No clear winner. The survival gap is under 10 percentage points and the balanced-score gap is under 0.05, so neither SolvSRK nor the best baseline clears the win threshold. Either works — choose on cost, licensing, or integration effort. All verdicts →
Coupled target-background thermal evolution for IR detection. Target states (6): T_surface, T_core, emissivity, convection, radiation_flux, contrast_signal. Background states (6): identical structure with ambient parameters. Both subsystems share solver drift characteristics. Output metric: target-background contrast ratio for TOST non-inferiority (delta=5%).
Optica (2012) thermal drift compensation for microbolometers; arXiv 2512.15211 TBC metric
Canonical RHS excerpt from the registered callable used for this benchmark cell. Expand it to verify the state equations; it is not a standalone runnable fixture.
def _ir_thermal_symmetric_rhs(t: float, y: np.ndarray) -> np.ndarray:
dy = np.zeros(12)
T_s_tgt = max(y[0], 200.0)
T_c_tgt = max(y[1], 200.0)
eps_tgt = np.clip(y[2], 0.01, 1.0)
h_tgt = max(y[3], 0.1)
q_rad_tgt = y[4]
contrast = y[5]
T_s_bg = max(y[6], 200.0)
T_c_bg = max(y[7], 200.0)
eps_bg = np.clip(y[8], 0.01, 1.0)
h_bg = max(y[9], 0.1)
q_rad_bg = y[10]
rho_c_tgt = 2.5e6
k_tgt = 15.0
L_tgt = 0.05
A_tgt = 0.1
Q_int = 500.0
q_conv_tgt = h_tgt * A_tgt * (T_s_tgt - _T_AMB)
q_rad_actual_tgt = eps_tgt * _STEFAN_BOLTZMANN * A_tgt * (T_s_tgt**4 - _T_AMB**4)
q_cond_tgt = k_tgt * A_tgt * (T_c_tgt - T_s_tgt) / L_tgt
vol_tgt = A_tgt * L_tgt
cap_surf_tgt = rho_c_tgt * vol_tgt * 0.3
cap_core_tgt = rho_c_tgt * vol_tgt * 0.7
dy[0] = (q_cond_tgt - q_conv_tgt - q_rad_actual_tgt) / cap_surf_tgt
dy[1] = (Q_int - q_cond_tgt) / cap_core_tgt
tau_eps = 2.0
eps_eq = 0.85 + 0.05 * np.sin(0.5 * t)
dy[2] = (eps_eq - eps_tgt) / tau_eps
tau_h = 1.0
h_eq_tgt = 10.0 + 2.0 * np.sin(0.3 * t)
dy[3] = (h_eq_tgt - h_tgt) / tau_h
dy[4] = (q_rad_actual_tgt - q_rad_tgt) / 0.1
rho_c_bg = 1.8e6
k_bg = 0.6
L_bg = 0.1
A_bg = 1.0
q_conv_bg = h_bg * A_bg * (T_s_bg - _T_AMB)
q_rad_actual_bg = eps_bg * _STEFAN_BOLTZMANN * A_bg * (T_s_bg**4 - _T_AMB**4)
q_cond_bg = k_bg * A_bg * (T_c_bg - T_s_bg) / L_bg
q_solar = 200.0 * max(0.0, np.sin(0.5 * t))
vol_bg = A_bg * L_bg
cap_surf_bg = rho_c_bg * vol_bg * 0.3
cap_core_bg = rho_c_bg * vol_bg * 0.7
dy[6] = (q_cond_bg + q_solar * A_bg * eps_bg - q_conv_bg - q_rad_actual_bg) / cap_surf_bg
dy[7] = (-q_cond_bg) / cap_core_bg
tau_eps_bg = 5.0
eps_eq_bg = 0.92 + 0.02 * np.cos(0.3 * t)
dy[8] = (eps_eq_bg - eps_bg) / tau_eps_bg
h_eq_bg = 5.0 + 1.0 * np.cos(0.2 * t)
dy[9] = (h_eq_bg - h_bg) / tau_h
dy[10] = (q_rad_actual_bg - q_rad_bg) / 0.1
radiance_tgt = eps_tgt * _STEFAN_BOLTZMANN * T_s_tgt**4
radiance_bg = eps_bg * _STEFAN_BOLTZMANN * T_s_bg**4
target_contrast = (radiance_tgt - radiance_bg) / max(radiance_bg, 1e-10)
tau_contrast = 0.05
dy[5] = (target_contrast - contrast) / tau_contrast
dy[11] = 0.0
return dyCanonical RHS excerpt captured from the same registered callable used for the published benchmark. Frozen closure values are summarized below; helper imports and solver settings are intentionally omitted.
Fingerprint
Spread: high
Default noise: low
Recommendation snapshot
Clean best: SolvSRK
Noisy best: SciPy LSODA
Coverage
14 solver arms · clean + 5 noise levels
Ranked on survival, precision, and speed
20 seeds/cell default · 14 arms · TRL 4–5 · simulation-lab validated · this page: IR Target/Background Symmetric Thermal (dim=12) (ir-target-background-symmetric-thermal-dim-12)
Governed SolvTune benchmark freeze; per-arm medians only. RHS definitions and raw trial rows are not published.
Self-reported by Resonix Labs · not independently verified
Pick an objective and a noise level to rank all arms on survival, median SCD, median nfev, and median wall time. Medians across seeds.
Objective
Best overall trade-off of survival, precision, and speed.
Noise level
| # | Solver | Survival | SCD | nfev | Wall | Score |
|---|---|---|---|---|---|---|
| 1 | SolvSRK | 100% | 11.5 | 1,260 | 51 ms | 0.892 |
| 2 | Vern7external | 100% | 10.7 | 1,102 | 4.05 s | 0.875 |
| 3 | SciPy RadauSciPy | 100% | 10.1 | 1,399 | 43 ms | 0.860 |
| 4 | SciPy DOP853SciPy | 100% | 10.0 | 722 | 12 ms | 0.857 |
| 5 | SciPy RK45SciPy | 100% | 9.9 | 932 | 17 ms | 0.855 |
| 6 | Vern9external | 100% | 9.5 | 1,730 | 4.05 s | 0.844 |
| 7 | SciPy LSODASciPy | 100% | 9.3 | 651 | 9 ms | 0.840 |
| 8 | CVODE BDFexternal | 100% | 9.1 | 346 | 24 ms | 0.836 |
| 9 | SciPy BDFSciPy | 100% | 8.6 | 628 | 26 ms | 0.825 |
| 10 | Tsit5external | 100% | 8.5 | 804 | 757 ms | 0.821 |
| 11 | CVODE Adamsexternal | 100% | 8.3 | 299 | 21 ms | 0.818 |
| 12 | SciPy RK23SciPy | 100% | 8.2 | 1,553 | 30 ms | 0.813 |
| 13 | FBDFexternal | 100% | 7.8 | 441 | 5.31 s | 0.806 |
| 14 | TRBDF2external | 100% | 6.4 | 1,005 | 5.22 s | 0.771 |
At Clean, best balanced arm is SolvSRK.
Values are medians across seeds, measured by Resonix Labs on Resonix hardware and not independently verified; nfev and wall are on reference lab hardware (indicative). Under injected noise only SolvSRK and the SciPy arms are run. How we measure accuracy → · Verification status →
SolvScout · free
This page shows one published benchmark cell. SolvScout fingerprints your ODE, compares it to the full corpus, and recommends a solver with the same survival / precision / speed ranking you see here — including when a SciPy arm wins.
SolvSRK · 30-day trial
SolvSRK is the stiffness-adaptive integrator behind the SolvSRK column in these tables. Create an account, activate a machine, and take a 30-day trial — same binary you'd ship after purchase.
Replace the access date. Pin the freeze ID and library versions when comparing against a later export. Cite it as what it is — a self-reported vendor benchmark, not an independently verified result. The note field says so; please keep it.
@misc{resonix_evidence_ir_target_background_symmetric_thermal_dim_12_2026,
title = {Resonix Evidence Portal: IR Target/Background Symmetric Thermal (dim=12)},
author = {{Resonix Labs (Canada) Inc.}},
year = {2026},
howpublished = {\url{https://resonix.tech/evidence/problems/ir-target-background-symmetric-thermal-dim-12}},
note = {Self-reported vendor benchmark; internally generated by Resonix Labs and not independently verified. Accessed YYYY-MM-DD. Freeze 2026-08-13; libsolvsrk 2.3.0; SciPy 1.14.}
}TRL 4–5 · simulation-lab validated · 398 problems · 14 solver arms · clean + 5 noise levels
Freeze: 2026-08-13 · scipy 1.14 · libsolvsrk 2.3.0 · Methodology
Self-reported by Resonix Labs · not independently verified · Verification status