FA-100481 / Sparse tensor canonicalization / Open access

Sparse tensor: column reduction confuses row positions with column indices · case 01

column reduction confuses row positions with column indices.

Verified by executionVariant 1 · 6 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Coordinate values are accumulated into output positions determined by their column labels, not their storage index.

THE FAILURE

Coordinate values are accumulated into output positions determined by their column labels, not their storage index.

Unsuccessful approach: A strict prefix count still excludes contributions whose label equals the output column.

Case contract

Return a dense list of width column sums, adding each value at its stored column.

Why this case matters

A small offline sparse-array operation has explicit coordinate and storage semantics.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(columns,values,width):
    return [sum(values[i] for i,c in enumerate(columns) if i==j) for j in range(width)]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('reordered columns',solve([2,0,2],[N,3,4],3),[3,0,N+4])
check('empty columns',solve([],[],N),[0]*N)
check('single bucket',solve([0],[7],1),[7])
check('repeated bucket',solve([1,1],[N,2],3),[0,N+2,0])
check('negative contribution',solve([0,2],[-N,5],3),[-N,0,5])
check('no entries',solve([],[],0),[])
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 fixture	Actual	Expected	Outcome
reordered columns	`[1, 3, 4]`	`[3, 0, 5]`	Failed
empty columns	`[0]`	`[0]`	Passed
single bucket	`[7]`	`[7]`	Passed
repeated bucket	`[1, 2, 0]`	`[0, 3, 0]`	Failed
negative contribution	`[-1, 5, 0]`	`[-1, 0, 5]`	Failed
no entries	`[]`	`[]`	Passed

SHA-256 / 1d8bf771690e3e8bd5451803bd7d0c5878d11b51b031718a5f3fdd37eda0a307

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(columns,values,width):
    return [sum(v for c,v in zip(columns,values) if c<j) for j in range(width)]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('reordered columns',solve([2,0,2],[N,3,4],3),[3,0,N+4])
check('empty columns',solve([],[],N),[0]*N)
check('single bucket',solve([0],[7],1),[7])
check('repeated bucket',solve([1,1],[N,2],3),[0,N+2,0])
check('negative contribution',solve([0,2],[-N,5],3),[-N,0,5])
check('no entries',solve([],[],0),[])
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 fixture	Actual	Expected	Outcome
reordered columns	`[0, 3, 3]`	`[3, 0, 5]`	Failed
empty columns	`[0]`	`[0]`	Passed
single bucket	`[0]`	`[7]`	Failed
repeated bucket	`[0, 0, 3]`	`[0, 3, 0]`	Failed
negative contribution	`[0, -1, -1]`	`[-1, 0, 5]`	Failed
no entries	`[]`	`[]`	Passed

SHA-256 / 51096166fd0a58b053951bff3b15c59ed82007bcd00dd79e48050f53189bb378

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

These are toy integer-valued COO or CSR contracts; they make no claim about a framework API, floating point policy, or large tensor performance. 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 2026-09-30T16:48:36.241690+00:00.

Case digest / 547dc1722b40b1bba6eebe1b5f798d0b627c45eba93ff11acf743bb6ec7d48f9