Skip to content

SPDX-License-Identifier: AGPL-3.0-or-later

Commercial license available

© Concepts 1996–2026 Miroslav Šotek. All rights reserved.

© Code 2020–2026 Miroslav Šotek. All rights reserved.

ORCID: 0009-0009-3560-0851

Contact: www.anulum.li | protoscience@anulum.li

scpn-quantum-control — Kuramoto competitive benchmark

Kuramoto Competitive Benchmark

This page documents the external competitive harness that measures our Kuramoto toolkit against real third-party solvers on one deterministic Kuramoto forward problem. It is the cross-package counterpart to the in-repository Kuramoto Tier Benchmark: the tier benchmark compares our own Rust/Julia/Python backends against each other, while this harness compares our toolkit against independent external libraries.

The harness lives in scpn_quantum_control.benchmarks.kuramoto_competitive_benchmark; the runner is scripts/bench_kuramoto_competitive.py; the committed artefact is docs/benchmarks/kuramoto_competitive.json.

What is compared

All solvers integrate the identical networked Kuramoto field

\[ \dot\theta_i = \omega_i + \sum_j K_{ij}\sin(\theta_j - \theta_i) \]

on one deterministic, seed-built problem, and report the final order parameter r, its absolute error against the high-precision reference, and the wall-clock time.

Method Family Backend Status
ours_rk4 ours kuramoto.kuramoto_rk4_trajectory (fixed step) live
ours_dopri ours kuramoto.kuramoto_dopri_trajectory (adaptive DOPRI5) live
scipy_solve_ivp external SciPy solve_ivp(RK45), tight tolerances live (reference)
julia_diffeq external Julia DifferentialEquations.jl(Tsit5) live
networkdynamics_jl external NetworkDynamics.jl declared target
dynamicalsystems_jl external DynamicalSystems.jl declared target
scimlsensitivity_jl external SciMLSensitivity.jl (differentiable competitor) declared target
jitcdde external jitcdde (just-in-time C) declared target

The reference is the SciPy high-precision run (rtol=1e-10, atol=1e-12) when SciPy is installed, otherwise our adaptive DOPRI5.

Fail-closed contract

Every external competitor is probed before it is run. A solver that is not installed — or whose subprocess errors or times out — yields an available=False row that carries the documented install command and the reason, never a fabricated number. The four declared targets above are recorded as such on a host without them, so the artefact is complete and reproducible everywhere and the rows flip to live once the package is added.

Measured comparison (committed artefact)

The committed artefact was generated on the development workstation (11th-gen Intel Core i5-11600K). Correctness errors are host-independent; the timings were captured on a non-isolated, heavily loaded host (powersave governor, load average ≈ 14.7), so per the claim boundary below they are functional and reproducibility evidence, not an isolated performance claim. For clean timing, run the runner on a quiesced, core-reserved host.

Problem: n = 12, t_max = 6.0, dt = 0.01, seed 20260628; reference scipy_solve_ivp.

Method Available r_final Error vs reference Time (ms)
ours_rk4 yes 0.77562419 2.96e-11 3.036
ours_dopri yes 0.77562558 1.39e-06 2.326
scipy_solve_ivp yes 0.77562419 — (reference) 19.627
julia_diffeq yes 0.77562419 1.76e-09 8.689
networkdynamics_jl no
dynamicalsystems_jl no
scimlsensitivity_jl no
jitcdde no

Reading: our integrators, SciPy solve_ivp, and Julia DifferentialEquations.jl agree on the final order parameter to between 3e-11 and 2e-9 — three independent implementations corroborate the result, which is the primary, host-independent claim. On this loaded host our adaptive DOPRI5 was also the fastest available solver, but that ordering must be re-measured on an isolated host before it is quoted as a performance result.

Reproduce

python scripts/bench_kuramoto_competitive.py            # default n=12 problem
python scripts/bench_kuramoto_competitive.py --n 32 --t-max 10

To bring a declared target live, install it (the artefact prints the exact command, e.g. julia -e 'using Pkg; Pkg.add("NetworkDynamics")') and re-run.

Claim boundary

  • The order-parameter values and their cross-implementation agreement are the reproducible, host-independent quantities.
  • Timings are functional and reproducibility evidence on the recorded host, not a production-latency, SLA, or universal-hardware claim. Competitor package versions and numerical tolerances are recorded in the artefact.
  • The verdict reports honestly where a competitor is faster than our toolkit; absent competitors do not contribute to it.