File size: 6,404 Bytes
c6eaad2
 
 
 
 
 
 
e7c9bd3
c6eaad2
 
e7c9bd3
 
c6eaad2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e7c9bd3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c6eaad2
 
 
 
 
 
 
 
 
 
 
 
e7c9bd3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c6eaad2
 
 
 
 
 
 
 
 
 
 
e7c9bd3
 
 
 
 
 
c6eaad2
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
"""Independent counterexample and raw-audit checker for Claim 4."""

from __future__ import annotations

import json
import re
import xml.etree.ElementTree as ET
from itertools import combinations
from pathlib import Path

import numpy as np


GRAPHML_NS = "http://graphml.graphdrawing.org/xmlns"
SUBJECT_RE = re.compile(r"^sub-(OAS\d+)_ses-d(\d+)_")


def independent_ids(path: Path) -> list[int]:
    root = ET.parse(path).getroot()
    key_names = {
        element.attrib["id"]: element.attrib.get("attr.name")
        for element in root.findall(f"{{{GRAPHML_NS}}}key")
    }
    ids: list[int] = []
    for node in root.iter(f"{{{GRAPHML_NS}}}node"):
        values = [
            data.text
            for data in node.findall(f"{{{GRAPHML_NS}}}data")
            if key_names.get(data.attrib.get("key", ""))
            == "dn_multiscaleID"
        ]
        if len(values) != 1 or values[0] is None:
            raise ValueError("independent parser found an ambiguous node ID")
        ids.append(int(values[0]))
    return sorted(ids)


def check(runtime_dir: Path) -> dict[str, object]:
    primary = json.loads((runtime_dir / "claim_4_result.json").read_text())
    session_rows: list[dict[str, object]] = []
    for path in sorted(runtime_dir.glob("claim_4_session_audit_*.json")):
        session_rows.extend(json.loads(path.read_text()))
    subject_rows = json.loads(
        (runtime_dir / "claim_4_subject_audit.json").read_text()
    )
    files = list((runtime_dir / "work" / "extracted").rglob("*.graphml"))
    counterexample_name = primary["data_provenance"]["counterexample_file"]
    counterexample_paths = [
        path for path in files if path.name == counterexample_name
    ]
    if len(counterexample_paths) != 1:
        raise RuntimeError("counterexample file is not uniquely materialized")
    ids = independent_ids(counterexample_paths[0])
    subject_counts: dict[str, int] = {}
    for row in session_rows:
        match = SUBJECT_RE.match(str(row["file"]))
        if not match or match.group(1) != row["subject"]:
            raise RuntimeError("session filename/subject mismatch")
        subject_counts[str(row["subject"])] = (
            subject_counts.get(str(row["subject"]), 0) + 1
        )
    recorded_invalid = [
        row for row in session_rows if row["subject"] == "OAS30938"
    ]
    table3 = json.loads(
        (runtime_dir / "claim_4_table3_result.json").read_text()
    )
    pair_rows: list[dict[str, object]] = []
    for path in sorted(runtime_dir.glob("claim_4_table3_pairs_*.json")):
        pair_rows.extend(json.loads(path.read_text()))
    expected_pairs = {
        (left, right) for left, right in combinations(range(100), 2)
    }
    subject_order = [
        str(row["subject"]) for row in table3["selected_subjects"]
    ]
    observed_pairs = {
        (
            subject_order.index(str(row["left_subject"])),
            subject_order.index(str(row["right_subject"])),
        )
        for row in pair_rows
    }
    independent_summary: dict[str, float] = {}
    for metric in ("diffusion", "geodesic"):
        for method in ("CDOT", "FGW"):
            values = np.asarray(
                [
                    float(row["accuracy"])
                    for row in pair_rows
                    if row["metric"] == metric
                    and row["method"] == method
                ]
            )
            independent_summary[f"{metric}_{method}"] = float(values.mean())
    recorded_summary = {
        key: float(value["mean_accuracy"])
        for key, value in table3["rerun"].items()
    }
    summary_error = max(
        abs(independent_summary[key] - recorded_summary[key])
        for key in independent_summary
    )
    diffusion_margin = (
        independent_summary["diffusion_CDOT"]
        - independent_summary["diffusion_FGW"]
    )
    geodesic_margin = (
        independent_summary["geodesic_FGW"]
        - independent_summary["geodesic_CDOT"]
    )
    gates = {
        "all_chunked_sessions_loaded": len(session_rows) == 975,
        "all_subject_rows_loaded": len(subject_rows) == 696,
        "filename_parser_independently_finds_696_subjects": len(subject_counts)
        == 696,
        "counterexample_has_one_session": len(recorded_invalid) == 1,
        "independent_xml_parser_finds_exact_ids_1_to_168": ids
        == list(range(1, 169)),
        "no_counterexample_session_contains_169_or_170": 169 not in ids
        and 170 not in ids,
        "primary_falsification_status_rechecked": primary["status"]
        == "FALSIFIED",
        "table3_all_19800_method_metric_rows_loaded": len(pair_rows)
        == 4_950 * 4,
        "table3_exact_4950_pair_inventory": observed_pairs
        == expected_pairs,
        "table3_each_cell_has_4950_rows": all(
            sum(
                row["metric"] == metric and row["method"] == method
                for row in pair_rows
            )
            == 4_950
            for metric in ("diffusion", "geodesic")
            for method in ("CDOT", "FGW")
        ),
        "table3_summary_recomputed_from_raw_rows": summary_error < 1e-9,
        "table3_diffusion_direction_rechecked": diffusion_margin > 0,
        "table3_geodesic_direction_rechecked": geodesic_margin > 0,
        "table3_primary_gates_rechecked": table3["all_gates_pass"],
    }
    result = {
        "checker": (
            "independent filename inventory plus direct ElementTree reparse "
            "of the primary-archive counterexample"
        ),
        "counterexample_file": counterexample_name,
        "independent_node_count": len(ids),
        "independent_min_id": min(ids),
        "independent_max_id": max(ids),
        "independent_subject_count": len(subject_counts),
        "table3_raw_rows": len(pair_rows),
        "table3_unique_pairs": len(observed_pairs),
        "table3_independent_summary": independent_summary,
        "table3_summary_max_abs_error": summary_error,
        "table3_diffusion_CDOT_minus_FGW": diffusion_margin,
        "table3_geodesic_FGW_minus_CDOT": geodesic_margin,
        "gates": gates,
        "all_gates_pass": all(gates.values()),
    }
    (runtime_dir / "claim_4_independent_checker.json").write_text(
        json.dumps(result, indent=2) + "\n", encoding="utf-8"
    )
    if not result["all_gates_pass"]:
        raise RuntimeError("Independent Claim 4 checker failed")
    return result