back to table · edit · history · where entries came from · files · download
3213 bytes, as of the version from 2026-10-25 11:31 (current). Recorded here, not run.
# T484 presentation repair, 2026-10-25
The live authenticated document still contained all three reported issues. The
anonymous API concealed the table and `/T484` returned 404, confirming that it
remained an unpublished draft. The current live and local table skills were
read before editing.
The repair changes two size-value display labels to `cells` and `edges`, gives
three related-table links the titles read from their live API documents, and
adds the missing verb `is` to both bond-row comments. The same two comments in
`generate.py` were corrected. All four entry identities, numeric strings,
uncertainty intervals, parameter order, constraints, references and relation
sentences are unchanged.
The complete proposed diff was printed and read before the API submission.
Before and after fragments were rendered with the site's `/preview` endpoint
and inspected in Chromium at 1440 pixels. All four number rows appeared
together; the row preview omits parameter constraints solely to fit the
preview request limit. The related-links preview includes the existing
polyomino value to satisfy the preview parser. The stored document retains
all parameter constraints and all four rows. The preview editor form was
hidden for screenshots. Both shorter size labels occupy one line, and all
three captions render as readable titles linked to the unchanged addresses.
Each linked table returned HTTP 200. Final previews have no parser alerts,
MathJax errors or JavaScript exceptions.
`repair_checks.py` checks the snapshots and the revised generator, then runs
the existing `checks.py` suite against the proposed numbers. The checks pass:
all four numbers agree, 192- and 384-bit Arb conversions give the same stored
interval, independent 70-digit Decimal endpoints are contained, enumeration
controls return 1 in one dimension, fixed-site counts agree through 8 cells
in dimension 2 and 6 cells in dimension 3, and two bond enumerators agree
through 4 edges. Both finite-series extrapolations pass their exact control
and lie in the retained polyomino interval. These are reruns of the existing
reproducibility checks and source-data comparisons, not fresh simulations,
independent source verification, or certification of the limiting constants.
After the API edit, the complete readback equals the checked proposal.
`verify_stored.py` checks the live document, all four comments and values,
and the package's full `verify(sample=None)` result. Its output records the
Python, Sage and numberdb versions. Run the files with SageMath as explained
in their docstrings; download the existing `checks.py` and `source-data.json`
as well as the two snapshots used by `repair_checks.py`.
The post-edit audit is unchanged: three warnings flag the named class,
lattice and size parameters. No new audit finding was introduced. Splitting
would be inappropriate here: every row is the exponential growth constant
of a specified lattice-animal counting class, and size is fixed by that
class. The existing definition and family comment already explain this.
The repair was submitted through the authenticated API as zeta3 and returned
`reviewed: false`. No publication, review or offer-for-review action was taken.