ssnreduce 0.1.0

Faithful Python port of Joel Mortimer's supersonic-nozzle nucleation data-reduction pipeline (Ar+CO2 mixture; isentropic -> diabatic ODE -> nucleation -> growth).

ssnreduce

A faithful Python port of Joel Mortimer's supersonic-nozzle (SSN) nucleation data-reduction pipeline (the MATLAB analyseSSN_mix5 / diabaticEquations / compareSSN chain), built to analyse and plot the lab's future Ar + CO₂ runs.

It deliberately reuses what already exists in D:/PhD/:

• postprocessing/ — the proven rig-CSV loader and settled-station extractor (load_run, extract_steady_stations). ssnreduce.io wraps these.
• co2ssn/ — the pure-CO₂ physics library. Kept as an independent cross-check for the nucleation rates (different EOS/σ choices), not a dependency of the reduction.

The new code is the bridge: real rig data → Joel's mixture-aware three-stage reduction → Joel's plots.

Status (v0.1)

Stage	Module	State
Front-end: rig CSV → run scalars + settled stations	io.py	✅ done, tested
Mixture thermo (Ar+CO₂, CoolProp-anchored, sub-triple clamp)	mixture.py	✅ done, tested
Stage 1 isentropic: p/p₀ → M, T, A/A*, choked throat	isentropic.py	✅ done, validated on real data
Stage 2 diabatic ODE: → T, ρ, g, q	diabatic.py	✅ core done, dry-isentrope self-test; ⚠ needs z-calibration + dry trace for real runs
Stage 3 nucleation: CNT/Courtney/Girshick-Chiu/Wölk/Hale + MKNT	nucleation.py	✅ done, tested (internal invariants)
Stage 4 growth: Gyarmathy 1963 + Ostwald-Freundlich	growth.py	✅ done, tested
Property DB (getGases), case metadata (getCases)	gases.py, cases.py	✅ done, tested
Plots, pipeline	plots.py, pipeline.py	✅ Stage 1 figures + J(z) overlay

33 pytest invariants green. Publishing roadmap in PUBLISHING.md (MIT license + CITATION.cff added; landscape scan found no existing CNT/MKNT Python package — a genuine gap).

Validated on 260601_..._4 percent CO2_long.csv: p/p₀ 0.104→0.574, M up to 2.10, T down to 87.6 K, discharge coefficient CD ≈ 0.90 — consistent with the postprocessing report.

Quick start

import ssnreduce as ssn

# one call: load -> settled stations -> Stage 1 -> figures
res = ssn.reduce_run("260601_..._4 percent CO2_long.csv", figdir="figures")
print(res.run.p0_bar, res.run.T0_K, res.run.x_condensable)
iso = res.isentropic            # M, T, A/A*, throat properties
print(iso.M.max(), iso.T.min(), iso.p_throat_p0)

Run the worked example and the tests:

python examples/01_stage1_isentropic.py
pytest -q

Known gaps to close with Joel (apparatus knowledge)

1. position(%) → z(mm) calibration — the rig logs probe position in % of stage stroke; the diabatic ODE integrates in mm from the throat. Need the stage travel-per-% and the throat % reading. Wired in diabatic.calibrate_z.
2. Dry companion trace — Stage 2 needs the effective A/A*(z) from a dry run paired with each wet run (Joel's dryTrace pairing).
3. linearStage_steady flag is still all-zero in the logs (station detection falls back to position dwell — already handled).

See plan.md for the full design contract and the next-increment plan.

Faithfulness

Physics equations and numerical choices follow Joel's MATLAB exactly (Wyslouzil 2000 Eqs 5/10/11/12/17, Streletzky 2002 Eqs 1/2/3, the choked-flow throat relations, the MKNT cubic to come). The software is rewritten cleanly (dataclass returns, no evalin/globals/persistent). Where RefProp was used, CoolProp replaces it with Joel's own sub-triple fallback (clamp to the carrier triple temperature). The goal is to reproduce his g(x), J(x) to numerical tolerance — tests/ enforces the invariants we can check without his outputs.