Physicist review index

This is a reproducible verification companion to release r4, not an experimental validation of a spacecraft. The specialist notebooks expose equations, inputs, source, residuals and uncertainty. They distinguish mathematical recomputation, bookkeeping reconciliation and incomplete physical/numerical claims.

Read this first: the tabulated-ice point-nadir estimator remains unconverged. The raw ice table is not distributed. A saved response can be reconciled without reproducing the physical scattering experiment. A pulse-shape classifier is not a validated source-depth classifier.

from pathlib import Path
import sys, json, hashlib, math, platform
ROOT = next(p for p in (Path.cwd(), *Path.cwd().parents) if (p/'modules/common.py').is_file())
sys.path.insert(0, str(ROOT/'modules'))
from common import H, C, PHOTON_ENERGY_337, PHOTON_ENERGY_777, R_PHOTONS_PER_R_M2_SR_S, table
import numpy as np
import scipy
print('Python', platform.python_version(), 'NumPy', np.__version__, 'SciPy', scipy.__version__)
print('All package paths resolve relative to', ROOT)

Python 3.12.14 NumPy 2.3.5 SciPy 1.18.1
All package paths resolve relative to D:\Photometer_research_20261022\verification_20261022_r1

1. Check the copied reference inputs

SHA256 detects changes to the reference artefacts. It does not validate the physics or prove source licensing. Missing original physical inputs are separately declared in the manifest.

manifest = json.loads((ROOT/'input_manifest.json').read_text(encoding='utf-8'))
integrity = []
for item in manifest['files']:
    path = ROOT/item['package_path']
    actual = hashlib.sha256(path.read_bytes()).hexdigest()
    integrity.append({'file':item['package_path'], 'bytes':path.stat().st_size, 'hash_matches':actual == item['sha256']})
assert all(row['hash_matches'] for row in integrity)
print(len(integrity), 'pinned copied inputs match')
table(integrity, max_rows=8)
print('Not bundled:', '; '.join(manifest['excluded_inputs']))

233

 pinned copied inputs match

Not bundled: Raw GHM ice LUT: redistribution permission not established; Restricted RAIDS/GLIMS observations; Proprietary observed waveform data

2. Inspect a complete SI worked example

The input is an assumed aperture-plane source fluence. Atmospheric propagation has already occurred. The continuum is an illustrative processed ground component, not a total polar night background.

E_\gamma=hc/\lambda,\quad S=\eta A\Phi,\quad G=A\pi\sin^2(\Theta/2).

The admitted rayleigh brightness is already band-integrated. Do not multiply that contribution by the optical bandwidth again.

p = json.loads((ROOT/'parameters/worked_example.json').read_text(encoding='utf-8'))
A = math.pi*p['aperture_diameter_m']**2/4
G = A*math.pi*math.sin(p['full_circular_fov_rad']/2)**2
energy = H*C/p['wavelength_m']
source_PE = p['end_to_end_primary_efficiency']*A*p['received_source_fluence_ph_m2']
continuum_cps = p['end_to_end_primary_efficiency']*G*p['continuum_W_m2_sr_nm']*p['equivalent_rectangular_bandwidth_nm']/energy
aurora_cps = p['end_to_end_primary_efficiency']*G*p['admitted_auroral_brightness_R']*p['projected_aurora_filling']*R_PHOTONS_PER_R_M2_SR_S
background_PE = (continuum_cps+aurora_cps+p['dark_primary_rate_s_inverse'])*p['on_window_s']
rows = [{'quantity':'photon energy337.1nm','value':energy,'unit':'J'}, {'quantity':'entrance area','value':A,'unit':'m^2'}, {'quantity':'projected etendue','value':G,'unit':'m^2 sr'}, {'quantity':'expected source count','value':source_PE,'unit':'primary PE'}, {'quantity':'continuum count rate','value':continuum_cps,'unit':'s^-1'}, {'quantity':'on-window background expectation','value':background_PE,'unit':'counts'}]
assert source_PE > 0 and background_PE > 0
assert abs(energy-PHOTON_ENERGY_337)<1e-32
table(rows)

3. Follow a calculation or original question

The map identifies source functions/lines and methods. Its status distinguishes executed formulas from blocked physical predictions and retained legacy calculations. Reading a source implementation and independently verifying it are different actions.

mapping = json.loads((ROOT/'calculation_map.json').read_text(encoding='utf-8'))
print('Crosswalk structure:', list(mapping))
coverage = json.loads((ROOT/'question_coverage.json').read_text(encoding='utf-8'))
table([{'question':q['original_label'], 'scope':q['question'], 'notebooks':q['notebooks'], 'remaining_gap_count':len(q['remaining_gaps'])} for q in coverage], max_rows=22)

Crosswalk structure: ['schema_version', 'generated_utc', 'title', 'source_manifest', 'calculation_count', 'calculations', 'plots', 'retained_legacy', 'warnings', 'path_convention', 'function_line_policy', 'source_snapshots']

4. Review in this order

1. 01_units_budget_optics: photon units, etendue, background/count conversion, filter assumptions, electronics and optical algebra.
2. 02_transport: RTE, phase normalization, direct light, next-event estimator, native batch reconstruction, timing convolution and rare-path convergence controls.
3. 03_statistics: likelihood derivation, worked fitted counts, exact on/off tests, red upper limits and all saved confusion denominators.
4. 04_analytic_diagnostics: molecular column, Lambert reflection, diffusion ambiguity, deadtime and FOV population proxies.

Each specialist identifies what it does not independently recompute. A small residual or a passed assertion is not evidence that the hypothetical source, cloud phase, background or detector describes the mission.

5. Full jobs and missing physical inputs

The retained source code and explicit full-run commands are available under reference and scripts. Full Monte Carlo/pulse-grid jobs are opt-in. They must use new output folders and do not run when opening this notebook. Input source hashes and the raw-ice acquisition/extraction recipe are provided; raw licensed/restricted inputs are excluded.

No software check establishes the target fluence population, source-depth ground truth, actual polar/Kp/lunar radiance distribution, selected-device radiation lifetime, minimum thermal777/337ratio, or photon-to-discharge-current relation.