Electromagnetics validation
Each check compares a RunMat result with an independent reference: a closed-form solution, a manufactured solution or a published benchmark. The tests listed under a check run on every change and in the evidence report; receipts record a public study run with its build, commit and mesh. The validation methodology explains how checks are built and how tolerances are set, and Status summarizes every family.
Electrostatics and steady current
Homogeneous slab capacitor and conductor
Problem. A homogeneous slab between two terminals, solved as a capacitor (electrostatics) and as a conductor (steady current) on Tet4 and Tet10 meshes.
Reference. Parallel-plate closed form: linear potential, uniform field, terminal charge or current, stored energy and Joule power.
Compared. Nodal potential, field, terminal flux, energy, Joule power.
Tolerance. Potential and field within 1e-8; flux imbalance and energy identity below 1e-8; energy and displacement field within relative 1e-8.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: homogeneous_capacitor_and_conductor_match_independent_slab_field_terminals_and_energy.
Two-material slab in series
Problem. A slab of two aligned materials between two terminals, on Tet4 and Tet10 meshes.
Reference. Series capacitance and series resistance, with continuous normal flux across the interface.
Compared. Potential, energy, Joule power, interface flux.
Tolerance. Relative 1e-8.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: layered_materials_match_series_capacitance_and_resistance_with_interface_flux_continuity.
Harmonic polynomial potential
Problem. The harmonic potential , solved with both the electrostatic and the conduction equation; Tet10 on one mesh and Tet4 on three.
Reference. The exact field and energy of the harmonic polynomial.
Compared. Field and energy.
Tolerance. Tet10: field and energy exact to relative 1e-8. Tet4: field error falls below 0.6 of the previous mesh at each refinement.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: harmonic_quadratic_potential_has_tet10_exact_field_and_tet4_convergence.
Uniformly charged dielectric slab
Problem. A dielectric slab with uniform volume charge on three Tet4 meshes.
Reference. Closed form: quadratic potential, linear field, conserved charge and energy in the fixture's units.
Compared. Potential, field and energy errors; flux imbalance.
Tolerance. Per refinement: potential error order above 1.7, field order above 0.8, energy error below 0.35 of the previous mesh. Finest mesh: potential below 0.005, field below 0.1, energy below 0.025. Flux imbalance below 1e-9.
Result. Passed (1 test) in the evidence run of 2026-11-08. Potential error 4.6e-2, 1.1e-2, 2.9e-3; field error 0.29, 0.14, 0.072; energy error 0.25, 0.063, 0.016.
Tests: tet4_charged_slab_has_independent_spatial_field_and_energy_convergence.
Steady current in a copper bar
Problem. A 1 m copper bar ( = 59.6 MS/m) with 1 mV across its ends; Tet10.
Reference. Uniform current density and Joule power .
Compared. Current density; Joule power from the loss density and from the terminal current.
Tolerance. Relative 1e-9.
Result. Passed (1 test) in the evidence run of 2026-11-08. Peak current density 59 600.000 A/m² against 59 600 A/m² (relative difference 1e-13); Joule power 0.596000 W against 0.596 W (1e-14).
Tests: copper_bar_carries_sigma_v_over_l_and_dissipates_v_squared_over_r.
Magnetostatics
Magnetic field of a square conductor
Problem. A square conductor carries a uniform axial current density and is grounded on every exterior face; Tet4 and Tet10 geometry with the lowest-order edge basis, three meshes per order.
Reference. An independent Fourier-series solution for the vector potential, flux density and energy; the series tail is bounded by at N = 255 terms.
Compared. Vector potential, flux density and magnetic energy.
Tolerance. Each error falls by more than 1.4× per refinement. Finest mesh: potential below 0.02, flux density below 0.10, energy below 0.25. Physical and gauge residuals below 8e-10.
Result. Passed (1 test) in the evidence run of 2026-11-08. Flux-density error 0.166, 0.116, 0.064 and energy error 0.80, 0.39, 0.12 on 1, 2 and 4 cells.
Tests: square_uniform_current_has_independent_potential_field_and_energy_convergence.
Harmonic
Skin effect in a driven conductor
Problem. A conducting strip ( = 1 mm, = 2 mm, = 1 mm, = 1 MS/m) under an air gap = 0.5 mm, returned through a grounded plane, is driven by a 2 A terminal current at = 0.5, 2 and 4 with 8, 16 and 32 cells through the thickness and along the current.
Reference. Parallel-plate line impedance with and . (S. Ramo, J. R. Whinnery and T. Van Duzer, Fields and Waves in Communication Electronics, 3rd ed., Wiley, 1994.)
Compared. Terminal impedance ; Joule loss against delivered power; Poynting balance and charge continuity.
Tolerance. Impedance error below 2e-3 at 32 cells and reduced by more than 3× from 8 to 32 cells; Joule loss equals within 1e-8; Poynting and charge-continuity residuals below 1e-8.
Result. Passed (1 test) in the evidence run of 2026-11-08. Relative impedance error at 8, 16 and 32 cells: 4.2e-4, 1.0e-4, 2.6e-5 at = 0.5; 7.3e-3, 1.8e-3, 4.5e-4 at 2; 1.6e-2, 3.9e-3, 9.8e-4 at 4. The error falls fourfold per refinement.
Tests: current_driven_strip_impedance_converges_to_the_skin_effect_solution.
Low-frequency limit of a driven strip
Problem. The strip of the skin-effect check, voltage driven at 1 Hz.
Reference. DC resistance and the low-frequency gap inductance.
Compared. Real and imaginary parts of the terminal impedance.
Tolerance. Resistance within relative 1e-6; reactance within relative 2e-2.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: voltage_drive_and_low_frequency_limit_match_dc_resistance_and_inductance.
Transmission-line parameters
Microstrip impedance against Hammerstad–Jensen
Problem. A zero-thickness surface microstrip on a 0.2 mm substrate over a reference plane, open above, with w/h = 0.5, 1 and 2 on εr = 4.4, w/h = 4 on εr = 2.2 and w/h = 1 on εr = 9.8, on the default graded mesh.
Reference. Hammerstad and Jensen's closed forms for Z0 and εeff of a zero-thickness microstrip; the authors state Z0 in air within 0.01 % (w/h ≤ 1) and 0.03 % (w/h ≤ 1000) and εeff within 0.2 % (εr ≤ 128, 0.01 ≤ w/h ≤ 100) of exact quasi-static values. E. Hammerstad and Ø. Jensen, "Accurate models for microstrip computer-aided design," IEEE MTT-S International Microwave Symposium Digest, 1980, pp. 407–409.
Compared. Z0, εeff and delay per length of the line.
Tolerance. 0.5 % on Z0 and εeff; the delay follows from εeff.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: microstrip_matches_hammerstad_jensen.
Stripline and coupled stripline against Cohn's exact forms
Problem. Zero-thickness strips centred between planes 0.5 mm apart in εr = 3.0: single strips with w/b = 0.2, 0.5 and 1.0, and edge-coupled pairs with (w/b, s/b) = (0.5, 0.2), (0.5, 0.5) and (0.3, 1.0).
Reference. Cohn's conformal-mapping solutions, exact for zero-thickness strips between infinite planes: Z0 = (30π/√εr) K(k)/K(k′) with k = sech(πw/2b), and the even- and odd-mode impedances with k_e = tanh(πw/2b) tanh(π(w+s)/2b), k_o = tanh(πw/2b) coth(π(w+s)/2b). S. B. Cohn, "Characteristic impedance of the shielded-strip transmission line," IRE Trans. Microwave Theory Tech., vol. MTT-2, pp. 52–57, 1954; S. B. Cohn, "Shielded coupled-strip transmission line," IRE Trans. Microwave Theory Tech., vol. MTT-3, pp. 29–38, 1955.
Compared. Z0 of single strips; even- and odd-mode impedances of pairs, with Zdiff = 2 Zodd, Zcommon = Zeven / 2, εeff = εr and zero far-end coupling in the homogeneous dielectric.
Tolerance. 0.3 % on each impedance; the identities hold to 1e-9.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: stripline_matches_cohn_exact_forms.
Microstrip and stripline against IPC-2141A
Problem. 35 µm copper. Surface microstrips (w, h, εr) = (0.35 mm, 0.2 mm, 4.1), (0.2 mm, 0.15 mm, 4.3) and (0.15 mm, 0.25 mm, 3.5); centred striplines (w, b, εr) = (0.15 mm, 0.6 mm, 4.2), (0.1 mm, 0.5 mm, 3.6) and (0.2 mm, 0.8 mm, 4.4). Every case is inside the forms' stated ranges.
Reference. IPC-2141A surface microstrip Z0 = 87/√(εr + 1.41) · ln(5.98h/(0.8w + t)) for 0.1 < w/h < 2.0 and 1 < εr < 15, and symmetric stripline Z0 = 60/√εr · ln(1.9b/(0.8w + t)) for w/b < 0.35 and t/b < 0.25. IPC-2141A, Design Guide for High-Speed Controlled Impedance Circuit Boards, IPC, 2004.
Compared. Z0 of each line.
Tolerance. 5 %: the IPC-2141A forms are empirical fits, so this check bounds agreement with the board-industry forms; Hammerstad–Jensen and Cohn are the tight references.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: microstrip_and_stripline_agree_with_ipc_2141a_inside_their_ranges.
Mesh and enclosure convergence of Z0
Problem. A 0.35 mm, 35 µm-thick surface microstrip on 0.2 mm of εr = 4.1, meshed with edge sizes of 7, 3.5, 1.75 and 0.875 µm, then with the default mesh in an enclosure twice the default width and height.
Reference. Self-convergence: the change in Z0 between successive refinements, and the change from doubling the enclosure.
Compared. Z0.
Tolerance. Successive changes decrease and the last is below 0.1 %; the enclosure changes Z0 by under 0.05 %.
Result. Passed (1 test) in the evidence run of 2026-11-08.
Tests: microstrip_impedance_converges_with_the_mesh_and_enclosure.