Research: shared target donors save 3,288 R420 XORs (no new kappa)

research/r420-target-star/CLAIM.json

@@ -0,0 +1,36 @@
+{
+  "track": "bit supplier",
+  "assignment_seed": 0,
+  "randomness": "Python secrets.randbelow(100), OS entropy, drawn once before task selection",
+  "base_commit": "66849e1475c930fd7732b1a0067ffee7e1e95f60",
+  "claimed_at_utc": "2026-10-13T04:32:22.523809+00:00",
+  "expected_finish_utc": "2026-10-13T05:07:56Z",
+  "coordination": "No pinned issues returned by GitHub pinnedIssues; local claim as instructed. No separate reservation issue opened.",
+  "object": "PR543 R420 frame-zero D compensation, target 0 (register 120) as one fixed donor",
+  "method": "A single involutory target-star basis: pre-add donor to each chosen target, replace receiver helper rows by symmetric differences, post-add donor. Enumerate all 119 receiver choices independently; include each strictly positive net saving after both restoration XORs. Seed 0 selects donor target 0.",
+  "expected_deliverable": "Exact finite circuit with arbitrary dirty-target restoration, source pins, complete native cost/frame/prefix replay, scoped proof and rejection controls; no new kappa.",
+  "candidates_considered": [
+    "Phase-B 12-register block resynthesis: overlaps #670/#674/#688.",
+    "Phase-A decoder-total joint synthesis: would need a broader changed retained-total interface.",
+    "Target-star D compensation: selected; changes target rows, while #686/#694 factor helper columns."
+  ],
+  "prior_work": [
+    "#476/#478 already use reversible target-interval factoring on older R715 B words; the elementary identity is credited, not claimed new.",
+    "#661 closes fixed-order R420 phase-B frame changes, not D circuit changes.",
+    "#686/#694 change D by restoring a helper basis; no code or recipe imported."
+  ],
+  "initial_overlap_max_pr": 704,
+  "result": {
+    "status": "completed finite bit-circuit contribution; no new kappa",
+    "updated_at_utc": "2026-10-13T05:02:13.305041+00:00",
+    "url": "https://github.com/CrocSwap/integer-mult-bounds/pull/742",
+    "code_commit": "40f0d419a9012c0243e87a686ea22327619c766c",
+    "evidence_driven_extension": "After #713/#716 arrived with stronger disjoint target pairs, the one bounded extension permitted shared donors while retaining disjoint donor/receiver sets. Limit 16 donors fixed before running the extension; no other pivot.",
+    "completed_domain": "All 14,280 oriented pair costs and 1,680 candidate donor sets in 15 greedy additions, starting with seed-selected donor target 0; stopped at 16 donors. Best assignment for fixed donors is exact; donor-set optimality is not claimed.",
+    "result": "16 donors, 100 receivers, 3288 ADDs saved including all preparation/restoration. D 19922->16634; complete native word 26916->23628. Rank mass 5112 and complete frame/copy paths unchanged. Native payload 102 bits under 104-bit cap.",
+    "verification": "Nine construction controls, all 660 native columns and both directional exports; Python and unchanged C++ 5/9/11-stage replay on both words; actual-stream bridge omissions, native controls, isolated optimized/minimal-checkout reproduction, source/certificate corruption controls and selected-record check passed.",
+    "time_allocation": "Initial orientation/selection ended at local claim before 04:33 UTC (< first 20% of 40-minute budget). Final circuit fixed by 04:55 UTC; remainder reserved for verification, overlap checks and handoff.",
+    "concurrent_credit": "#476/#478 identity; #686/#694 helper basis; #713/#716 target pairs. #731 baseline rank audit and #733 stronger two-sided helper/target composition were inspected at final submission; no imported code or recipe, no overall record claim.",
+    "remaining": "Review/integrate semantic D interface; hosted-parent/whole-owner and global prefix/invoice replays; changed-word assembly. All inherited all-size/physical/hosting/precision/analytic assumptions remain."
+  }
+}