# Fusion Observatory: cross-lab transfers

Open **Results → Export** after a Torus Lab calculation. Each export identifies the selected stored run, its input hash and kernel. Draft edits are not included. These adapters preserve scientific boundaries; they do not automatically combine incompatible physics or power ledgers.

| Export | Contents | Where to use it |
|---|---|---|
| Cross-lab research package | `hbf-research-package@1`, full immutable Observatory run, source records, methods, quantity definitions and content SHA-256 | Archive or exchange the package. The original run is inside `record.result`. |
| Review panel settings | `{ "package": <research package> }` | In another HBF lab, open **Research records & tools → Controlled run review**, replace its settings with this complete JSON and calculate. |
| Uniform-source record | `hbf-lab-transfer@1`, rates, volume, temperatures, reaction frame and the exact selected evaluation | Source-data review or an explicit adapter. This record is not a DirectConvert particle-flux input. |
| Collector panel settings | `{ "transfer": <hbf-particle-flux@1>, "collector": <settings> }` | In **DirectConvert → Research records & tools → Correlated collector transport**, paste this complete JSON into the settings and calculate. |

Research packages can retain complete or partial runs with their original status. Uniform-source and collector exports require a **complete p–B source calculation**. A field line, particle visualization or incomplete calculation does not provide a normalized escaping alpha distribution.

## Collector scenario inputs

The export view lets the researcher edit the entrance normal, accounting time window and collected fraction of escaping alpha energy. The normal must have unit length; the application does not silently normalize it. The fraction must be between zero and one, and the finite positive time window is in seconds. These are separate export assumptions and do not change the original source configuration or its scientific hash.

The constructed scenario assumes one alpha energy, `Q/3`, and every direction equal to the entrance normal. It omits incident reactant kinetic energy and does not model physical three-alpha correlations, an experimental energy spectrum, angular transport, extraction probability or a collector transit time. The preview and exported record identify it as an **explicit monoenergetic, all-forward scenario**, not a simulated escaping distribution.

Only the source ledger's escaping alpha power is available:

`collector input power = escaping alpha power × collected escaping fraction`

`incoming alpha rate = collector input power / ((Q/3) × joules per MeV)`

Deposited alpha heating remains outside this collector input. Uncollected escaping power is recorded separately, so nuclear source power is partitioned once. The downstream planar-retarder model reports its own conversion losses and residual. Its conditional electric output is not net plant electricity or experimentally demonstrated conversion performance.

Uniform-source records retain `SIKORA-WELLER-2016` or `TENTORI-BELLONI-2023-HT` with the original source version and applicable metadata. They use relative center-of-mass energy for reactivity. Since the source task has no dynamical observation duration, its source transfer leaves duration and expected reaction count unavailable rather than assuming a time interval.

## Verification

Run `node --test tests/observatory-transfers.test.mjs`. Five groups exercise the existing research-package checksum contract, matching source identities, real `validateParticleTransfer`/`correlatedCollector` compatibility, source/collector energy partition, zero escaping power, invalid settings and incompatible task rejection. These tests verify software contracts and scenario accounting, not physical reactor validation.