Postdoctoral Researcher - Operator Networks for Hydrodynamics & ICF (IRC145077)

Los Alamos National Laboratory; Los Alamos (NM); On-site; USD 65,000 - 90,000; Full time

Job summary

Los Alamos National Laboratory invites applications for 1-2 postdoctoral researchers to advance physics-consistent operator networks for hydrodynamics and inertial confinement fusion (ICF). The role spans theory, algorithms, and scalable HPC implementations on cutting-edge hardware.

Responsibilities include designing neural operators, developing mathematical foundations, integrating physics into learning, and running large-scale experiments.

Qualifications

• PhD earned within the last 5 years in a related field (Applied Mathematics, Mathematics, CS, Statistics, Physics, or Engineering).
• Strong background in functional analysis and numerical PDEs; ability to craft rigorous arguments and proofs.
• Experience in neural operators / scientific ML with publications.
• Proficient in Python and ML frameworks (PyTorch/JAX/TensorFlow).
• Excellent written and oral communication demonstrated through publications and letters.

Responsibilities

• Design, analyze, and implement operator neural networks for hydrodynamics and ICF.
• Develop mathematical foundations for operator learning and provide rigorous analyses.
• Integrate physics into learning via variational principles and differentiable solvers.
• Run large-scale computational experiments on CPU/GPU clusters and compare with high-fidelity solvers.
• Collaborate across math, CS, and plasma physics; publish and present findings.

Skills

Mathematical depth

Scientific ML

Scientific computing

Communication skills

Education

PhD in a relevant field

Tools

Python

PyTorch

JAX

TensorFlow

Job description

The Applied Math and Plasma Physics Group (T-5) in the Theoretical Division at Los Alamos National Laboratory seeks 1-2 outstanding postdoctoral researchers to advance physics-consistent operator networks for hydrodynamics and inertial confinement fusion (ICF) applications. The work will span theory, algorithms, and scalable implementations on cutting-edge HPC systems. As a postdoc you will:

• Design, analyze, and implement operator neural networks (e.g., DeepONets, Fourier/Neural Operators, graph/mesh operators) that respect conservation laws, symmetries, and stability constraints relevant to compressible hydro, radiation-hydrodynamics, and MHD.


• Develop mathematical foundations for operator learning: approximation/error bounds, well-posedness, stability, and generalization for PDE-to-PDE maps; contribute proofs and rigorous analyses.


• Integrate physics into learning (e.g., variational/energy principles, constrained training, differentiable solvers, PINO/PINN-style residuals) and build robust surrogates for stiff, multi-scale flows.


• Run large-scale computational experiments on CPU/GPU clusters; compare against high-fidelity solvers and community benchmarks.


• Collaborate across math, CS, and plasma physics, publish in leading journals, present at conferences, and engage with external research partners.

Minimum Job Requirements

• Education: Ph.D. (earned within the last 5 years) in Applied Mathematics, Mathematics, or a closely related field (e.g., CS, Statistics, Physics, Engineering).


• Mathematical depth: Demonstrated strength in functional analysis/operator theory, numerical PDEs, approximation theory, and/or stochastic processes; ability to craft rigorous arguments (stability, error estimates, convergence).


• Scientific ML: Research experience in neural operators / scientific machine learning (operator learning, PINNs/PINOs, PDE-constrained learning, model reduction) with a strong publication record.


• Programming: Proficiency in Python and one or more of PyTorch/JAX/TensorFlow; working knowledge of scientific computing (vectorization, profiling, testing).


• Communication: Clear written and oral communication, as shown through publications and the cover letter.

Desired Qualifications

• Experience with hydrodynamics/ICF/HEDP physics (e.g., shocks, turbulence, EOS, radiation transport, MHD) or with high-order/structure-preserving numerical methods.


• HPC skills (MPI, OpenMP, CUDA, GPU acceleration), containers, CI, and collaborative software development.


• Contributions to open-source ML/scientific software; code reviews; reproducible workflows.


• Background in UQ/data assimilation/Bayesian inference, multi-fidelity or active learning for PDE models.


• Evidence of competitive excellence (e.g., research awards, notable competitions) and effective teamwork.

Education/Experience

A Ph.D. in Applied Mathematics, Mathematics, Computer Science, Statistics, Physics, or a closely related field, earned within the last 5 years.