q4 — leftover (19,3) nests

The published census is 2,367 nonempty degree-8 T-curve schemes (arXiv:2602.06888 v3, §5.3). Seventeen more sit in ../certs/new_schemes.json, so the lower bound in this folder is 2,384. q1–q3 finished the radius-1 leftover thicken of every leftover census triangulation of twist-rank at most 26. The two open (19,3) deep nests remain.

Replay

From problems/hilbert16-degree-8/compute:

sh q4/run_all.sh

That re-checks the seventeen and the bow-tie collection.

Searches

From the same directory, after python3 prep.py:

python3 q4/ladder3.py q4/dn_out/ladder3_193.jsonl 0 193
python3 q4/ladder3.py q4/dn_out/ladder3_depth3.jsonl 0 depth3
python3 q4/even_walk.py bfs q4/even_out/bfs.jsonl 1200000
python3 q4/odd_skel.py 5 q4/even_out/odd_skel5.jsonl
python3 q4/write_certs.py
python3 q4/collect.py

A new scheme is a T-curve only after python3 verify_new.py q4/certs/new_schemes.json. An incomplete search is not a lower bound.

What this run found

The three-split around the twelve published depth-3 M-collections finished 12/12: 3,909,329 evaluations, 29 known schemes, neither open nest. A 1.2 million even-split BFS stays on the five published (19,3) M-schemes with the queue still open. No certs/new_schemes.json. The bound stays 2,384. See certs/ after collect.py. A new scheme is a T-curve only after the exact verifier.