SolvSRK wins. At the comparison noise level, SolvSRK beats the best baseline by at least 10 percentage points of survival, or by at least 0.05 balanced score when survival is tied. Use SolvSRK for this class of problem. All verdicts →
3-interceptor vs 3-target cooperative PN guidance with deconfliction. Per-vehicle state [x,y,theta] (dim=18). Target maneuver: straight-line. d_miss=2m terminal cutoff. Inter-interceptor repulsive potential at d_safe=50m.
Zarchan, 'Tactical and Strategic Missile Guidance', Ch. 8; Shima & Rasmussen, 'Cooperative Interceptor Guidance'
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 rhs(t: float, y: np.ndarray) -> np.ndarray:
d = np.zeros(dim)
positions = np.empty((N_total, 2))
for k in range(N_total):
bk = k * _STATES_PER_VEHICLE
positions[k, 0] = y[bk + 0]
positions[k, 1] = y[bk + 1]
for i in range(N_interceptors):
bi = i * _STATES_PER_VEHICLE
x_m, y_m, th_m = y[bi + 0], y[bi + 1], y[bi + 2]
vx_m = V_m * np.cos(th_m)
vy_m = V_m * np.sin(th_m)
d[bi + 0] = vx_m
d[bi + 1] = vy_m
tgt_idx = N_interceptors + (i % N_targets)
bt = tgt_idx * _STATES_PER_VEHICLE
x_t, y_t, th_t = y[bt + 0], y[bt + 1], y[bt + 2]
dx = x_t - x_m
dy = y_t - y_m
r_sq = dx * dx + dy * dy
if r_sq > d_miss_sq:
r = np.sqrt(r_sq) + _EPS_RANGE
vx_t = V_t * np.cos(th_t)
vy_t = V_t * np.sin(th_t)
dlam_dt = (dx * (vy_t - vy_m) - dy * (vx_t - vx_m)) / (r * r)
V_c = -(dx * (vx_t - vx_m) + dy * (vy_t - vy_m)) / r
a_pn = N_pn * V_c * dlam_dt
else:
a_pn = 0.0
a_deconf_x = 0.0
a_deconf_y = 0.0
for j in range(N_interceptors):
if j == i:
continue
bj = j * _STATES_PER_VEHICLE
dxij = y[bi + 0] - y[bj + 0]
dyij = y[bi + 1] - y[bj + 1]
dist_sq = dxij * dxij + dyij * dyij
if dist_sq < d_gate_sq:
dist = np.sqrt(dist_sq) + _EPS_RANGE
separation = max(dist - d_safe, 0.1)
repulsion = 50.0 / (separation * separation)
repulsion = min(repulsion, 200.0)
a_deconf_x += repulsion * dxij / dist
a_deconf_y += repulsion * dyij / dist
a_total = a_pn + (a_deconf_x * np.sin(th_m) - a_deconf_y * np.cos(th_m))
d[bi + 2] = np.clip(a_total, -50.0, 50.0) / V_m
for j in range(N_targets):
bj = (N_interceptors + j) * _STATES_PER_VEHICLE
th_t_j = y[bj + 2]
d[bj + 0] = V_t * np.cos(th_t_j)
d[bj + 1] = V_t * np.sin(th_t_j)
d[bj + 2] = omega_t
return dCanonical 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: low
Default noise: medium
Recommendation snapshot
Clean best: SciPy RK45
Noisy best: SolvSRK
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: Cooperative Interceptor 3v3 (straight-line) (cooperative-interceptor-3v3-straight-line)
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 | SciPy RK45SciPy | 100% | 8.5 | 320 | 10 ms | 0.822 |
| 2 | SciPy RadauSciPy | 100% | 8.5 | 1,298 | 52 ms | 0.821 |
| 3 | SolvSRK | 100% | 8.0 | 589 | 11 ms | 0.810 |
| 4 | SciPy DOP853SciPy | 100% | 7.4 | 398 | 12 ms | 0.796 |
| 5 | SciPy LSODASciPy | 100% | 7.0 | 417 | 11 ms | 0.786 |
| 6 | SciPy RK23SciPy | 100% | 6.7 | 422 | 14 ms | 0.778 |
| 7 | SciPy BDFSciPy | 100% | 6.7 | 664 | 27 ms | 0.778 |
| 8 | Tsit5external | 100% | 6.5 | 354 | 845 ms | 0.775 |
| 9 | CVODE Adamsexternal | 100% | 6.5 | 252 | 18 ms | 0.773 |
| 10 | CVODE BDFexternal | 100% | 6.3 | 351 | 18 ms | 0.770 |
At Clean, best balanced arm is SciPy RK45 · SolvSRK survival 100%, SCD 8.0.
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_cooperative_interceptor_3v3_straight_line_2026,
title = {Resonix Evidence Portal: Cooperative Interceptor 3v3 (straight-line)},
author = {{Resonix Labs (Canada) Inc.}},
year = {2026},
howpublished = {\url{https://resonix.tech/evidence/problems/cooperative-interceptor-3v3-straight-line}},
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