What Is Physics Computing?

Physics computing, also called computational physics, is the discipline concerned with implementing numerical and computational methods to solve physical problems that are analytically intractable or experimentally prohibitive to measure directly. It occupies the intersection of physics, applied mathematics, and computer science, treating the computer as a scientific instrument for investigating physical systems through simulation, modeling, and data analysis. The field emerged as a formal discipline in the mid-twentieth century alongside the development of electronic computers and has since become a standard component of research across nearly every branch of physics.

Computational physics is distinguished from theoretical physics by its emphasis on numerical approximation rather than exact analytical solutions, and from experimental physics by its reliance on simulation rather than laboratory measurement. In practice, the three modes of inquiry are deeply intertwined: theoretical models guide the construction of algorithms, computational results suggest experimental hypotheses, and experimental data constrain simulation parameters.

Numerical Methods and Algorithms

The core technical content of computational physics consists of numerical methods for solving the mathematical equations that describe physical systems. Ordinary and partial differential equations arise in classical mechanics, electromagnetism, quantum mechanics, and fluid dynamics; methods such as Runge-Kutta integration, finite difference, finite element, and spectral approaches convert continuous equations into discrete systems solvable by computer. Eigenvalue solvers address quantum mechanical problems where the energy levels and wavefunctions of a system must be extracted from a Hamiltonian operator. Monte Carlo methods use random sampling to evaluate high-dimensional integrals that appear in statistical mechanics and quantum field theory. The Journal of Computational Physics, published since 1966, documents advances in these methods across physics domains and remains the primary archival record of the field.

Simulation and Modeling

Simulation is the primary output mode of computational physics. Molecular dynamics (MD) simulations propagate the positions and momenta of many particles under prescribed interatomic potentials, enabling study of equilibrium and nonequilibrium phenomena in condensed matter at length scales inaccessible to experiment. Particle-in-cell (PIC) codes simulate plasmas by tracking representative particles on a discretized grid with self-consistent electromagnetic fields. Lattice Boltzmann methods simulate fluid flow through a statistical mechanics model on a regular grid. In astrophysics and cosmology, N-body codes integrate the gravitational interactions of up to billions of particles to study structure formation in the universe. The computational physics group at Lawrence Livermore National Laboratory, described by LLNL Computing, develops simulation codes for high-energy-density physics, radiation hydrodynamics, and nuclear science that serve both basic research and national security missions.

High-Performance Computing in Physics

Many computational physics problems require computing resources that exceed single workstations by several orders of magnitude. Plasma simulations of thermonuclear fusion reactors, climate-coupled atmosphere-ocean models, and first-principles electronic structure calculations for complex materials routinely run on the largest available supercomputers. Modern physics codes are written to exploit parallelism through MPI message-passing across compute nodes and OpenMP thread-level parallelism within nodes, and increasingly through GPU acceleration for operations that decompose into many independent floating-point tasks. Validation against known analytic solutions and verification through mesh-refinement studies are standard practices for establishing code correctness. Carnegie Mellon University's computational physics research group illustrates the range of physical systems, from quantum chromodynamics to soft matter, addressed with high-performance methods.

Applications

Physics computing has applications in a range of fields, including:

• Nuclear fusion and plasma physics through MHD and PIC simulations
• Materials science via density functional theory and molecular dynamics
• Astrophysics and cosmology through N-body and hydrodynamic codes
• Quantum chemistry and drug discovery through ab initio calculations
• Weather prediction and climate modeling through atmospheric physics codes