Skip to content

Installation

Web App (no install)

The fastest way to use HYRR: exoma-ch.github.io/hyrr

Everything runs in your browser — no Python, no install, no data upload.

Desktop App (offline)

For air-gapped machines or offline use, download the native desktop app from GitHub Releases:

Platform Artifact
Windows 10+ .msi installer or .exe (NSIS)
macOS 10.15+ (Apple Silicon) .dmg
macOS 10.15+ (Intel) .dmg
Ubuntu 22.04+ .deb or .AppImage

The installer ships with the always-needed meta/ and stopping/ data (~54 MB) bundled. The chosen cross-section library (default tendl-2023-iso, ~50 MB) is downloaded on first launch into ~/.hyrr/nucl-parquet/v{version}/. A returning user pays no network cost; switching libraries triggers another small download. Installer size ~80 MB (was ~15 MB before #52 — the difference is bundled meta+stopping).

Python Package

uv add hyrr

From source

git clone --recurse-submodules https://github.com/exoma-ch/hyrr.git
cd hyrr
uv sync

Nuclear Data

HYRR uses the nucl-parquet data package, included as a git submodule. The data is resolved in this order:

  1. --data-dir CLI argument
  2. HYRR_DATA environment variable
  3. Managed cache at ~/.hyrr/nucl-parquet/v{version}/ (sentinel-protected)
  4. ../nucl-parquet sibling directory (submodule, dev layout)

Fetching data

The Python CLI populates the same managed cache the desktop app uses:

hyrr fetch-data                       # default: meta + stopping
hyrr fetch-data --library tendl-2023-iso  # specific library (~50 MB)
hyrr fetch-data --all                 # every library (~400 MB)

The default library is tendl-2023-iso. Override with --library or HYRR_LIBRARY env var. The cache is sentinel-protected: a partial download or interrupted extract leaves the dir behind for cleanup but is never picked up as a usable cache.

Air-gapped install

For machines without internet (Faraday-cage labs, isolated networks):

  1. On a connected machine, populate the cache and pack it:
hyrr fetch-data --all
hyrr fetch-data --offline-bundle hyrr-data.tar.zst

  1. Copy hyrr-data.tar.zst to the air-gapped machine (USB stick, internal share, etc.)

  2. On the air-gapped machine, install the bundle:

hyrr fetch-data --from hyrr-data.tar.zst

Both machines must run matching hyrr versions for the bundle to apply cleanly — the cache is version-pinned, so a v0.9 bundle won't be picked up by a v0.10 install. Alternatively, drop the upstream GitHub Releases tarball directly. The release URL pattern — host, path layout, tarball filename — is defined by hyrr_core::data_fetch::release_url() / tarball_filename(). Read your installed version from Python with python -c "import hyrr._native as n; print(n.py_data_version())" and substitute it for <V> below:

# On a connected machine — substitute <V> with the version printed above
# (data version is CalVer YYYY.MM.MICRO, e.g. 2026.5.0; no `v` prefix)
curl -LO "https://github.com/exoma-ch/nucl-parquet/releases/download/data-<V>/nucl-parquet-data-<V>.tar.zst"

# On the air-gapped machine
hyrr fetch-data --from "nucl-parquet-data-<V>.tar.zst"