{"abstract":"Frozen parameters accumulate optimizer state.","category":"Optimizer update state","checks":6,"contract":"If inactive return [p,m,v]. Otherwise set nm=.5*m+.5*g, nv=.5*v+.5*g*g, and return [p-nm/(sqrt(nv)+1),nm,nv]; v is nonnegative.","contract_signature":"p, g, m, v, active","evaluation_group":"optimizer-toy-frozen-parameter-state","failed_approach":"Skipping inactive moments repairs freezing but leaves active parameter values unchanged despite their new moments.","family":"w3-optimizer-update-state-frozen-parameter-state","id":"FA-99016","implementations":{"attempt":{"sha256":"c31419ec432bd5206ac380d25f7fb7d4b52006d7d79def7a6aa39e4f76cde0c4","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(p, g, m, v, active):\n    nm=m if not active else 0.5*m+0.5*g; nv=v if not active else 0.5*v+0.5*g*g\n    return [p,nm,nv]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('frozen nonzero gradient',solve(3,4,2,6,False),[3,2,6])\ncheck('active update',solve(3,4,2,6,True),[3-3/(11**0.5+1),3,11])\ncheck('frozen negative gradient',solve(-2,-4,5,8,False),[-2,5,8])\ncheck('frozen zero state',solve(0,0,0,0,False),[0,0,0])\ncheck('active zero gradient',solve(5,0,4,8,True),[5-2/(4**0.5+1),2,4])\ncheck('active singleton',solve(N,2,0,0,True),[N-1/(2**0.5+1),1,2])\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":"89c29fee0b6004462e0683b21b648b3b0fd1f88582fe784f2773b588adb85079","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(p, g, m, v, active):\n    nm=0.5*m+0.5*g; nv=0.5*v+0.5*g*g\n    return [p,nm,nv]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('frozen nonzero gradient',solve(3,4,2,6,False),[3,2,6])\ncheck('active update',solve(3,4,2,6,True),[3-3/(11**0.5+1),3,11])\ncheck('frozen negative gradient',solve(-2,-4,5,8,False),[-2,5,8])\ncheck('frozen zero state',solve(0,0,0,0,False),[0,0,0])\ncheck('active zero gradient',solve(5,0,4,8,True),[5-2/(4**0.5+1),2,4])\ncheck('active singleton',solve(N,2,0,0,True),[N-1/(2**0.5+1),1,2])\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-frozen-parameter-state","generated_at":"2026-09-30T16:48:19.456305+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":"Gradient moments are updated without consulting the active flag, including when a parameter is explicitly frozen.","sha256":"8554241a0ba0746611a5e8d924f18e79f4b22976ff249324af3828246c6f9958","title":"Frozen parameters accumulate optimizer state · 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":28.479,"exit_code":1,"observations":[{"actual":[3,2,6],"check":"frozen nonzero gradient","expected":[3,2,6],"passed":true},{"actual":[3,3.0,11.0],"check":"active update","expected":[2.30501256289338,3,11],"passed":false},{"actual":[-2,5,8],"check":"frozen negative gradient","expected":[-2,5,8],"passed":true},{"actual":[0,0,0],"check":"frozen zero state","expected":[0,0,0],"passed":true},{"actual":[5,2.0,4.0],"check":"active zero gradient","expected":[4.333333333333333,2,4],"passed":false},{"actual":[1,1.0,2.0],"check":"active singleton","expected":[0.5857864376269049,1,2],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"frozen nonzero gradient\", \"actual\": [3, 2, 6], \"expected\": [3, 2, 6], \"passed\": true}, {\"check\": \"active update\", \"actual\": [3, 3.0, 11.0], \"expected\": [2.30501256289338, 3, 11], \"passed\": false}, {\"check\": \"frozen negative gradient\", \"actual\": [-2, 5, 8], \"expected\": [-2, 5, 8], \"passed\": true}, {\"check\": \"frozen zero state\", \"actual\": [0, 0, 0], \"expected\": [0, 0, 0], \"passed\": true}, {\"check\": \"active zero gradient\", \"actual\": [5, 2.0, 4.0], \"expected\": [4.333333333333333, 2, 4], \"passed\": false}, {\"check\": \"active singleton\", \"actual\": [1, 1.0, 2.0], \"expected\": [0.5857864376269049, 1, 2], \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.9,"exit_code":1,"observations":[{"actual":[3,3.0,11.0],"check":"frozen nonzero gradient","expected":[3,2,6],"passed":false},{"actual":[3,3.0,11.0],"check":"active update","expected":[2.30501256289338,3,11],"passed":false},{"actual":[-2,0.5,12.0],"check":"frozen negative gradient","expected":[-2,5,8],"passed":false},{"actual":[0,0.0,0.0],"check":"frozen zero state","expected":[0,0,0],"passed":true},{"actual":[5,2.0,4.0],"check":"active zero gradient","expected":[4.333333333333333,2,4],"passed":false},{"actual":[1,1.0,2.0],"check":"active singleton","expected":[0.5857864376269049,1,2],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"frozen nonzero gradient\", \"actual\": [3, 3.0, 11.0], \"expected\": [3, 2, 6], \"passed\": false}, {\"check\": \"active update\", \"actual\": [3, 3.0, 11.0], \"expected\": [2.30501256289338, 3, 11], \"passed\": false}, {\"check\": \"frozen negative gradient\", \"actual\": [-2, 0.5, 12.0], \"expected\": [-2, 5, 8], \"passed\": false}, {\"check\": \"frozen zero state\", \"actual\": [0, 0.0, 0.0], \"expected\": [0, 0, 0], \"passed\": true}, {\"check\": \"active zero gradient\", \"actual\": [5, 2.0, 4.0], \"expected\": [4.333333333333333, 2, 4], \"passed\": false}, {\"check\": \"active singleton\", \"actual\": [1, 1.0, 2.0], \"expected\": [0.5857864376269049, 1, 2], \"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."}}