Example team
Quantum Science in Space
Three researchers and their agents slow down a cloud of ultracold atoms, so that one day it can be watched falling freely in space. Each result their agents record is a claim, with the code, data and figures behind it, reviewed by the team, checked against the papers it uses, and written up step by step in a labbook.
The people and their results are made up; every number was computed by the code its claim shows. The two papers and the checks of them are real. What you see are the Library’s real screens.
Start here
Labbook
Slowing an atom cloud for a test of free fall in space
The project as a document: 5 steps, one per claim, written as the claims came in. Built by the Library as a PDF.Map
How the claims build on each other
Every claim, what it uses, extends or replicates, and the papers it rests on.Reviewed and corrected
A 0.9 ms magnetic lens slows the cloud's expansion about seven-fold
The supervisor found a swapped digit; the author corrected it the next morning. The original stays, the correction sits beneath it.A paper, checked by an agent
Müller et al., Trap-Quenched Matter-Wave Optics in Space (2026)
Before a claim used Tab. II, the agent checked it, and found a unit misprinted. Verdict: holds with correction.
The team
Lena Hartmann
Supervisor — reads and corrects
Jonas Weber
Experimentalist — runs the chip
Mira Okafor
Theorist — models and simulations
5 claims, 4 reviewed · 2 papers checked · 1 open issue · the team’s page
This fortnight
- 2026-10-12
Lena Hartmann reviewed One lens is not enough for a 10⁻¹⁵ test of free fall in space; two would be · an open issue
- 2026-10-11
Mira Okafor published a result One lens is not enough for a 10⁻¹⁵ test of free fall in space; two would be
- 2026-10-11
Mira Okafor added a step to the labbook Slowing an atom cloud for a test of free fall in space · 5 steps, PDF rebuilt
- 2026-10-11
Mira Okafor had an agent check Tab. I of Struckmann et al., Platform and environment requirements of a satellite quantum test of the Weak Equivalence Principle (2023) · holds
- 2026-10-09
Mira Okafor replied on A longer lens is not a stronger one: at 1.8 ms the cloud expands as fast as without
- 2026-10-09
Jonas Weber replied on A longer lens is not a stronger one: at 1.8 ms the cloud expands as fast as without
- 2026-10-09
Mira Okafor reviewed A longer lens is not a stronger one: at 1.8 ms the cloud expands as fast as without · a question, settled
- 2026-10-08
Jonas Weber published a negative result A longer lens is not a stronger one: at 1.8 ms the cloud expands as fast as without
- 2026-10-06
Jonas Weber corrected A 0.9 ms magnetic lens slows the cloud's expansion about seven-fold · corrected
- 2026-10-06
Lena Hartmann reviewed A 0.9 ms magnetic lens slows the cloud's expansion about seven-fold · an error found
- 2026-10-05
Jonas Weber published a result A 0.9 ms magnetic lens slows the cloud's expansion about seven-fold
- 2026-10-01
Mira Okafor published a method A model of the cloud after release: the residual magnetic push and the best lens duration
- 2026-10-01
Mira Okafor had an agent check Tab. II of Müller et al., Trap-Quenched Matter-Wave Optics in Space (2026) · holds with correction
- 2026-09-30
Lena Hartmann reviewed A Bose–Einstein condensate of 1.2 × 10⁵ rubidium atoms on the chip · nothing to raise
- 2026-09-29
Jonas Weber published a result A Bose–Einstein condensate of 1.2 × 10⁵ rubidium atoms on the chip
Open issues
The simulation keeps the residual push measured on ground. In orbit there is no gravitational sag, so the cloud sits about 30 µm higher in the lens and sees a different field. One run with the cloud displaced would settle whether the margin survives.
On One lens is not enough for a 10⁻¹⁵ test of free fall in space; two would be
Papers checked
Claims
Shared in Quantum Science in Space · Result
One lens is not enough for a 10⁻¹⁵ test of free fall in space; two would be
The 76 pK reached with one lens [1] is above the 50 pK that a satellite test of the universality of free fall at the 10⁻¹⁵ level requires [2]. Simulating a second, weaker lens 40 ms after the first predicts 31 ± 5 pK, below the requirement.
Mira Okafor · Claude Code · Opus 5.5 · Berry Science agent skills · 2026-10-11 · Reviewed · 1 candidate issue
Shared in Quantum Science in Space · Negative result
A longer lens is not a stronger one: at 1.8 ms the cloud expands as fast as without
Scanning the lens duration from 0.3 to 1.8 ms, the expansion energy is lowest at τ* = 0.91 ± 0.02 ms, as the model of [1] predicts (0.88 ms), and rises again beyond it: at 1.8 ms it is 501 ± 46 pK, the energy without a lens within the uncertainty. A longer pulse over-focuses the cloud instead of cooling it further.
Jonas Weber · Claude Code · Opus 5.5 · Berry Science agent skills · 2026-10-08 · Reviewed · 0 candidate issues
Shared in Quantum Science in Space · Result
A 0.9 ms magnetic lens slows the cloud's expansion about seven-fold
Re-trapping the condensate for 0.9 ms, 20 ms after release (a magnetic lens), lowers its expansion energy from 506 ± 21 pK to 67 ± 10 pK, so that after 300 ms of free fall the cloud is 38 µm instead of 70 µm wide (rms).
Jonas Weber · Claude Code · Opus 5.5 · Berry Science agent skills · 2026-10-05 · Reviewed · 1 candidate issue
Shared in Quantum Science in Space · Method
A model of the cloud after release: the residual magnetic push and the best lens duration
Treating the chip's residual magnetic field after release as a polynomial force, with the coefficients fitted for a comparable chip experiment [1], puts the cloud's centre 0.22 mm away from a ballistic path after 300 ms. A thin-lens model of the magnetic lens predicts the lowest expansion energy for a pulse of τ* = 0.88 ms.
Mira Okafor · Claude Code · Opus 5.5 · Berry Science agent skills · 2026-10-01 · Unreviewed
Shared in Quantum Science in Space · Result
A Bose–Einstein condensate of 1.2 × 10⁵ rubidium atoms on the chip
After 1.6 s of evaporative cooling the atom chip holds a Bose–Einstein condensate of (1.19 ± 0.06) × 10⁵ ⁸⁷Rb atoms, with a condensed fraction of 0.74 ± 0.02, measured by absorption imaging 20 ms after release.
Jonas Weber · Claude Code · Opus 5.5 · Berry Science agent skills · 2026-09-29 · Reviewed · 0 candidate issues
How this example was made
The team, its members and its measurements are fictional. The images and curves come from a simulation; the fits, the model and the uncertainties are computed by the code each claim shows, and the labbook PDF was built from the claims by the Library’s own generator.
The two papers are real, and so are the passages the reviews check: the misprinted unit in Tab. II of arXiv:2606.14577, and the requirement table of arXiv:2310.04212, which holds.