Last verified: 2026-08-02.
PhaseProbe is positioned narrowly: simulation-specific transition discovery, bounded counterexample minimization, deterministic replay, and regression-test generation. It is not a replacement for a simulator, ODE solver, sensitivity-analysis package, formal bifurcation package, property-based testing framework, or rigorous proof tool.
The PhaseProbe name is retained. Exact-name checks returned no repository named phaseprobe from the GitHub repository-search API and HTTP 404 (no registered project) from both the PyPI project JSON endpoint and npm registry on 2026-08-01. A broader web search found the descriptive term “phase probe” in accelerator instrumentation, but no material software-package collision for this developer tool. This is a practical collision audit, not a legal opinion or trademark clearance. PyPI publication remains explicitly out of scope.
Evidence: GitHub repository search, PyPI project endpoint, npm registry endpoint, and an example of the unrelated accelerator phase-probe usage.
| tool | primary job | target user | existing strength | overlap | remaining gap | PhaseProbe differentiation | last verified date |
|---|---|---|---|---|---|---|---|
| Hypothesis | Property-based test generation and shrinking | Python developers | Broad input strategies, edge-case generation, mature shrinking | Search and counterexample reduction | Does not provide a simulation trace/classification/replay artifact workflow by itself | Domain-specific bounded scans, trace evidence, replay fixtures, and generated pytest tests | 2026-08-01 |
SciPy solve_ivp |
Numerical initial-value integration | Scientific Python users | Multiple adaptive ODE solvers, dense output, events, and error controls | Produces trajectories for models | Does not classify regimes or turn discovered transitions into integrity-protected tolerance fixtures and pytest regressions | Optional trajectory adapter keeps SciPy responsible for integration while PhaseProbe owns bounded search, evidence, tolerance replay, and test materialization | 2026-08-02 |
| Mesa | Agent-based model construction, execution, visualization, and analysis | Agent-based modelers | Rich ABM components and data collection | Simulation configuration and repeated runs | No focused transition-to-replay-to-pytest pipeline in the core job | Adapter-oriented testing layer, not an ABM framework | 2026-08-01 |
| NetLogo | Programmable multi-agent modeling environment | Researchers and educators | Mature modeling language, model library, and interactive environment | Parameter sweeps and qualitative behavior exploration | Developer-facing Python fixture/test generation is not its central job | Headless local regression evidence for adapters | 2026-08-01 |
| cadCAD | Modeling complex adaptive and dynamical systems | System modelers | Structured generalized dynamical-system workflows | Parameterized simulation runs | No dedicated smallest-found transition fixture and pytest generation workflow | Testing pipeline around a small typed adapter | 2026-08-01 |
| Golly | Explore cellular automata | Cellular-automata users | Fast cross-platform CA exploration and many rule formats | Qualitative state evolution | Not a general simulation regression tool | Model-agnostic adapter and CI artifacts | 2026-08-01 |
| py-pde | Solve PDEs including diffusion and stochastic equations | PDE modelers | Field abstractions, boundary conditions, and PDE solvers | Reaction/diffusion trajectory generation | Does not center on qualitative-transition minimization and pytest materialization | Consumes bounded observables/classifiers rather than replacing PDE solvers | 2026-08-01 |
| PyDSTool / PyCont | Dynamical-system simulation, continuation, and bifurcation analysis | Dynamical-systems researchers | Integrated trajectories, continuation, phase planes, and bifurcation tools | Parameter changes and bifurcation evidence | Formal analysis is broader and deeper, while direct developer regression-fixture output is not the primary job | Explicitly reports numerical brackets rather than formal bifurcation points and produces deterministic test artifacts | 2026-08-01 |
| AUTO-07p | Continuation and bifurcation analysis for ODEs | Applied mathematicians | Established numerical continuation and bifurcation capabilities | Transition analysis | Higher setup cost and a different goal from bounded black-box regression discovery | Lightweight black-box classification and replay; never presents itself as formal continuation | 2026-08-01 |
| Attractors.jl | Find attractors, basins, basin boundaries, stability measures, and continuation | Nonlinear-dynamics researchers | Deep attractor and basin analysis | Regime and multistability exploration | Regression-test generation and Python replay fixtures are not the central workflow | Testing-focused evidence handoff with deliberately simpler analysis | 2026-08-01 |
| SALib | Global sensitivity analysis | Modelers and uncertainty-quantification practitioners | Sobol, Morris, FAST, and other established methods | Bounded input exploration and sensitivity | Sensitivity indices answer a different question from finding a reproducible class boundary | Searches a declared qualitative predicate and records the smallest case found in that search | 2026-08-01 |
| Exploratory metamorphic testing | Test relations among related scientific-simulation inputs and outputs | Scientific software teams | Addresses test-oracle difficulty using relations across executions | Baseline/perturbed twin runs and invariants | A research method, not a packaged transition/replay/pytest workflow | Implements a narrow, deterministic operational workflow without claiming to subsume metamorphic testing | 2026-08-01 |
| General record/replay tools (for example VCR) | Capture and replay external interactions | Application developers | Stable deterministic fixtures for I/O interactions | Fixture integrity and replay concepts | External-call cassettes do not re-execute a scientific model from model/config/seed/tolerances | Re-executes supported simulation adapters and verifies trace/classification hashes | 2026-08-01 |
The 0.2.0 SciPy integration does not change this boundary: SciPy integrates trajectories; PhaseProbe searches declared dimensions, preserves qualitative boundaries, and creates regressions.
PhaseProbe v0.1.0 offers bounded empirical evidence. It does not prove chaos, compute a formal Lyapunov exponent, prove global minimality, locate an exact bifurcation point, validate a scientific model against nature, or replace solver-convergence studies. Its “smallest” result always means the smallest reproducible change found by the declared finite search and refinement procedure.