Capability discovery
Check what your file contains before choosing an analysis. File readers are included in the normal installation; see installation.
Run this first:
Use capabilities for each input type, especially in a mixed-file batch.
It shows the detected format, reader version, available data and analyses.
Use
openwfn --format json FILE capabilities for automation. In Python,
load(path).capabilities() returns the same inferred component states.
available means the relevant data are directly present. derived means
openWFN can calculate the component from present data; for example, AO overlap
may be derived from a supported basis. missing means the file does not
provide it. These are the states currently returned for listed components.
unsupported is reserved in the capability type for unrecognized requirements;
an analysis lacking required components returns a structured Unsupported
result. Capability states describe what can be attempted, not an independent
accuracy certification. A missing wavefunction must not be filled in from
atomic electron counts alone.
For ambiguous inputs such as GAMESS .dat, use
openwfn --input-format gamess FILE capabilities or
load(path, format_hint="gamess"). doctor includes the same normalized
capability and analysis report. An available summary requirement means the
file has structure; a geometry-only or periodic summary is still marked
partial because electronic fields are unavailable.
Only one record is ingested from a multi-record file. Periodic structures, stored volumetric grids, and FCIDUMP integrals remain distinct canonical components; the molecular frontier/population/density methods do not become available merely because a file has a grid or integrals. See the format matrix and limitations.
Grid resource safety
Direct molecular grid operations reject grids above 2,000,000 points before allocating coordinate arrays. The limit applies to density integration, cube export and grid-based ESP. A rejected request must be retried with larger spacing or smaller padding; openWFN does not change either setting automatically. Smaller padding can omit density tails, so check conservation and grid convergence afterwards. AO evaluation chunks also shrink for larger bases. These bounds reduce memory use but are not a guarantee that every calculation fits on every machine.