Known limitations¶
SPM-Kit is alpha software. Its strongest current public evidence is deliberately narrower than its feature list. This page collects the practical boundaries a researcher should check before treating an output as scientific evidence.
Scientific scope¶
- No capability currently carries a general
LEVEL 4 — PHYSICALLY_VALIDATEDorLEVEL 5 — REPRODUCIBILITY_VALIDATEDclaim. - The retained Gwyddion campaigns cover Sa, Sq, and Sz under declared preprocessing, matrix, version, and tolerance conditions. They do not prove universal equivalence with Gwyddion.
- Contact-mechanics, chain, relaxation, and force-volume recovery evidence is synthetic. Tip shape, spring constant, sensitivity, contact point, adhesion, sample geometry, and fit window remain experiment-specific responsibilities.
- KPFM, spectral, grain, and resonance utilities have software evidence but no broad public physical-reference campaign.
File and metadata scope¶
- Reader support is variant-specific, not a promise that every file sharing an extension will load correctly.
- Nanoscope
.spmevidence covers six demonstrated files and was affected by a documented pre-freeze unblinding incident. - NanoSurf
.nidcomparisons include selected laboratory-context files, but the private instrument corpus is not distributed. - Optional adapters inherit behavior from their installed dependency versions.
- Instrument metadata is preserved when available; its presence does not prove calibration, traceability, or scientific suitability.
Distribution and interface scope¶
- The current source tree is
0.1.5.dev0, the latest GitHub release is0.1.4, and PyPI currently serves0.1.2. Choose an installation source explicitly. - Fathom depends on Qt and platform graphics behavior. Offscreen GUI tests do not establish identical behavior on every desktop.
- The CLI and public Python API are the automation contracts. GUI labels and layouts may evolve during alpha releases.
Evidence handling¶
Data Hunter records candidates; it does not certify them. Phantoms provide declared truth; they are not physical references. Validation records campaign outcomes; a successful run only supports the frozen claim and scope written in that campaign.
For capability-level detail, use the scientific status matrix, the format matrix, and the retained campaign record.