{"abstract":"Decoupled decay contaminates gradient moments.","category":"Optimizer update state","checks":6,"contract":"For zero prior state return [p*(1-lr*wd)-lr*g,g,g*g].","contract_signature":"p, g, lr, wd","evaluation_group":"optimizer-toy-decoupled-weight-decay","failed_approach":"Applying multiplicative decay and then subtracting a decay-augmented gradient double counts decay.","family":"w3-optimizer-update-state-decoupled-weight-decay","id":"FA-99006","implementations":{"attempt":{"sha256":"7d5697764d7664ddaa9942fac64a60438fbca75c252562dcecf102f341813f56","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(p, g, lr, wd):\n    q=g+wd*p\n    return [p*(1-lr*wd)-lr*q,q,q*q]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('decay only',solve(4,0,0.5,0.25),[3.5,0,0])\ncheck('gradient state untouched',solve(4,2,0.5,0.25),[2.5,2,4])\ncheck('zero decay',solve(3,-2,1,0),[5,-2,4])\ncheck('zero rate',solve(3,2,0,1),[3,2,4])\ncheck('negative parameter',solve(-4,0,0.5,0.25),[-3.5,0,0])\ncheck('combined sign',solve(2,-1,1,0.5),[2,-1,1])\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":"90481c345a705e047578045f8be2ab3565881fab37c4b64880fc4284b4a55138","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(p, g, lr, wd):\n    q=g+wd*p\n    return [p-lr*q,q,q*q]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('decay only',solve(4,0,0.5,0.25),[3.5,0,0])\ncheck('gradient state untouched',solve(4,2,0.5,0.25),[2.5,2,4])\ncheck('zero decay',solve(3,-2,1,0),[5,-2,4])\ncheck('zero rate',solve(3,2,0,1),[3,2,4])\ncheck('negative parameter',solve(-4,0,0.5,0.25),[-3.5,0,0])\ncheck('combined sign',solve(2,-1,1,0.5),[2,-1,1])\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":"These are deterministic scalar or short-vector toy contracts, not claims about production optimizer APIs or training quality. 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-optimizer-update-state-decoupled-weight-decay","generated_at":"2026-09-30T16:48:19.455427+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"Optimizer implementations must associate each parameter with its own mathematically specified update state.","root_cause":"Parameter decay is inserted into the gradient accumulator.","sha256":"9e8c554b15444b76138811c931a7469f7baf19a60b8ee4d0b6e06f5f89c5f6e7","title":"Decoupled decay contaminates gradient moments · case 01","variant":1,"variant_policy":"The same explicitly authored fixtures repeat across all five numbered records; runtime N is unused. These records share one mechanism and must remain together in evaluation splits.","verified":true,"visibility":"public","verification":{"attempt":{"elapsed_ms":29.457,"exit_code":1,"observations":[{"actual":[3.0,1.0,1.0],"check":"decay only","expected":[3.5,0,0],"passed":false},{"actual":[2.0,3.0,9.0],"check":"gradient state untouched","expected":[2.5,2,4],"passed":false},{"actual":[5,-2,4],"check":"zero decay","expected":[5,-2,4],"passed":true},{"actual":[3,5,25],"check":"zero rate","expected":[3,2,4],"passed":false},{"actual":[-3.0,-1.0,1.0],"check":"negative parameter","expected":[-3.5,0,0],"passed":false},{"actual":[1.0,0.0,0.0],"check":"combined sign","expected":[2,-1,1],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"decay only\", \"actual\": [3.0, 1.0, 1.0], \"expected\": [3.5, 0, 0], \"passed\": false}, {\"check\": \"gradient state untouched\", \"actual\": [2.0, 3.0, 9.0], \"expected\": [2.5, 2, 4], \"passed\": false}, {\"check\": \"zero decay\", \"actual\": [5, -2, 4], \"expected\": [5, -2, 4], \"passed\": true}, {\"check\": \"zero rate\", \"actual\": [3, 5, 25], \"expected\": [3, 2, 4], \"passed\": false}, {\"check\": \"negative parameter\", \"actual\": [-3.0, -1.0, 1.0], \"expected\": [-3.5, 0, 0], \"passed\": false}, {\"check\": \"combined sign\", \"actual\": [1.0, 0.0, 0.0], \"expected\": [2, -1, 1], \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.653,"exit_code":1,"observations":[{"actual":[3.5,1.0,1.0],"check":"decay only","expected":[3.5,0,0],"passed":false},{"actual":[2.5,3.0,9.0],"check":"gradient state untouched","expected":[2.5,2,4],"passed":false},{"actual":[5,-2,4],"check":"zero decay","expected":[5,-2,4],"passed":true},{"actual":[3,5,25],"check":"zero rate","expected":[3,2,4],"passed":false},{"actual":[-3.5,-1.0,1.0],"check":"negative parameter","expected":[-3.5,0,0],"passed":false},{"actual":[2.0,0.0,0.0],"check":"combined sign","expected":[2,-1,1],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"decay only\", \"actual\": [3.5, 1.0, 1.0], \"expected\": [3.5, 0, 0], \"passed\": false}, {\"check\": \"gradient state untouched\", \"actual\": [2.5, 3.0, 9.0], \"expected\": [2.5, 2, 4], \"passed\": false}, {\"check\": \"zero decay\", \"actual\": [5, -2, 4], \"expected\": [5, -2, 4], \"passed\": true}, {\"check\": \"zero rate\", \"actual\": [3, 5, 25], \"expected\": [3, 2, 4], \"passed\": false}, {\"check\": \"negative parameter\", \"actual\": [-3.5, -1.0, 1.0], \"expected\": [-3.5, 0, 0], \"passed\": false}, {\"check\": \"combined sign\", \"actual\": [2.0, 0.0, 0.0], \"expected\": [2, -1, 1], \"passed\": false}], \"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."}}