PIC Simulation Planner
Before you request an allocation or launch a job: estimate grid resolution, time step, memory and compute cost for a particle-in-cell simulation, and get warnings about common pitfalls.
Inputs
Results
Email me this report + expert setup advice
Get a formatted copy of these results in your inbox. Reply to it and a physicist will review your planned setup for free.
Method, assumptions & references
Cell size is set by the most restrictive physical length scale: λ/30 for laser problems, c/ωp÷10 (the collisionless skin depth) for electromagnetic problems, and Δx ≤ λD for explicit electrostatic PIC, which avoids finite-grid heating. The time step follows the Courant condition c·Δt < Δx/√D for electromagnetic codes, or ωpΔt ≤ 0.2 for electrostatic codes.
Memory assumes ~64 bytes per macro-particle and ~160 bytes per cell (fields, currents and buffers), with 30% overhead. Cost assumes ~5×10⁶ particle pushes per CPU core-second and ~5×10⁸ per modern GPU. Real codes vary by about ×3 either way depending on shape order, diagnostics and load balance.
References: Birdsall & Langdon, Plasma Physics via Computer Simulation; Hockney & Eastwood, Computer Simulation Using Particles; Arber et al., PPCF 57, 113001 (2015).
Need this simulation set up, run and verified?
We handle the physics, the numerics and the HPC. You get trustworthy results.