M1 RK1 and FLD
Movies
19 movies available from 19 completed runs. Open the movie gallery.
Saved states are shown with zero-order hold and no solution interpolation.
Status: ready. 48 uniform-grid runs: eight problems, both models, levels 2/3/4 (32/64/128 cells per side). 121 requested physical snapshots and 60-second movies. Positive streaming optical depth 1e-4 for both models. Full shock/blast durations, longer Gaussian/Marshak evolution. Preserve validated samples and initial/final 3D states; retire intermediate 3D files only after verified extraction. First-order moving FLD uses a serial root-gathered composite solve and remains subject to memory limits. Full-resolution runtime and timestep convergence are unverified. Start only on explicit command. Model discrepancies are not FLD error estimates. FLD uses a serial root-gathered multigrid solve. No moving-gas or scaling claim.
M1 evolves radiation energy and flux using RK1 with radiation substeps. FLD solves energy and material heating together implicitly; its flux is a derived diagnostic. Cross-model differences are not truncation-error estimates.
Model comparison plots
Time is in seconds except for normalized Marshak time. Gaussian, equilibrium sphere and Marshak have analytic diffusion-limit references for FLD energy and flux; Marshak also has material energy. These are not exact nonlinear LP FLD solutions. Thin positive-opacity streaming cases compare closures and conservation without claiming an analytic FLD profile. Reference discrepancies remain separate from cross-model differences. Prepared runs are not validation evidence.
Supported scope
All eight problems are planned for both models at physical light speed on CPU. FLD uses a mixed-frame moving-gas update for shocks/blast. Streaming opacity is explicitly positive, and zero-opacity advection curves are limit references. Any legacy exclusions are listed here:
Plan and provenance · Prepared commands
Execution requires a separate explicit start. This workflow does not publish to a remote website.