05 / Experiment

Life of an experiment

Beam time — your scheduled hours with the neutron or muon beam pointed at your sample — is one of the scarcest resources in science: ~1,200 experiments a year, allocated by peer review, scheduled months ahead, and over in a few days. Here’s the whole journey — and where the software you build or run fits into it.

5.1 — Seven phases

Proposal to paper

Click each phase to see what happens, who is involved, and which software is in the critical path.

01 — Propose

A research team writes a short scientific case: what they want to measure, on which instrument, and why it matters. There are three application routes: the main twice-a-year call (“Direct Access”, deadlines around mid-April and mid-October), plus fast-track “Rapid Access” and “Xpress” routes for urgent or simple measurements.

who’s involved
Visiting researchers, with advice from ISIS instrument scientists.
systems & software
Proposal submission system and user database.

The short version: proposal → panel review (~6 weeks) → beam time roughly 3–8 months after the deadline. During Measure, every detected event is timestamped and the sample conditions are logged alongside.

5.2 — Why an hour matters

A day on an instrument

ISIS runs in cycles — stretches of several weeks with beam on around the clock, separated by maintenance shutdowns. During a cycle, an instrument is handed from one user team (ISIS-speak for the visiting researchers) to the next with barely a gap; experiments run through the night with support staff on call.

A typical allocation is two to five days. There are no do-overs: if equipment, beam, or software loses six hours, that might be a quarter of someone’s experiment — possibly one they waited eight months for. It’s the closest science gets to a live production deployment.

00–03 measure 03–06 measure 06–09 measure 09–10 swap sample 10–15 measure 15–16 recalibrate 16–21 measure 21–24 measure

A run schedule is mostly “measure” — every other block is something the team tries to minimize.

5.3 — Follow the bytes

Where the data goes

instrument control + DAQ raw events NeXus / HDF5 files archive + catalogue, kept forever reduction Mantid — often automatic analysis → paper models, fits, DOI one run (one continuous measurement, with its own ID) = millions to billions of events; every byte traceable to proposal + run number reduced data is what users actually take home — small, calibrated, in physical units
FIG 5.3 — the pipeline every experiment’s bytes travel. DAQ = data acquisition; NeXus = the neutron/muon community’s standard file format, built on HDF5. raw files stay; reduced data travels; conclusions get published.

Want the reduction step hands-on? Drive the pipeline yourself →

5.4 — Checkpoint

Quick check