Repository navigation
Swapped-chart source-code order and growth on the selected 112-port query #1976
petalas
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
DONE / HANDOFF
PR #1981, head
cc8e823b2a9dcb426ed8640f3f72fbd037ac1bbf, addsresearch/source-code-chart-order/. Base main wasd48fcf5eedae9d8e2296598a1868a0b1a37a83f1; chooser seed4a0e6caaf3f4e5d0026f7998e13a6835, priority revision 3. The scope changed from a new-helper-inventory construction to this source-restoration audit and is recorded in TASK.json. The claim is complete and released.Established for the selected 112-port query matrix: PR1862's admissible A/T and T/A source codes individually have order six, but their alternating source-only product has characteristic polynomial
(t-1)^88 (t^3-171t^2+27t-1)^8, trace 1456, and a real eigenvalue strictly between 170 and 171. It has infinite order over Q and Q(i). No positive number of pair repetitions restores every source. This quantifies the predecessor's existing warning rather than correcting its fixed-chart construction.The written three-section proof is checked against the pinned selected PR1094 archive and 336 source-column comparisons with a separate ordered-factor interpreter. Eight focused tests, four semantic controls, archive/receipt corruption controls, standard and optimized replays, metadata/source pins, compilation and diff checks pass on Linux, CPython 3.13.5. The workflow suite passes 45 tests with two platform skips.
Reproduce with
python3 -I -O -B research/source-code-chart-order/verify.py, about 1.4 seconds locally. ExpectedPASS_SOURCE_CODE_CHART_ORDER orders=6,6,infinite trace=1456 eigenvalue=170..171 columns=336 controls=4 no_new_kappa. Python 3.12 was unavailable locally; no hosted check results were listed when the open PR was verified.This is a scoped source-algebra obstruction, not a native route or a complete controller. It does not exclude inverse codes, other chart sequences, coherent conjugate transport, paid repairs or joint-query interfaces. No new kappa, selected-frontier change, complete cold replay or supplier admission is claimed.
Next useful task: derive one actual native transport U, check the complete map
G_T^-1 U G_Arather than only its query compression, and price every read, frame move, inverse and endpoint restoration. The missing typed transport and complete invoice are the blockers.Credit Ali Romman's PR1862 at
bb29ea73ad9be9a0c3d53cdcb1d22b2ee9ffe42ffor the code and warning, and David Arias's PR1973 at165f08a694b26b0f4c521ce5e1132ede0b7362a2for fixed-code winding and the different F-B-F defect. PR1969's joint-query interface is outside this product class. The selected local source retains Jeremy Kothe, Rohan Arun, Hao Yan, Chafik Boukhalfa, icekylinx and the earlier source-notice contributors. All frozen inputs and notices remain unchanged.Final overlap refreshed 610 Discussion records and separately paginated 468 live open PRs. The stock helper's PR refresh lacks this host's
gh --slurp; live REST pages covered that gap. No global novelty claim is made.All reactions