For optical computing groups

You are building new hardware or a new reconstruction method, and a reviewer will ask what you compared it against. ApertureAPI gives you a named, recorded baseline reconstruction to put beside your own: every result records exactly which input, settings and software build produced it, so your comparison says precisely what it was compared against.

What you get

• A baseline you can name. A conventional reconstruction of each of your holograms, run on a recorded software build. The provenance ID commits to the input, the settings and the software version; if the software changes, the record names the new build. Keep the result package you compared against: on the in-line path, two runs of the same input and settings do not yet return identical output arrays (see Validation docket).
• An evidence record per run. evidence.json commits to your input and your settings by SHA-256 without publishing either, names the software version, and states in writing what is and is not established.
• Check, not trust. Anyone holding the same input can rerun the call and confirm by provenance ID that the same input, settings and software were used. See Reproducibility bundle for a worked example.

What this is not

It is not a claim that our reconstruction is the right answer. It is a reference point you can name, cite and check. If your method beats it, the comparison is yours to publish. We do not quote agreement or accuracy figures for any setup, including ours.

How groups use it

1. Run your holograms through the API with the settings your setup needs.
2. Keep each result's evidence_report_id and provenance_id.
3. Compare your method's output against our arrays in reconstruction_data.npz.
4. Cite the baseline with the methods line (see Methods supplement).

Next step

Send one hologram with its wavelength, pixel pitch and rough distance to system@apertureapi.com and we will send back a result and its evidence record. Or start now with no account at https://api.apertureapi.com/. For a structured engagement on your own data, see Design partnership.