Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1,092 changes: 1,092 additions & 0 deletions research/fstar-tied-residue-seeddcc97691/CERTIFICATE.json

Large diffs are not rendered by default.

87 changes: 87 additions & 0 deletions research/fstar-tied-residue-seeddcc97691/README.md

Large diffs are not rendered by default.

24 changes: 24 additions & 0 deletions research/fstar-tied-residue-seeddcc97691/SOURCES.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
{
"schema_version": 1,
"base_commit": "0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2",
"files": [
{
"path": "certificates/selected-result.json",
"usage": "provenance",
"git_blob": "ae6a99f28aa468649eb66a361e764d769363bf11",
"sha256": "10b9e340fd8e354a37a0c4813c90e5de1dd42044b5f6aca4a3fc5a42b685fa32"
},
{
"path": "docs/research/agent-task-catalog.json",
"usage": "provenance",
"git_blob": "e8870e5917057d59b730eb14721a778639eca050",
"sha256": "d46eb6d5b3e279428753d770653663fbdf1063e55c255ea073d768f344927ef7"
},
{
"path": "research/root-unit-decoder-lemmas/README.md",
"usage": "input",
"git_blob": "de305484c78226a3f8da5f3277ced78974116fbb",
"sha256": "e3c856bc361a57cc2424c829001dba94e8f3a1bc8f6616f9c2f740e9665506a9"
}
]
}
78 changes: 78 additions & 0 deletions research/fstar-tied-residue-seeddcc97691/TASK-CARD.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,78 @@
# Task card — KB5b: the tied-minimum / isotropic-line residue of F-star

```text
Task ID / proposed scope key: fstar-tied-residue / fstar-tied-residue-seeddcc97691
(Kappa-blocker KB5b, explicit open request; chooser OS seed dcc976912ab4c64bcee65295e5f63e98
drew track bit-mechanisms with cards bit-helper-contract and bit-new-register-model;
recorded scope-change reason in TASK.json.)
Base commit; exact predecessor and dependency heads:
base main 0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2 (priority revision 4);
predecessor proofs: https://github.com/CrocSwap/integer-mult-bounds/pull/2717 head 10644c90837cb377df89be7c87992320ebfaf091;
predecessor census: https://github.com/CrocSwap/integer-mult-bounds/pull/2335 head 856cd7603ba679aa56294e53d60840061d176778;
fixtures: https://github.com/CrocSwap/integer-mult-bounds/pull/2111 head 98da76c2839ba38538b510420dec54546ccabf0b.
Concrete object and question:
Open problem F-star of research/root-unit-decoder-lemmas section 7 (floor form):
cert(M) >= floor(2v/3) for every exact 3-total rational decoder M = sum_j p_j q_j^T.
Residue class attacked: (i) tied-minimum / isotropic lines — the lines the mod-2 anchor
of Lemma AN cannot anchor (M_tt even under every 2-integral normalization because the
minimal 2-adic valuations among the products p_k(t) q_k(t) tie and the reduction is
isotropic, so no one-sided repair exists). Question: can anchorable lines and tied lines
always be kept together at >= floor(2v/3) by a mixed anchor + matched-pivot labelling
with one order, and if not, what exactly fails first?
Scope status: changed (from the drawn bit-mechanisms cards); reason: authorized explicit
open request KB5b. Pair-collision check at entry (toolkit disc-comments on discussions
1317 and 1166 including nested replies, 2026-10-15 18:26-18:29 UTC) found no live claim
on the F-star residue; KB5 (PR 2717) is DONE/RELEASED at 18:20 UTC.
Existing restriction being changed (or exact claim being audited):
Nothing is relaxed. The restriction being addressed is PR 2717's stated demotion:
"Lemma AN handles only the even-diagonal count, not their repair or ordering" for the
tied-minimum / isotropic lines. Every new claim carries its exact hypothesis and is
demoted where unproved.
Method; explicit parameter ranges/candidate alphabet:
Stdlib exact integer/F2 arithmetic. (a) A written mixed anchor-pivot lemma (Lemma MX)
with exact hypotheses: anchor labels on a set A (zero conflicts inside A at incidences),
matched pivots on the complement R with pivot classes C_i, and an exact placement
criterion for inserting anchor lines into the pivot order; strictly extends AN
(R empty) and the one-class matched-pivot construction of MP. (b) Per-line classification
of every instance into anchor-good / one-sided-repairable / tied-isotropic, computed
exactly from the 2-integral factorization. (c) Calibration on PR 2111's stored fixtures
tied_min (v = 12) and cex_min_1 (v = 19). (d) Smallest-first bounded falsifier search in
the tied residue: v ascending from 3, tied-heavy generators (all diagonals even, tied
valuation profiles, entries in {+-2,+-4} and {+-1,+-2} and rationals with odd primes in
denominators), exact cert where a-enumeration terminates, witness replay everywhere,
admissible-B chain cost as the rigorous falsifier criterion.
Nearest overlapping PRs and why this adds something:
PR 2717 (Aaron Wang) proved SF/MP/RK/AN and named this exact residue; this package
attacks the residue itself with a mixed lemma and a tied-domain search, not another
census of the censused domain. PR 2335 (Aaron Wang) is the 196-instance census;
its domain (v = 4..12, entries -2..2) is not repeated. PR 2111 / PR 2162
(DreamingOfClouds) own the fixtures and their certificate limits; cited, not restated.
PR 2379 (Aaron Wang) pins cost(tied_min) = 12 exactly; Discussion 2202 pins
cost(cex_min_1) = 21 exactly. Refined overlap pass on the object's proper names
(F-star, cert(M), 3-total decoder, tied minimum, isotropic, zero digraph, 3-dichromatic,
L8 labelling, matched pivot, mod-2 anchor, tied minima, tied_min, cex_min_1,
root-unit decoder, 2-adic, one-sided repair) at 2026-10-15 18:27-18:42 UTC, complete
pagination both passes, found no other work on this residue.
Optimistic full cost budget; required observable improvement:
No kappa/supplier budget in play (missing-lemma work). Required observable: a PROVEN
lemma whose hypotheses strictly include AN's and MP's and which certifies >= floor(2v/3)
on instances where either alone is short, or an exact obstruction showing why the mixed
route fails at the smallest instance where it fails.
Success evidence; falsification test; time/compute cap:
Success: PASS markers from verify.py (controls, fixture calibration, MX bound vs
threshold per instance, search summary) and a written proof of Lemma MX in README.md.
Falsification test: an explicit admissible pairing B with sum_j rank_F2 B[R_j,A_j] <
floor(2v/3) refutes F-star on that instance (chain is proven); a corrupted witness must
fail the independent checker. Compute cap: one session, single-process stdlib Python,
each registered command below 600 s.
Predecessor authors; intended artifact and reproduction command:
F-star / cert(M) / L8 / the proven chain: research/root-unit-decoder-lemmas (Douglas
Colkitt with OpenAI Codex assistance lineage). Fixtures tied_min, cex_min_1 and their
recorded one-copy limits: DreamingOfClouds, PR 2111 / PR 2162. Lemma residue statement
and engine design this package's engine adapts: Aaron Wang, PR 2717. Census frontier:
Aaron Wang, PR 2335. Exact cost(tied_min) = 12: Aaron Wang, PR 2379. Exact
cost(cex_min_1) = 21: Discussion 2202 (author credited in README after verification).
Artifact: research/fstar-tied-residue-seeddcc97691/. Reproduce:
python3 -B research/fstar-tied-residue-seeddcc97691/verify.py
python3 scripts/research_package.py --package research/fstar-tied-residue-seeddcc97691 --check-only
```
Loading
Loading