Acoustic scattering — BEM

Acoustic scattering — BEM · 153

Burton-Miller and transmission BEM, Galerkin formulations, and compressed operators with convergence ladders and adjoint certification. Method shown on hover; tolerance is the pass gate. back to QA

Case	Tier	Reference	Gate
`bem.layered-porous-surface.v1`	physics-gate	independent characteristic-impedance transfer-matrix recursion and centered frequency derivative	single-layer and transfer errors <= 1e-12; derivative error <= 2e-5; absorption in [0,1]; four typed rejections
`bem.galerkin-quad4-compressed-native.v1`	physics-gate	independent dense global Müller action plus admitted fine-source certified solve	both relative action errors <= 2e-7; exact-source residual <= 1e-8; all metrics finite
`bem.fem-bem-trace-transfer.v1`	smoke	constant preservation, lumped-measure conservation, weighted Hermitian adjoint, coupled block materialization, typed rejection	all transfer, coupled forward/adjoint, and residual errors <= 1e-14; two typed invalid-input rejections
`bem.fem-bem-trace-operator.v1`	contract-gate	dense CoupledTraceSystem action parity and exact adjoint capability	all three action errors <= 1e-12
`bem.fem-bem-preconditioner.v1`	contract-gate	independent dense block actions, exact algebraic-transpose/Hermitian parity, typed rejection	all action errors <= 1e-12; one unknown-adjoint and one bad-state rejection
`bem.2d-rigid-cylinder-reference.v1`	physics-gate	independent 2-D rigid-cylinder series field/far-field amplitude and RMS/peak per-length power	four field/far-field/power and RMS/peak plus invalid-query admission checks pass
`bem.2d-fluid-cylinder-reference.v1`	physics-gate	independent cylindrical Fourier pressure, far-field amplitude, and power reference	exterior/interior fields, interface traces, far-field amplitude, power, and invalid-query admission pass documented bounds
`bem.2d-transmission-adjoint.v1`	physics-gate	independent assembled matrix action, true adjoint residual, and complex inner-product identity	forward/adjoint/batch residuals and solution-action error <= 1e-10; adjoint invalid-point rejection >= 1; inner-product error <= 2e-12
`bem.transmission-adjoint.v1`	physics-gate	independent Aᴴ residual and complex inner-product identity	adjoint/batch residuals <= 1e-10; inner-product error <= 1e-9
`bem.transmission-reciprocity.v1`	physics-gate	weighted bilinear identity J A = (J A)^T with J pressure weight i/(s omega)	maximum weighted reciprocity relative error <= 1e-6
`bem.transmission-operator-contract.v1`	physics-gate	independent dense and matrix-free weighted transmission contracts	dense reciprocity <= 1e-6; compressed reciprocity <= 5e-4; power balance <= 1e-12; passivity minimum >= 0
`bem.transmission-higher-order.v1`	physics-gate	finite block matrices, certified linear residual, finite exterior field	linear residual <= 1e-9
`bem.transmission-continuous-quadratic.v1`	physics-gate	shared-DOF reduction, finite matrix/field, certified solve residual, independent FluidSphereTransmission3d field	forward/adjoint residuals <= 1e-8; maximum Mie field error < 0.15
`bem.transmission-compressed-parity.v1`	physics-gate	independently assembled dense transmission matrix action, conjugate-transpose action, forward/adjoint ILU-GMRES residuals, exact dense-source residual sentinel	dense-source actions <= 2e-2; on-demand parity <= 1e-12; compressed and exact source solve residuals <= 1e-8; at least one compressed block; adjoint batch count = 2; non-finite ACA/output rejection >=
`bem.transmission-compressed-block-action.v1`	physics-gate	independent column-wise LinearOperator actions	block-action relative error <= 1e-13; at least one compressed block; three RHS columns
`bem.transmission-compressed-adjoint-block-action.v1`	physics-gate	independent column-wise compressed Hermitian actions	Hermitian block-action relative error <= 1e-13; at least one compressed block; three RHS columns
`bem.transmission-kernel-native-parity.v1`	physics-gate	dense constant-panel transmission matrix assembled by the same production panel kernels	forward and transpose action errors <= 2e-7
`bem.galerkin-single-layer.v1`	physics-gate	matrix finiteness and transpose symmetry across self, common-edge, common-vertex, and separated pair routes	finite entries; matrix asymmetry <= 2e-8
`bem.galerkin-quad4-pair.v1`	physics-gate	independent reverse-pair Galerkin block	reciprocity error <= 2e-10; finite linear/quadratic blocks; one common-edge and one common-vertex mapped touching pair
`bem.galerkin-quad4-transmission-pair.v1`	physics-gate	finite curved physical-map block and changed output under non-planar corner geometry	curved geometry delta > 1e-6; self/touching blocks finite; permissive adjacent-order policy passes; zero-budget dense and compression gates reject
`bem.galerkin-quad4-transmission-self.v1`	physics-gate	finite nonzero linear/quadratic self blocks and finite global self matrix	linear 8x8 and quadratic 18x18 blocks finite/nonzero; global 8x8 matrix finite/nonzero
`bem.galerkin-linear-single-layer.v1`	physics-gate	finite panel-local linear Galerkin matrix with reported transpose diagnostic	all entries finite; transpose asymmetry is reported diagnostically
`bem.galerkin-quadratic-single-layer.v1`	physics-gate	finite panel-local quadratic Galerkin matrix with reported transpose diagnostic	all entries finite; transpose asymmetry is reported diagnostically
`bem.galerkin-conforming-linear-single-layer.v1`	physics-gate	finite conforming global matrix with reported transpose diagnostic	all entries finite; transpose asymmetry reported diagnostically
`bem.galerkin-double-layer-principal-value.v1`	physics-gate	finite touching-pair matrices, quadrature convergence, and separated K/K' transpose action	finite entries; touching convergence <= 5e-4; separated reciprocity <= 2e-8
`bem.galerkin-quadratic-double-layer.v1`	physics-gate	finite quadratic touching-pair matrices and separated K/K' transpose action	self-panel norm <= 1e-14; finite entries; separated reciprocity <= 2e-7
`bem.galerkin-hypersingular-weak-form.v1`	physics-gate	finite complex-symmetric W matrix and closed-surface edge/DOF contract	finite entries; complex-symmetric asymmetry <= 1e-12
`bem.galerkin-quadratic-hypersingular-weak-form.v1`	physics-gate	finite complex-symmetric quadratic W matrix and closed-surface edge/DOF contract	finite entries; complex-symmetric asymmetry <= 1e-12
`bem.galerkin-hypersingular-open-boundary.v1`	physics-gate	direct finite-part Galerkin kernel parity with finite matrix entries	finite entries; each direct parity error <= 1e-10
`bem.galerkin-hypersingular-open-compressed.v1`	physics-gate	independent dense global Müller action on the same open finite-part entries	all action errors <= 1e-9; all metrics finite
`bem.galerkin-muller-transmission.v1`	physics-gate	finite nonzero two-medium pressure and normal-velocity diagonal blocks with checked frequency/harmonic admission	all entries and field samples finite; both diagonal blocks strictly positive; both compressed action errors and compressed field error <= 1e-7 with at least one ACA block each; solve residuals <= 1e-8
`bem.galerkin-curved-quad4-compressed-field.v1`	physics-gate	dense curved Quad4 Galerkin Müller exterior total-pressure field	compressed field error <= 2e-6; matrix dimension = 52; all checked field values finite
`bem.galerkin-muller-mie-convergence.v1`	physics-gate	independent two-medium spherical-Mie exterior total-pressure samples	fine L2 <= 0.001; fine pointwise <= 0.002; both adjacent L2 and pointwise refinement ratios <= 0.5; all three solve residuals <= 1e-8; three numerical solves; fine field magnitude > 0
`bem.transmission-distributed-block-rows.v1`	physics-gate	independent column-wise compressed forward and transpose row actions	forward and transpose relative errors <= 1e-13; three rows; three RHS columns
`bem.transmission-distributed-hermitian-block-rows.v1`	physics-gate	independent column-wise compressed Hermitian row actions	Hermitian relative error <= 1e-13; three rows; three RHS columns
`bem.transmission-distributed-rows.v1`	physics-gate	global compressed forward and transpose actions on the same CPU operator	local action errors <= 1e-10; all 40 trace rows owned exactly once
`bem.transmission-distributed-higher-order-rows.v1`	physics-gate	global dense-ACA forward and transpose actions on the same CPU operator	local action errors <= 1e-10; every pressure and velocity row owned exactly once
`bem.transmission-quadrature-policy.v1`	physics-gate	direct and typed-XEM checked builders agree in dimension; invalid order is rejected with typed medium error	valid direct and typed-XEM paths agree; invalid policy rejection = 1
`bem.transmission-compressed-field-parity.v1`	physics-gate	independent dense source operator and forward/Hermitian solutions on the identical mesh with fluid contrast	all compressed solve residuals <=1e-10; fine field and dense residual errors <=1e-5; fine action errors <=1e-6; every adjacent field/action ratio <=0.25 above roundoff floor 1e-12; at least one low-ra
`bem.adjoint-many.v1`	physics-gate	independent dense and compressed Hermitian-transposed matrix residual certification	three RHS columns; dense residual <= 1e-10; H-matrix residual <= 1e-6
`bem.adjoint-frequency-batch.v1`	physics-gate	independent dense and compressed exact-TBEM Hermitian-adjoint residual certification for every frequency/RHS column	four compressed backends across two frequencies and two RHS columns; dense residual <= 1e-10; compressed residual <= 1e-6; independent exact-reference residual <= 1e-4; coefficient error <= 1e-4
`bem.adjoint-compressed-backends.v1`	physics-gate	independent dense adjoint field, exact residual, and checked invalid-point certification for each compressed backend	four compressed backends across two frequencies; compressed residual <= 1e-6; independent exact-reference residual <= 1e-4; field error <= 1e-6; all eight invalid points rejected
`bem.driver-loads.v1`	physics-gate	independent dense and compressed exact forward residual certification	two frequencies and two RHS columns; dense residual <= 1e-10; H-matrix residual <= 1e-6
`bem.driver-loads-compressed.v1`	physics-gate	independent matrix-free exact TBEM forward residual certification for compressed backends	three compressed backends; maximum residual <= 1e-6; exact-reference residual <= 1e-5; coefficient and field error <= 5e-5
`bem.higher-order-exact-residual.v1`	physics-gate	independent matrix-free basis-weighted TBEM forward/Hermitian action with curved collocation/nodal normals for FMM paths	eight cases (H-matrix linear/quadratic/continuous quadratic, impedance, SLFMM discontinuous/continuous quadratic, MLFMM discontinuous/continuous quadratic, directional-MLFMM continuous quadratic); for
`bem.wavenumber-objective-gradient.v1`	physics-gate	independent central finite differences in wavenumber and sound speed	both relative errors <= 2e-5
`bem.wavenumber-objective-gradient-batch.v1`	physics-gate	independent per-frequency BemSolver::solve_wavenumber_objective_gradient results	two-frequency batch/scalar error <= 1e-12; finite gradients; mismatched batch cardinality rejected
`bem.hmatrix-wavenumber-objective-gradient.v1`	physics-gate	independent centered finite difference in wavenumber	backend-maximum relative error <= 5e-2
`bem.compressed-field-parity.v1`	physics-gate	independent dense TBEM pressure field plus checked finite-value/invalid-point contract	maximum relative field error <= 5e-2; all four checked backends and invalid-point rejections
`bem.checked-field-evaluation.v1`	physics-gate	infallible field parity plus typed FieldEvaluationError rejection contracts	all checked values finite; non-finite points and malformed buffers reject
`bem.field-gradient.v1`	physics-gate	independent central difference of the production pressure evaluator	maximum relative error <= 3e-3; invalid point rejected
`bem.hmatrix-material-objective-gradients.v1`	physics-gate	independent centered finite differences in sound speed and density	both backend-maximum relative errors <= 5e-2
`bem.compressed-impedance-material-objective-gradient.v1`	physics-gate	independent centered H-matrix solves for speed and density perturbations	both relative errors <= 3e-4; all metrics finite
`bem.compressed-higher-order-material-objective-gradients.v1`	physics-gate	independent centered finite differences in sound speed and density	both backend-maximum relative errors <= 5e-2
`bem.compressed-higher-order-wavenumber-objective-gradient.v1`	physics-gate	independent centered finite difference in wavenumber	maximum backend relative error <= 5e-2
`bem.higher-order-mlfmm-wavenumber-derivative.v1`	physics-gate	independent centered wavenumber perturbation of the same MLFMM value factors	both relative errors <= 3e-3
`bem.higher-order-soft-dirichlet.v1`	physics-gate	finite-value and field-evaluation contract across three basis configurations	at least three configurations, at least one DOF, and finite surface/field values
`bem.density-objective-gradient.v1`	physics-gate	independent centered finite difference in medium density	relative error <= 2e-5
`bem.near-singular-quadrature-reference.v1`	physics-gate	independent higher-order regular integration rule	strictly monotone errors and order-32 absolute error < 1e-10
`bem.distributed-ownership.v1`	physics-gate	independent sorted global-row ownership and deterministic replay	duplicate row count = 0; all shared rows owned exactly once; scatter/gather round-trip = 1
`bem.distributed-krylov.v1`	physics-gate	loopback all-gather, independent diagonal solution, and global residual replay	relative residual <= 1e-12; maximum solution error <= 1e-10; typed rejection count >= 2
`bem.compressed-wavenumber-derivative.v1`	physics-gate	independent central difference of compressed H-matrix actions	relative action error <= 3e-4; at least one low-rank block
`bem.compressed-fmm-wavenumber-derivative.v1`	physics-gate	independent centered compressed SLFMM/MLFMM/directional-MLFMM actions	all relative action errors <= 2e-3
`bem.batch-schedule.v1`	reference-only	independent schedule validation and exact workload coverage	covered items >= 30; peak reserved RAM <= 130 bytes
`bem.transmission-transparent-limit.v1`	physics-gate	independent analytic plane-wave field in the equal-medium transparent-interface limit	total-field and scattered-pressure errors <= 1e-3; certified relative residual <= 1e-10
`bem.sphere-mie-3d.v1`	physics-gate	analytical Mie solution	complex L2 <= 1e-2; surface L2 <= 2e-2; scattered power <= 6e-2
`bem.sphere-field-pattern-mie.v1`	physics-gate	sonium-wave::analytical::sphere_scattering_3d independent rigid-sphere Mie field pattern	l2_relative_error <= 1.5e-1; max_pointwise_relative_error <= 2.5e-1
`bem.sphere-reciprocity-passivity.v1`	physics-gate	independent rigid-sphere Mie scattering series	reciprocity <= 2e-2; scattered-power <= 8e-2; field L2 <= 1.5e-1; pointwise <= 2.5e-1
`bem.sphere-hmatrix-backend-parity.v1`	physics-gate	dense TBEM and independent rigid-sphere Mie scattering series	H-matrix backscatter <= 2e-5; reciprocity <= 2e-2; Mie scattered-power <= 8e-2
`bem.sphere-backend-formulation-matrix.v1`	physics-gate	independent rigid-sphere Mie scattering series	all eight rows satisfy the production test gates
`bem.fluid-interface-transmission.v1`	physics-gate	independent planar Fresnel pressure and normal-power coefficients	coefficient/power error <= 2e-13; two typed rejection checks
`bem.transmission-operator-contracts.v1`	smoke	independent off-diagonal kernel fixture, typed mixed-frequency rejection, incident-trace RHS finiteness, and solved residual	finite matrix entries; matrix L1 norm > 0; mixed-frequency rejection dimension-matched finite incident RHS and solve residual <= 1e-10
`bem.transmission-material-derivatives.v1`	physics-gate	independent centered speed and density rebuilds plus dk/dc chain rule	speed finite-difference <= 4e-5; speed chain <= 2e-12; density finite-difference <= 2e-8; continuous-quadratic speed finite-difference <= 4e-5; linear and continuous-quadratic derivative entries finit
`bem.compressed-transmission-material-derivative.v1`	physics-gate	dense and kernel-native compressed forward actions against the dense material-derivative matrix	both relative compressed-action errors <= 2e-7
`bem.impedance-speed-derivative.v1`	physics-gate	independent centered speed perturbation including Y=1/(rho*c)	both relative errors <= 2e-5; all metrics finite
`bem.galerkin-transmission-material-derivatives.v1`	physics-gate	independent centered Galerkin matrix rebuilds	all derivative entries finite; centered-reassembly error <= 1e-10; invalid medium and constant basis rejected
`bem.galerkin-compressed-transmission-material-derivative.v1`	physics-gate	compressed forward action against the fine-order dense Galerkin derivative	relative compressed-action error <= 2e-7
`bem.galerkin-quad4-pair-material-derivative.v1`	physics-gate	finite nonzero centered pair derivative matrix	8x8 finite derivative; maximum entry norm positive
`bem.galerkin-quad4-material-derivative.v1`	physics-gate	fine-order dense global-DOF material derivative action	analytic-centered relative error <= 5e-3; compressed action relative error <= 2e-6
`bem.galerkin-quad4-self-wavenumber-derivative.v1`	physics-gate	centered production self-panel Duffy derivative	maximum relative derivative error <= 5e-4
`bem.galerkin-quad4-global-wavenumber-derivative.v1`	physics-gate	centered production global Quad4 Müller matrix and dense derivative action	matrix relative error <= 2e-3; compressed action error <= 1e-6
`bem.typed-raw-transmission-parity.v1`	physics-gate	independent off-diagonal Green-kernel and k = 2πf/c fixture	typed/raw relative matrix error <= 1e-14; independent wavenumber checks
`bem.operator-consistency.v1`	physics-gate	dense TBEM action, exact small-system identities, over-integrated quadrature, MLFMM sphere parity, and independent Mie far-field power/radius-invariance shared by dense/H-matrix/SLFMM/MLFMM solves	every dense/H-matrix/SLFMM/MLFMM sphere solve: power <= 6e-2, far-field radius-invariance <= 1.5e-2, and true residual <= 2e-8; separated-cluster H-matrix action <= 2e-4, SLFMM action <= 2e-3, and at
`bem.mlfmm-slow-validation.v1`	physics-gate	independently assembled stabilized dense TBEM actions	MLFMM dense action <= 1e-2; M2L pair action <= 1e-4; SLFMM 16-term far-field action <= 1e-5
`bem.frequency-sweep-backends.v1`	physics-gate	independent matrix-free stabilized TBEM action	exact residual <= 2e-5; operator action error <= 1e-3
`bem.forward-frequency-batch-compressed.v1`	physics-gate	independent dense direct forward frequency-batch solve	four compressed backends including directional-MLFMM; residual <= 1e-6; coefficient error <= 5e-5
`bem.formulation-equivalence-reference.v1`	physics-gate	independent rigid-sphere Mie field and regular/Burton-Miller equivalence	Mie relative field error <= 1.5e-1; formulation relative error <= 8e-2
`bem.linear-basis-execution.v1`	physics-gate	independent surface-panel topology and finite-field execution contract	exact linear DOF count; all coefficients and point evaluation finite
`bem.higher-order-hmatrix.v1`	physics-gate	finite coefficients and exterior point evaluation on deterministic higher-order compressed fixture	all coefficients and field values finite
`bem.higher-order-formulation-reference.v1`	physics-gate	independent spherical Mie fields, checked field evaluation, linear/quadratic panel paths, formulation equivalence, and axial invariance	linear fine Mie/formulation thresholds plus bounded quadratic Mie reference
`bem.2d-fluid-cylinder-convergence.v1`	physics-gate	independent cylindrical Fourier transmission field and scattered-power-per-length reference	field and power errors refine monotonically; fine field errors <= 0.12; residual <= 1e-3
`bem.2d-fluid-cylinder-batch.v1`	physics-gate	independent single-RHS production solves using the same assembled operator	maximum coefficient error <= 1e-11; residual <= 1e-10
`bem.2d-rigid-cylinder-batch.v1`	physics-gate	independent single-RHS production solves using the same assembled Neumann operator	maximum coefficient error <= 1e-11; residual <= 1e-10
`bem.gmres-ir-failure-matrix.v1`	smoke	independent checked failure/status matrix	7 invalid-input cases and 1 inner-failure case reject without panic
`bem.fmm-hierarchical-preconditioner.v1`	physics-gate	independent full-SLFMM operator residual certification	true residual <= 1e-6
`bem.mlfmm-nearfield-ilu.v1`	smoke	independent dense near-field matrix action	forward/adjoint residual <= 1e-11 and inner-product error <= 1e-12
`bem.slfmm-nearfield-ilu.v1`	physics-gate	independent full-SLFMM operator residual certification	true residual <= 1e-6
`bem.slfmm-adjoint-preconditioners.v1`	contract-gate	independent full matrix-free SLFMM Hermitian-adjoint action	both residuals <= 1e-11; solution delta <= 1e-11; one non-finite RHS rejection
`bem.mlfmm-adjoint-preconditioners.v1`	contract-gate	matrix-free MLFMM Hermitian-adjoint action	both residuals <= 1e-11; solution delta <= 1e-11; one non-finite RHS rejection
`bem.high-level-hilu-dispatch.v1`	contract-gate	independent exact full-operator forward and Hermitian-adjoint residuals	both residuals <= 1e-4; one non-FMM rejection
`bem.continuous-quadratic-topology.v1`	physics-gate	independent global vertex/edge DOF deduplication finite production solve	shared vertex/edge DOFs are deduplicated and the production solve is finite
`bem.continuous-quadratic-quad4.v1`	physics-gate	independent closed-cube Q2 vertex/edge/interior DOF count and finite production solve	global DOFs = 26; local Q2 slots = 54; shared slots = 28; all pressure coefficients finite
`bem.continuous-quadratic-mie-convergence.v1`	physics-gate	independent spherical-Mie exterior fields plus rotational-invariance identity	fine L2 <= 1e-2; fine pointwise <= 2e-2; both refinement ratios <= 0.85; rotational error <= 1e-2
`bem.continuous-quadratic-quad4-mie-convergence.v1`	physics-gate	independent spherical-Mie exterior fields plus rotational-invariance identity	fine L2 <= 1e-2; fine pointwise <= 2e-2; both refinement ratios <= 0.85; rotational error <= 1e-2
`bem.transmission-continuous-quadratic-quad4-mie-convergence.v1`	physics-gate	independent FluidSphereTransmission3d exterior pressure fields; direct coefficient cross-check and separately recomputed dense equation residual	fine field error <=0.01; both adjacent refinement ratios <=0.65; all solve residuals and direct coefficient errors <=1e-8; three numerical solves; fine DOFs/panels increase
`bem.axisymmetric-ring-mie-convergence.v1`	physics-gate	independent spherical-Mie exterior total pressure	fine L2 <= 1e-7; fine pointwise/far-field <= 2e-7; all refinement ratios <= 1e-2; residuals <= 1e-10
`bem.axisymmetric-penetrable-sphere-mie-convergence.v1`	physics-gate	independent FluidSphereTransmission3d Mie exterior/interior pressure reference	fine exterior/interior errors <= 1e-4; fine far-field error <= 1e-5; exterior/interior refinement ratios <= 0.1; far-field refinement ratio <= 0.01; Mie truncation change <= 1e-10; forward/adjoint/bat
`bem.axisymmetric-profile-forward-batch.v1`	physics-gate	independent scalar AxisymmetricPenetrableProfile::solve solutions	coefficient error <= 1e-10; residual <= 1e-9; empty batch accepted; non-finite amplitude rejected
`bem.axisymmetric-profile-transmission.v1`	physics-gate	sphere operator parity plus independent residual, continuity, field-finiteness, and single/batch adjoint contracts	sphere parity <= 1e-12; forward/single/batch adjoint/continuity <= 1e-9; compressed action <= 2e-2 with at least one low-rank block; finite field count >= 2; typed rejections >= 2
`bem.tbem-quadrature-policy-reference.v1`	physics-gate	independent permissive-pass and unrealistically strict-rejection contract	permissive reference policy passes; strict zero mixed budget rejects; checked DOFs > 0
`bem.penetrable-sphere-mie-convergence.v1`	physics-gate	independent two-medium spherical-Mie pressure/velocity traces and exterior fields	trace residual <= 3e-2 and refinement ratio <= 0.5; field L2 <= 5e-3; pointwise <= 1e-2; solve <= 1e-9; axisymmetry <= 1e-2; power imbalance <= 2e-2
`bem.quad4-higher-order-regular-singular-reference.v1`	reference-only	independent tensor-Gauss, Duffy, partition-of-unity, and matrix-finiteness checks	regular < 2e-7; singular < 3e-4; partition < 2e-11; 9x9 matrix finite and nonzero
`bem.higher-order-field-evaluation-reference.v1`	physics-gate	quadratic pressure/velocity DOF interpolation and checked buffer-layout rejection	nonzero pressure/velocity responses; expected velocity length 9 rejected for actual length 1
`bem.quadratic-vtu-export.v1`	physics-gate	independent VTU parser and required pressure point-field names	at least one surface element, six exported points, four exported cells, six pressure DOFs
`bem.source-position-objective-gradient.v1`	physics-gate	independent centered finite-difference objective evaluations	maximum relative error <= 2e-5
`bem.higher-order-source-position-objective-gradient.v1`	physics-gate	independent centered finite-difference objective evaluations	maximum relative error <= 2e-4
`bem.compressed-source-position-objective-gradient.v1`	physics-gate	independent centered finite differences for x/y/z source position	maximum relative and absolute error <= 5e-4
`bem.geometry-translation-objective-gradient.v1`	physics-gate	independent centered finite-difference objective evaluations after translating nodes and panel centers	maximum relative error <= 2e-5
`bem.compressed-geometry-translation-objective-gradient.v1`	physics-gate	independent centered finite differences for x/y/z translations	maximum relative and absolute error <= 5e-4
`bem.compressed-higher-order-geometry-translation-objective-gradient.v1`	physics-gate	independent centered finite differences for x/y/z translations	maximum relative and absolute error <= 1e-8
`bem.operator-cache-reuse.v1`	physics-gate	independent second dense solve with exact coefficient comparison	maximum solution difference <= 1e-10; one rebuild and one reuse hit
`bem.operator-cache-hmatrix-reuse.v1`	physics-gate	independent second H-matrix solve exact coefficient comparison	maximum solution difference <= 1e-10; one rebuild and one reuse hit
`bem.higher-order-hmatrix-quad4.v1`	physics-gate	finite coefficients and exterior point fields on one deterministic Quad4 fixture	all coefficients and field values finite
`bem.continuous-higher-order-hmatrix-quad4.v1`	physics-gate	finite coefficients and exterior point field on one deterministic Quad4 fixture	all reported values finite; DOF count > 0
`bem.continuous-linear-hmatrix-quad4.v1`	physics-gate	eight shared vertex DOFs and dense-TBEM exterior field parity on one deterministic fixture	DOF count = 8; all reported values finite; max relative field error <= 1e-4
`bem.slfmm-checked-admission.v1`	contract-gate	one valid constant-panel DOF plus typed higher-order/topology/quadrature rejections	valid_constant_dofs >= 1; basis_rejections >= 1; topology_rejections >= 1
`bem.fmm-block-row-actions.v1`	contract-gate	single-RHS row projections and fail-closed dimension rejection	maximum block action error <= 0; dimension rejection >= 1
`bem.mlfmm-checked-admission.v1`	contract-gate	valid constant-panel Tri3/Quad4 DOFs, typed higher-order/cluster rejections, and checked actions	valid_constant_dofs >= 1; valid_quad4_dofs >= 1; basis_rejections >= 1; topology_rejections >= 1; checked_actions >= 3
`bem.dense-operator-checked-admission.v1`	contract-gate	independent DenseMatrix forward, algebraic-transpose, Hermitian, and block products	maximum parity error <= 1e-12; checked_actions >= 6; dimension_rejections >= 3; nonfinite_rejections >= 3
`bem.mlfmm-curved-quad4-normals.v1`	physics-gate	analytic Burton-Miller local expansion action using geometric collocation normals	max_proxy_normal_error <= 1e-12; normal_variation > 1e-6
`bem.operator-checked-admission.v1`	contract-gate	constant-panel DOF count, node bounds, duplicate and dense numbering	at least 3 rejected layouts and 1 duplicate-layout rejection
`bem.higher-order-objective-gradients.v1`	physics-gate	independent centered finite differences in frequency and sound speed	relative error <= 5e-3
`bem.higher-order-mlfmm-quad4.v1`	physics-gate	independent dense quadratic TBEM forward/transpose/Hermitian actions plus directional parity and continuous shared-DOF solve	directional action_error <= 1e-10; full dense-reference action error <= 1e-8; directional dense-reference action error <= 1e-3; full_dofs = directional_dofs = 9; continuous_dofs = 26; far_field_entrie
`bem.compressed-higher-order-source-position-objective-gradient.v1`	physics-gate	independent centered finite-difference objective evaluations	maximum relative and absolute error <= 5e-3
`bem.operator-cache-soft-dirichlet.v1`	physics-gate	independent indirect soft-Dirichlet solve and rigid-boundary rebuild	maximum cache/batch/adjoint metrics <= 1e-10; at least two rebuilds and one reuse hit
`bem.transmission-kernel-native-shared-higher-order-parity.v1`	physics-gate	dense continuous-quadratic transmission matrix and analytic wavenumber derivative	value and derivative action errors <= 2e-7
`bem.transmission-kernel-native-exact-residual.v1`	physics-gate	dense transmission matrix retained by kernel-native owner	forward and Hermitian-adjoint exact-source relative residuals <= 1e-9
`bem.hmatrix-transpose-block.v1`	contract-gate	scalar H-matrix algebraic-transpose action	maximum block/scalar transpose error <= 1e-10
`bem.hmatrix-operator-accuracy.v1`	contract-gate	independently assembled dense stabilized TBEM matrix	each maximum relative action error <= 1e-8; all metrics finite
`bem.fmm-operator-accuracy.v1`	contract-gate	independently assembled dense stabilized TBEM matrix	each maximum relative action error <= 1e-6; four scalar/block non-finite row-action paths rejected per backend
`bem.slfmm-transpose-block.v1`	contract-gate	scalar SLFMM algebraic-transpose action	maximum block/scalar transpose error <= 1e-12
`bem.mlfmm-transpose-block.v1`	contract-gate	scalar MLFMM algebraic-transpose action	maximum block/scalar transpose error <= 1e-12
`bem.higher-order-dual-basis-field.v1`	physics-gate	independent quadratic Tri3 constant-trace mass moments and typed degenerate-geometry rejection	Tri3 recovery <= 1e-12; Quad4 outputs finite; one degenerate geometry rejection
`bem.nested-trace-operator.v1`	contract-gate	independent dense forward, algebraic-transpose, Hermitian, block and selected-row actions; typed edge/row rejection	all action errors <= 1e-12; two invalid-edge and two invalid-row rejections
`bem.nested-trace-solve.v1`	contract-gate	independent nested matrix-free action round-trip and true residual admission	all solve errors <= 1e-10; two invalid-input rejections
`bem.fem-bem-trace-solve.v1`	contract-gate	independent coupled matrix action round-trip and true residual admission	all solve errors and matrix-free true residuals <= 1e-10; two invalid-input rejections
`bem.thin-shell-surface.v1`	physics-gate	analytic mass-spring-damper impedance and centered frequency derivatives	derivative errors <= 2e-6; sweep error <= 1e-6; absorption in [0,1]; three typed rejections
`bem.incident-field-admission.v1`	contract-gate	IncidentField::validate and BemProblem::validate	four finite cases accepted; four malformed cases rejected
`bem.cuda-capability-and-assembly-matrix.v1`	accelerator-gate	independent CPU f64 TBEM assembly, CPU field evaluation, Mie radiation, reciprocity, and analytic derivative	matrix/product/field parity <= 1e-3; sphere CUDA-vs-CPU <= 1e-10; Mie/radiation/power <= 3e-1; refined mesh must reduce reference error; derivative <= 5e-3; kernels and transfers nonzero
`bem.cuda-hmatrix-fmm-device-residency.v1`	accelerator-gate	independent dense stabilized TBEM action, adjoint inner-product identity, reciprocity, and field reference	all action/residual/field errors <= 1e-8; steady-state H2D operations = 0; CUDA kernels and D2H evidence nonzero
`bem.rigid-sphere-mie-accuracy-ka1.v1`	physics-gate	analytical rigid-sphere Mie pressure; no external solver execution or imported output	complex L2 error <=0.01; numerical_solve_executed=1; external_reference_executed=0
`bem.rigid-sphere-mie-accuracy-ka2.v1`	physics-gate	analytical rigid-sphere Mie pressure; no external solver execution or imported output	complex L2 error <=0.015; numerical_solve_executed=1; external_reference_executed=0
`bem.rigid-sphere-mie-accuracy-ka5.v1`	physics-gate	analytical rigid-sphere Mie pressure; no external solver execution or imported output	complex L2 error <=0.015; numerical_solve_executed=1; external_reference_executed=0