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 →
7x7x7 3D glass-fiber front propagation (dim=686, S3)
Kamal & Sourour (1973); Pojman (2012)
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 _idx3d(i, j, k):
return i * _GF3_NY * _GF3_NZ + j * _GF3_NZ + k
def _kamal_sourour_rate_vec(T: np.ndarray, alpha: np.ndarray) -> np.ndarray:
T_safe = np.clip(T, _T_FLOOR, _T_CEIL)
alpha_safe = np.clip(alpha, 0.0, 1.0)
inv_RT = 1.0 / (_R_GAS * T_safe)
arg1 = np.clip(_E1 * inv_RT, 0.0, _EXP_ARG_MAX)
arg2 = np.clip(_E2 * inv_RT, 0.0, _EXP_ARG_MAX)
k1 = _A1 * np.exp(-arg1)
k2 = _A2 * np.exp(-arg2)
return (k1 + k2 * np.power(alpha_safe, _M)) * np.power(1.0 - alpha_safe, _N_ORD)
def _fp_glass_fiber_3d_rhs(t, y):
T_flat = np.clip(y[:_GF3_N3D], _T_FLOOR, _T_CEIL)
alpha_flat = np.clip(y[_GF3_N3D:], 0.0, 1.0)
dadt = _kamal_sourour_rate_vec(T_flat, alpha_flat)
dT = np.empty(_GF3_N3D)
for i in range(_GF3_NX):
for j in range(_GF3_NY):
for k in range(_GF3_NZ):
idx = _idx3d(i, j, k)
T_c = T_flat[idx]
# x-direction (fiber): Dirichlet at i=0, convective at i=NX-1
if i == 0:
lap_x = 0.0
elif i == _GF3_NX - 1:
T_ghost = T_c + (_H_CONV * _GF3_DX / _K_FIBER_G) * (_T_AMBIENT - T_c)
lap_x = (T_flat[_idx3d(i - 1, j, k)] - 2.0 * T_c + T_ghost) * _GF3_INV_DX2
else:
T_left = _GF3_T_LEFT if i == 1 else T_flat[_idx3d(i - 1, j, k)]
lap_x = (T_left - 2.0 * T_c + T_flat[_idx3d(i + 1, j, k)]) * _GF3_INV_DX2
# y-direction (transverse)
if j == 0:
T_ghost = T_c + (_H_CONV * _GF3_DY / _K_TRANS_G) * (_T_AMBIENT - T_c)
lap_y = (T_ghost - 2.0 * T_c + T_flat[_idx3d(i, j + 1, k)]) * _GF3_INV_DY2
elif j == _GF3_NY - 1:
T_ghost = T_c + (_H_CONV * _GF3_DY / _K_TRANS_G) * (_T_AMBIENT - T_c)
lap_y = (T_flat[_idx3d(i, j - 1, k)] - 2.0 * T_c + T_ghost) * _GF3_INV_DY2
else:
lap_y = (T_flat[_idx3d(i, j - 1, k)] - 2.0 * T_c + T_flat[_idx3d(i, j + 1, k)]) * _GF3_INV_DY2
# z-direction (transverse)
if k == 0:
T_ghost = T_c + (_H_CONV * _GF3_DZ / _K_TRANS_G) * (_T_AMBIENT - T_c)
lap_z = (T_ghost - 2.0 * T_c + T_flat[_idx3d(i, j, k + 1)]) * _GF3_INV_DZ2
elif k == _GF3_NZ - 1:
T_ghost = T_c + (_H_CONV * _GF3_DZ / _K_TRANS_G) * (_T_AMBIENT - T_c)
lap_z = (T_flat[_idx3d(i, j, k - 1)] - 2.0 * T_c + T_ghost) * _GF3_INV_DZ2
else:
lap_z = (T_flat[_idx3d(i, j, k - 1)] - 2.0 * T_c + T_flat[_idx3d(i, j, k + 1)]) * _GF3_INV_DZ2
dT[idx] = (_GF3_DIFF_X * lap_x + _GF3_DIFF_Y * lap_y
+ _GF3_DIFF_Z * lap_z + _SRC_COEFF * dadt[idx])
# Left face (i=0) is Dirichlet
for j in range(_GF3_NY):
for k in range(_GF3_NZ):
dT[_idx3d(0, j, k)] = 0.0
dy = np.empty(_GF3_DIM)
dy[:_GF3_N3D] = dT
dy[_GF3_N3D:] = dadt
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: extreme
Default noise: low
Recommendation snapshot
Clean best: SolvSRK
Noisy best: SciPy BDF
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: FP Glass Fiber 3D (fp-glass-fiber-3d)
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% | 12.0 | 2,449 | 1.93 s | 0.906 |
| 2 | SciPy RadauSciPy | 100% | 10.3 | 26,993 | 24.11 s | 0.866 |
| 3 | SciPy LSODASciPy | 100% | 8.4 | 4,365 | 3.59 s | 0.819 |
| 4 | CVODE Adamsexternal | 100% | 8.3 | 8,281 | 7.62 s | 0.817 |
| 5 | SciPy DOP853SciPy | 100% | 8.1 | 1,922 | 1.55 s | 0.811 |
| 6 | SciPy RK45SciPy | 100% | 7.9 | 2,018 | 1.63 s | 0.808 |
| 7 | Tsit5external | 100% | 7.9 | 2,064 | 2.78 s | 0.808 |
| 8 | CVODE BDFexternal | 100% | 7.9 | 6,699 | 5.90 s | 0.806 |
| 9 | SciPy BDFSciPy | 100% | 7.7 | 2,526 | 2.72 s | 0.802 |
| 10 | SciPy RK23SciPy | 100% | 7.3 | 3,335 | 2.71 s | 0.793 |
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_fp_glass_fiber_3d_2026,
title = {Resonix Evidence Portal: FP Glass Fiber 3D},
author = {{Resonix Labs (Canada) Inc.}},
year = {2026},
howpublished = {\url{https://resonix.tech/evidence/problems/fp-glass-fiber-3d}},
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