soniumRoom acoustics · 27
Shoebox benchmarks, mesh convergence, backend parity, energy balance, and gallery-manifest integrity for the room simulators. Method shown on hover; tolerance is the pass gate. back to QA
| Case | Tier | Reference | Gate |
|---|---|---|---|
roomsim.application-reference-correlation.v1 | physics-gate | independent monopole Green-function magnitude omega*rho*Q/(4*pi*r) | each numerical/reference magnitude ratio in [0.6, 1.7]; all samples solved |
roomsim.fast-modal-reference.v1 | physics-gate | independent rigid-shoebox Neumann mode and zero-mode pressure formulas | mode and zero-mode errors <= 1e-12; all outputs finite; malformed inputs reject |
roomsim.modal-convergence.v1 | physics-gate | production modal basis ladder with retained complex pressures and adjacent error budgets | adjacent complex-pressure changes <= 15% relative plus 1e-6 Pa absolute |
roomsim.spatial-field-reference.v1 | physics-gate | independent outgoing monopole Green-function 1/r and frequency scaling | finite layout; analytic provenance; radial and frequency ratio errors <= 1e-12 |
roomsim.source-calibration.v1 | physics-gate | independent monopole |p| = omega*rho*|Q|/(4*pi*r) | relative equation error <= 1e-12; round-trip SPL error <= 1e-10 dB |
roomsim.source-calibration-fit.v1 | physics-gate | independent monopole SPL observations generated outside the fit implementation | amplitude and residual error <= 1e-12 |
roomsim.source-power-calibration.v1 | smoke | P(a) = P(a0) (a/a0)^2 with explicit mesh/frequency provenance | successful construction and finite calibrated amplitudes |
roomsim.source-signal-ownership.v1 | smoke | complex speaker grids own crossover magnitude and phase unless chained filtering is explicitly enabled | typed setup validation |
roomsim.treatment-optimization.v1 | physics-gate | independent deterministic fixed-mesh treatment fixture with serial/parallel lane parity and objective reduction | objective reduction > 0; weighted treatment fraction <= 0.600000001; lane objective delta <= 1e-12 |
roomsim.treatment-decay-constraint.v1 | smoke | finite positive Eyring domain, monotonic absorption response, and typed public API surface | compile success |
roomsim.treatment-spatial-constraint.v1 | smoke | deterministic pair construction, exact signed MMA constraints, and bounded treatment API | compile success |
roomsim.treatment-density-filter.v1 | smoke | analytic/deterministic filter constants and linear pullback | successful construction and exact constant preservation |
roomsim.treatment-projection.v1 | smoke | analytic deterministic hat-kernel interpolation with explicit support rejection | successful construction and finite values |
roomsim.measured-response-correlation.v1 | reference-only | independently constructed complex-pressure fixture with branch-cut phase and provenance rejection checks | eight contract metrics valid; exact fixture errors <= 1e-12 |
roomsim.backend-parity-contract.v1 | reference-only | independently constructed deterministic complex-pressure CPU/accelerator fixture | RMS and maximum relative errors <= 1e-10 |
roomsim.mesh-convergence-contract.v1 | reference-only | independently constructed deterministic solved-response fixture | fine-grid RMS and maximum relative errors <= 0.07; both decrease monotonically |
roomsim.fem-mesh-convergence.v1 | physics-gate | production FEM response with retained listener-position and solver diagnostics | complex listener RMS <=0.03; maximum <=0.05; both adjacent refinement contraction ratios <=0.8; all native solves converge |
roomsim.reciprocity-contract.v1 | reference-only | independently constructed symmetric complex-pressure transfer fixture | RMS and maximum reciprocity relative errors <= 1e-12 |
roomsim.energy-balance-contract.v1 | reference-only | independently constructed exact active-power conservation fixture | RMS/max relative and absolute balance errors <= 1e-12 |
roomsim.frequency-segment-merge.v1 | physics-gate | ordered segment concatenation and exact-grid overlap SimulationOutput layout contract | ten exact contract metrics equal 1.0 |
roomsim.multi-listener-reference-correlation.v1 | physics-gate | independent monopole Green-function magnitude omega*rho*Q/(4*pi*r) at two listeners | each numerical/reference magnitude ratio in [0.6, 1.7]; all six samples solved |
roomsim.spatial-metrics-reference.v1 | physics-gate | independent RMS reduction, peak/RMS conversion, listener-order/phase invariance, and production fast-hybrid output | three independent metric checks finite and equal to analytic reductions |
roomsim.cross-backend-spatial-correlation.v1 | physics-gate | normalized listener-pattern Pearson correlation between fast-hybrid and FEM outputs | normalized listener-pattern correlation >= 0.75; finite output and spatial metrics |
roomsim.gallery-artifacts.v1 | physics-gate | validated native gallery manifest, source checksums, and fixed viewport | >=4 fixed-size PNG figures; >=6 fixed-size SVG figures; all required artifacts present |
roomsim.fem-nodal-gallery.v1 | physics-gate | finite accepted nodal field, solved provenance sidecar, 32x32 P1 slice, valid VTU | at least four mesh nodes; exactly 1025 CSV lines; solved/measured=false/fem_nodal_field=true |
roomsim.fem-spatial-field-reference.v1 | physics-gate | independent exterior fundamental solutions; scripts/generate_room_robin_reference.py; src/fixtures/room/robin_box_spatial_reference.json; analytic volume PDE, off-grid Robin residual and source-placement cross-check | fine complex RMS <=0.02, maximum <=0.03; both native refinement ratios <=0.65; all true residuals <=1e-9; reference final complex change, source-placement disagreement and off-grid Robin residual <=1e |
room.shoebox-fem-benchmark.v1 | physics-gate | closed-form Neumann eigenfrequencies including the degenerate pair | frequencies within 10%; solver pair residuals <= 1e-8; independent eigen-residuals <= 1e-6; zero absolute residual <= 1e-9 |