FAILURE MAP
← Case archive

FA-14136 / Numerical aggregation / Open access

Frequency cosine squared: Norms discard coordinates absent from the other sample. · case 01

The reduction disagrees with its explicit aggregation oracle.

Verified by executionVariant 1 · 8 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Norms discard coordinates absent from the other sample.

VERIFIED REPAIR

Preserve the frequency cosine squared contract at the identified reduction decision.

Unsuccessful approach: Dropping exclusive mass from just one norm still overstates similarity.

Case contract

Treat samples as frequency vectors over nominal integer labels. Return squared cosine dot**2/(sum count_a**2 * sum count_b**2) as exact Fraction string. A zero norm on either side returns None.

Why this case matters

Exact bounded examples isolate a reduction defect without floating-point or external-service effects.

1 / The failure

Exit 1
"""Failure Map reference implementation. Python standard library only."""
import json
from fractions import Fraction
from collections import Counter, defaultdict
import math
import itertools
N = 1
observations = []
def solve(a, b):
    ca,cb=Counter(a),Counter(b)
    keys=set(ca)|set(cb)
    dot=sum(ca[k]*cb[k] for k in keys)
    na=sum(ca[k]**2 for k in keys if cb[k])
    nb=sum(cb[k]**2 for k in keys if ca[k])
    return str(Fraction(dot*dot,na*nb)) if na and nb else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('regression 1', solve(*([1, 1, 2], [1, 2, 2])), '16/25')
check('regression 2', solve(*([1, 2], [1, 2])), '1')
check('regression 3', solve(*([], [])), None)
check('regression 4', solve(*([0], [1])), '0')
check('regression 5', solve(*([1, 1, 1, 2], [1, 2])), '4/5')
check('regression 6', solve(*([-1, 0, 0], [0, 0, 1, 1])), '2/5')
check('regression 7', solve(*([2, 2], [])), None)
check("variable cosine concentration",solve([0]*N+[1],[0]),str(Fraction(N*N,N*N+1)))
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
Boundary fixtureActualExpectedOutcome
regression 116/2516/25Passed
regression 211Passed
regression 3NoneNonePassed
regression 4None0Failed
regression 54/54/5Passed
regression 612/5Failed
regression 7NoneNonePassed
variable cosine concentration11/2Failed

SHA-256 / bdb93abdd5dcfa0e3fee469d285e96ed0185358aa347c2d697d47a1282f782ae

2 / The unsuccessful fix

Exit 1
"""Failure Map reference implementation. Python standard library only."""
import json
from fractions import Fraction
from collections import Counter, defaultdict
import math
import itertools
N = 1
observations = []
def solve(a, b):
    ca,cb=Counter(a),Counter(b)
    keys=set(ca)|set(cb)
    dot=sum(ca[k]*cb[k] for k in keys)
    na=sum(ca[k]**2 for k in keys if cb[k])
    nb=sum(v*v for v in cb.values())
    return str(Fraction(dot*dot,na*nb)) if na and nb else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('regression 1', solve(*([1, 1, 2], [1, 2, 2])), '16/25')
check('regression 2', solve(*([1, 2], [1, 2])), '1')
check('regression 3', solve(*([], [])), None)
check('regression 4', solve(*([0], [1])), '0')
check('regression 5', solve(*([1, 1, 1, 2], [1, 2])), '4/5')
check('regression 6', solve(*([-1, 0, 0], [0, 0, 1, 1])), '2/5')
check('regression 7', solve(*([2, 2], [])), None)
check("variable cosine concentration",solve([0]*N+[1],[0]),str(Fraction(N*N,N*N+1)))
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
Boundary fixtureActualExpectedOutcome
regression 116/2516/25Passed
regression 211Passed
regression 3NoneNonePassed
regression 4None0Failed
regression 54/54/5Passed
regression 61/22/5Failed
regression 7NoneNonePassed
variable cosine concentration11/2Failed

SHA-256 / 17554ab6061af811cab0330293e6c9fe900ee7e54ba177a901a8360eb4bd1ebf

3 / The verified repair

Exit 0
"""Failure Map reference implementation. Python standard library only."""
import json
from fractions import Fraction
from collections import Counter, defaultdict
import math
import itertools
N = 1
observations = []
def solve(a, b):
    ca,cb=Counter(a),Counter(b)
    keys=set(ca)|set(cb)
    dot=sum(ca[k]*cb[k] for k in keys)
    na=sum(v*v for v in ca.values())
    nb=sum(v*v for v in cb.values())
    return str(Fraction(dot*dot,na*nb)) if na and nb else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('regression 1', solve(*([1, 1, 2], [1, 2, 2])), '16/25')
check('regression 2', solve(*([1, 2], [1, 2])), '1')
check('regression 3', solve(*([], [])), None)
check('regression 4', solve(*([0], [1])), '0')
check('regression 5', solve(*([1, 1, 1, 2], [1, 2])), '4/5')
check('regression 6', solve(*([-1, 0, 0], [0, 0, 1, 1])), '2/5')
check('regression 7', solve(*([2, 2], [])), None)
check("variable cosine concentration",solve([0]*N+[1],[0]),str(Fraction(N*N,N*N+1)))
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
Boundary fixtureActualExpectedOutcome
regression 116/2516/25Passed
regression 211Passed
regression 3NoneNonePassed
regression 400Passed
regression 54/54/5Passed
regression 62/52/5Passed
regression 7NoneNonePassed
variable cosine concentration1/21/2Passed

SHA-256 / 15037d025e834f16d75cb1853d682a3dbb64d80828dd074813a99eb78afcb9d1

Verification & scope

Small offline integer/rational inputs only; no performance, statistical inference, or production-library conformance claim. 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.

Observations recorded using Python 3.12.14 at 2026-09-29T14:39:13.849630+00:00.

Case digest / d7446da4d7cf15e599aca369973fce9203d114cc01317adeba9c83e3182fb3b4