# Supernova remnants independent scientific reader v1

This package runs independent Python scientific producers. It contains no interactive Lab, Engine, Worker, event data, original paper, observed table or observation-derived image. The default replay uses **zero network** and recalculates every numerical step. Existing NumPy, SciPy and Matplotlib are prerequisites; this package does not install them or tools.

```sh
python3 replay.py --output /absolute/new/empty/run
```

The five steps integrate the n7/n12 ejecta-envelope Euler shells at two tolerances, integrate the separate zero-ejecta Sedov–Taylor benchmark at three central cuts, calculate five passive mean-cycle cases and five refusals/budget examples, quantify the angular mean-log discrepancy, and draw three own synthetic figures. Fresh Euler coefficients are passed to the isolated particle calculation. No completed numerical result is supplied as a substitute. Mean-cohort cases use their registered physical ages and remain distinct from the interactive Lab's fixed-age control study. The ST solution is never joined to the ejecta-envelope solution.

Inspect `replay-receipt.json` and individual logs. Numerical success concerns these finite cases and registered limits. A passive energy bound does not establish negligible cosmic-ray pressure; the prescribed Bohm scattering does not solve its magnetic waves. The particle endpoint belongs to a short mean-clock window, not a measured remnant maximum energy.

## Optional fixed HESS 2004 flux-point exercise

A separate command fetches exactly one official 9734-byte auxiliary HTML page, checks its complete SHA256, then extracts the 25 whole-remnant Figure12 photon-flux rows. The original observation used a 0.65-degree aperture and 30.5 hours selected live time. This is an old published high-level flux product, not an event/response reanalysis or a current best spectrum.

```sh
python3 optional_observed.py --fetch --output /absolute/new/private/hess-run
```

Exact source: https://www.mpi-hd.mpg.de/HESS/pages/publications/auxiliary/auxinfo_rxj1713_paper2.html

SHA256: `ea1c262671a439f35f8fb57d442dc6eff282c3d3610e2ad9b3689fc53d32daab`.

Later offline use can pass `--source /private/hess2004-auxiliary-private.html` in place of `--fetch`. Corrupted or changed bytes are rejected before an analysis directory is created. Keep the original HTML, extracted rows, fit outputs and native plot local; this package grants no redistribution rights to those third-party materials.

Both point-centre and explicitly uniform finite-bin approximations retain the original **failed** power-law comparison to the published chi-square: differences about 3.019 and 3.935 exceed the unchanged 2.0 threshold. The wrapper reports successful replay of an expected failure; it does not turn the scientific comparison into a pass. Unknown event response, covariance and systematic treatment remain unavailable. The common energy-scale sensitivity is correlated across every point and is not an independent extra observation. No parent proton spectrum is inferred from these photon fits.

## Integrity and provenance

`manifest.json` lists exact package bytes/SHA256 except itself. `input-lineage.json` identifies exact copied own scientific code and registrations. `replay.py` verifies that inventory before calculation. Every run must create a new output directory. Defaults need no cached observations and never automatically download an alternate source.