All problem comparisons · Campaign
Gaussian Pulse: M1 RK1 and FLD backward Euler
Saved-state movies
Movies use stored numerical states, held without solution interpolation.
M1 reference: linearized analytic reference. Fourier telegraph-equation reference for isotropic, linearized radiation moments. Nonlinear M1 can retain a closure-model error at finite amplitude.
FLD reference: periodic constant-diffusion analytic reference. Each retained Fourier mode decays as exp[-D |k|² t], with D = c/(3χ) and reference flux -D grad(E). It uses the same filtered initial energy as the M1 run, while M1 has its separate telegraph reference. This is an analytic diffusion-limit reference, not an exact solution of nonlinear flux-limited diffusion.
M1 and FLD solve different transport closures. Both updates are first order in time. A small model discrepancy does not establish accuracy against an exact solution. M1 evolves radiation flux; FLD reconstructs diagnostic flux from its energy gradient. Flux differences are not errors in a shared radiation-momentum conservation law. FLD here supports stationary ideal gas with physical light speed on the CPU.
Level 0
M1 RK1: complete results, parameters and reference discussion
FLD backward Euler: complete results, parameters and reference discussion
matched. No other refinement level or physical time is substituted for a missing partner.

Matched profile coordinates and values (CSV)
FLD minus M1: model discrepancy on the saved slice
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 2.203675e-05 | 5.805825e-05 | 2.261711e-04 |
| Radiation flux x | erg/(cm^2 s) | 2.828845e+05 | 5.418360e+05 | 1.800545e+06 |
| Radiation flux y | erg/(cm^2 s) | 2.828845e+05 | 5.418360e+05 | 1.800545e+06 |
| Radiation flux z | erg/(cm^2 s) | 1.877343e+05 | 4.736037e+05 | 1.800545e+06 |
M1 RK1: model-specific reference diagnostics
linearized analytic reference. Fourier telegraph-equation reference for isotropic, linearized radiation moments. Nonlinear M1 can retain a closure-model error at finite amplitude.
Area-weighted reference errors on this model’s original leaf slice.
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 5.724338e-05 | 1.415729e-04 | 5.451648e-04 |
| Radiation flux x | erg/(cm^2 s) | 9.345218e+05 | 2.020099e+06 | 7.169529e+06 |
| Radiation flux y | erg/(cm^2 s) | 9.345218e+05 | 2.020099e+06 | 7.169529e+06 |
| Radiation flux z | erg/(cm^2 s) | 5.657127e+05 | 1.813816e+06 | 7.169529e+06 |
Solver-reported full-volume reference norms
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 2.077154e-05 | 7.295665e-05 | 5.451648e-04 |
| Radiation flux x | erg/(cm^2 s) | 2.754654e+05 | 1.018561e+06 | 7.169529e+06 |
| Radiation flux y | erg/(cm^2 s) | 2.754654e+05 | 1.018561e+06 | 7.169529e+06 |
| Radiation flux z | erg/(cm^2 s) | 2.754654e+05 | 1.018561e+06 | 7.169529e+06 |
Source slice CSV · Run metadata
FLD backward Euler: model-specific reference diagnostics
periodic constant-diffusion analytic reference. Each retained Fourier mode decays as exp[-D |k|² t], with D = c/(3χ) and reference flux -D grad(E). It uses the same filtered initial energy as the M1 run, while M1 has its separate telegraph reference. This is an analytic diffusion-limit reference, not an exact solution of nonlinear flux-limited diffusion.
Area-weighted reference errors on this model’s original leaf slice.
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 2.312297e-05 | 5.064725e-05 | 1.861876e-04 |
| Radiation flux x | erg/(cm^2 s) | 2.092399e+05 | 3.542431e+05 | 8.783358e+05 |
| Radiation flux y | erg/(cm^2 s) | 2.092399e+05 | 3.542431e+05 | 8.783358e+05 |
| Radiation flux z | erg/(cm^2 s) | 1.601308e+05 | 3.261437e+05 | 8.783358e+05 |
Level 1
M1 RK1: complete results, parameters and reference discussion
FLD backward Euler: complete results, parameters and reference discussion
matched. No other refinement level or physical time is substituted for a missing partner.

Matched profile coordinates and values (CSV)
FLD minus M1: model discrepancy on the saved slice
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 4.032591e-05 | 8.036957e-05 | 2.775694e-04 |
| Radiation flux x | erg/(cm^2 s) | 5.839559e+05 | 1.340777e+06 | 7.581579e+06 |
| Radiation flux y | erg/(cm^2 s) | 5.839559e+05 | 1.340777e+06 | 7.581579e+06 |
| Radiation flux z | erg/(cm^2 s) | 1.188700e+05 | 2.375642e+05 | 1.235064e+06 |
M1 RK1: model-specific reference diagnostics
linearized analytic reference. Fourier telegraph-equation reference for isotropic, linearized radiation moments. Nonlinear M1 can retain a closure-model error at finite amplitude.
Area-weighted reference errors on this model’s original leaf slice.
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 2.482571e-05 | 7.264041e-05 | 3.228860e-04 |
| Radiation flux x | erg/(cm^2 s) | 4.062469e+05 | 1.379910e+06 | 7.285653e+06 |
| Radiation flux y | erg/(cm^2 s) | 4.062469e+05 | 1.379910e+06 | 7.285653e+06 |
| Radiation flux z | erg/(cm^2 s) | 3.593163e+05 | 1.235110e+06 | 7.285653e+06 |
Solver-reported full-volume reference norms
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 7.580997e-06 | 3.345063e-05 | 3.228860e-04 |
| Radiation flux x | erg/(cm^2 s) | 9.243889e+04 | 5.521543e+05 | 7.285653e+06 |
| Radiation flux y | erg/(cm^2 s) | 9.243889e+04 | 5.521543e+05 | 7.285653e+06 |
| Radiation flux z | erg/(cm^2 s) | 9.243889e+04 | 5.521543e+05 | 7.285653e+06 |
Source slice CSV · Run metadata
FLD backward Euler: model-specific reference diagnostics
periodic constant-diffusion analytic reference. Each retained Fourier mode decays as exp[-D |k|² t], with D = c/(3χ) and reference flux -D grad(E). It uses the same filtered initial energy as the M1 run, while M1 has its separate telegraph reference. This is an analytic diffusion-limit reference, not an exact solution of nonlinear flux-limited diffusion.
Area-weighted reference errors on this model’s original leaf slice.
| Field | Units | L₁ | L₂ | L∞ |
|---|---|---|---|---|
| Radiation energy density | erg/cm^3 | 1.089667e-05 | 3.629176e-05 | 2.444300e-04 |
| Radiation flux x | erg/(cm^2 s) | 7.780558e+04 | 1.604165e+05 | 8.617109e+05 |
| Radiation flux y | erg/(cm^2 s) | 7.780558e+04 | 1.604165e+05 | 8.617109e+05 |
| Radiation flux z | erg/(cm^2 s) | 3.036714e+04 | 8.303923e+04 | 5.479825e+05 |
How these comparisons are measured
All field values retain their saved units. Central profiles share x coordinates and the positive row adjacent to y = z = 0. Adaptive leaf-cell averages are prolonged as constants onto the finest common Cartesian x-y partition, without extrapolation or time interpolation. Different leaf sizes may average different z extents adjacent to the plane. These are saved finite-volume slices.
Slice L₁ is mean absolute discrepancy, L₂ is root-mean-square discrepancy, and L∞ is maximum absolute discrepancy, with area weighting. These are two-dimensional diagnostics; separately labeled solver reference norms cover the full volume. Each model is measured against its own saved reference. A diffusion-limit error includes departures of the nonlinear closure from that limit.