{"abstract":"search stops before a live path can beat the finished set.","category":"Beam search bookkeeping","checks":7,"contract":"Stop only if at least k finished scores exist and the best live score is no greater than the kth-best finished score; future deltas are nonpositive.","contract_signature":"live_scores, finished_scores, k","evaluation_group":"beam-early_bound","failed_approach":"Comparing against the worst finished score stops while the kth result is still improvable.","family":"w3-beam-search-bookkeeping-early-stop-bound","id":"FA-98496","implementations":{"attempt":{"sha256":"b012378c59571446a76af47a647814913d60e9b6c4c07a5dd17c14bc8ec62f20","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(live_scores, finished_scores, k):\n    return len(finished_scores)>=k and max(live_scores,default=float(\"-inf\"))<=min(finished_scores)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('one result can stop',solve([-N],[0],1),True)\ncheck('live still better',solve([1],[0],1),False)\ncheck('need k results',solve([-N],[0],2),False)\ncheck('kth threshold',solve([-N],[3,1],2),True)\ncheck('below kth threshold',solve([2],[3,1],2),False)\ncheck('first result still improvable',solve([2],[3,1],1),True)\ncheck('no live paths',solve([],[3],1),True)\nprint(json.dumps({\"observations\": observations, \"passed\": all(x[\"passed\"] for x in observations)}, ensure_ascii=False))\nraise SystemExit(0 if all(x[\"passed\"] for x in observations) else 1)\n"},"broken":{"sha256":"d52f28465d9503389472b215a2e7180248abfb9b6f90df50911a318e07b7a0c7","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(live_scores, finished_scores, k):\n    return bool(finished_scores) and max(live_scores,default=float(\"-inf\"))<=max(finished_scores)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('one result can stop',solve([-N],[0],1),True)\ncheck('live still better',solve([1],[0],1),False)\ncheck('need k results',solve([-N],[0],2),False)\ncheck('kth threshold',solve([-N],[3,1],2),True)\ncheck('below kth threshold',solve([2],[3,1],2),False)\ncheck('first result still improvable',solve([2],[3,1],1),True)\ncheck('no live paths',solve([],[3],1),True)\nprint(json.dumps({\"observations\": observations, \"passed\": all(x[\"passed\"] for x in observations)}, ensure_ascii=False))\nraise SystemExit(0 if all(x[\"passed\"] for x in observations) else 1)\n"}},"limitations":"This is a finite toy beam-state contract and makes no claim about neural model quality or production decoding. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.","method":"Deterministic executable model with adversarial boundary fixtures.","provenance":{"created_by":"Failure Map","dependencies":"Python standard library","family":"w3-beam-search-bookkeeping-early-stop-bound","generated_at":"2026-09-30T16:48:13.668747+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"Beam decoders rely on consistent hypothesis state while expanding and ranking candidates.","root_cause":"Stopping compares current live scores directly with finished scores despite nonpositive future deltas.","sha256":"2f52c5fadff7d5c4c69354ee8b7fc64cfd3583d3ebefc283e1b38cdf1dbb8b9c","title":"Beam search: search stops before a live path can beat the finished set · case 01","variant":1,"variant_policy":"The fixture code references N, supplied as 1 through 5. Fixed boundary checks may repeat. These records share one mechanism and must remain together in evaluation splits.","verified":true,"visibility":"public","verification":{"attempt":{"elapsed_ms":28.07,"exit_code":1,"observations":[{"actual":true,"check":"one result can stop","expected":true,"passed":true},{"actual":false,"check":"live still better","expected":false,"passed":true},{"actual":false,"check":"need k results","expected":false,"passed":true},{"actual":true,"check":"kth threshold","expected":true,"passed":true},{"actual":false,"check":"below kth threshold","expected":false,"passed":true},{"actual":false,"check":"first result still improvable","expected":true,"passed":false},{"actual":true,"check":"no live paths","expected":true,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"one result can stop\", \"actual\": true, \"expected\": true, \"passed\": true}, {\"check\": \"live still better\", \"actual\": false, \"expected\": false, \"passed\": true}, {\"check\": \"need k results\", \"actual\": false, \"expected\": false, \"passed\": true}, {\"check\": \"kth threshold\", \"actual\": true, \"expected\": true, \"passed\": true}, {\"check\": \"below kth threshold\", \"actual\": false, \"expected\": false, \"passed\": true}, {\"check\": \"first result still improvable\", \"actual\": false, \"expected\": true, \"passed\": false}, {\"check\": \"no live paths\", \"actual\": true, \"expected\": true, \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.393,"exit_code":1,"observations":[{"actual":true,"check":"one result can stop","expected":true,"passed":true},{"actual":false,"check":"live still better","expected":false,"passed":true},{"actual":true,"check":"need k results","expected":false,"passed":false},{"actual":true,"check":"kth threshold","expected":true,"passed":true},{"actual":true,"check":"below kth threshold","expected":false,"passed":false},{"actual":true,"check":"first result still improvable","expected":true,"passed":true},{"actual":true,"check":"no live paths","expected":true,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"one result can stop\", \"actual\": true, \"expected\": true, \"passed\": true}, {\"check\": \"live still better\", \"actual\": false, \"expected\": false, \"passed\": true}, {\"check\": \"need k results\", \"actual\": true, \"expected\": false, \"passed\": false}, {\"check\": \"kth threshold\", \"actual\": true, \"expected\": true, \"passed\": true}, {\"check\": \"below kth threshold\", \"actual\": true, \"expected\": false, \"passed\": false}, {\"check\": \"first result still improvable\", \"actual\": true, \"expected\": true, \"passed\": true}, {\"check\": \"no live paths\", \"actual\": true, \"expected\": true, \"passed\": true}], \"passed\": false}\n"}},"member_only":{"stages":["fixed"],"fields":["implementations.fixed","verification.fixed","harness","repair"],"note":"The verified repair, its recorded checks, the repair description, and the scoring harness are available to members."}}