Computational supplement

For the AI-generated paper Fujita freeness on complex projective sevenfolds (PDF) · posted here 2026-09-20 06:51:18 UTC

This supplement contains the complete finite-arithmetic data and the independent checkers for the thirteen certificate packages (75 certificate families) described in Appendix A.4 of the paper. The definitions, the geometric arguments showing that the certificates apply, the uniform error estimates, and the analytic reductions are in the paper itself; a successful computation does not replace those proofs, and it is not a formal verification of the paper.

Download

fujita-sevenfold-supplement-2026-09-20.zip (12.6 MB, 166 files).
SHA-256: a54bfd927c95cb78996b6edab1c830acc0fbc698ed068ffbe5e9c47b5cd4a52d

Individual files: README · manifest.json (byte counts and SHA-256 of every file) · check.py (the dispatcher) · validation-summary.json (the recorded full run).

How to run the checkers

Use Python 3.10 or later with its standard library only. After unpacking, run from the directory of the supplement:

python3 -B check.py --manifest manifest.json --report runs/full-001/aggregate.json

The dispatcher verifies the delivered inventories, runs every section, and checks that the inputs remain unchanged. No network, external package, or optimizer is needed. A zero exit code and status: pass with full_package_run: true mean that every declared arithmetic package passed. The recorded fresh full run of this release took 341 seconds and passed all thirteen sections. Do not use Python’s -O option.

Contents

Section packageCertificate familiesFiles
05-tails23
07-capacities103
09-original-sources513
10-ledger73
11-quartic-threefold45
12-cubic-fourfold59
13-triple-fivefold46
14-double-geometry23
16-double-twentynine315
17-ambient-endpoints2549
18-intrinsic-double58
19-large-deficit12
20-rank-two-feedback23

The data and the checkers were produced by the Danus system together with the paper and, like the paper, have not been verified by a human.