with positive and negative
- Often low funding rate (e.g., 10% of proposals get funded)
- Acadamics can apply to any agency, DOE labs only eligible for DOE funding
Block government funding
- NNSA funding for classified work; mission-oriented work at other agencies
- Internal review process, may be formal (“LDRD”) but likely doesn’t avoid CoI (reviewed by your management chain)
Industry
- Often informal process to influence management decisions
- Depends more on personal trust, likely more dependent on individual managers (can change rapidly with reorg)
- Since no formal process or external review, minimal feedback

Cloud and hardware vendors
- build or contribute to numerical libraries
- use numerical computing to design silicon, optimize fabrication process
- use numerical computing to design and optimize data centers, analyze workloads
Independent Software Vendors
- Geometric modeling/CAD, simulation (ANSYS, Abaqus, etc.)
- Matlab, Mathematica, etc.
Open source/services
- Distribution/integration: Posit (aka. RStudio), Enthought, Continuum, QuanSight
- Training/consulting/support: OpenFOAM
- Various large tech companies have people doing open source numerical library development

Engineering companies
- Growing in-house simulation tools
- Larger companies can afford to make these tools