File size: 7,267 Bytes
f325414
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
"""Aggregate CEC-2022 sweep results and render all paper figures + tables.

Reads every ``result.json`` under ``outputs/runs/cec2022/`` (or a
user-supplied directory), pivots into per-(algorithm, function, dim)
arrays of final best fitness, then writes:

* ``outputs/tables/cec2022_mean_std_d{dim}.tex`` — Table 1
* ``outputs/tables/cec2022_wilcoxon_d{dim}.tex`` — Table 2
* ``outputs/figures/cec2022_friedman.json`` — Friedman omnibus per dim
* ``outputs/figures/convergence_d{dim}_{func}.{pdf,png}`` — Figure 1
* ``outputs/figures/wilcoxon_heatmap_d{dim}.{pdf,png}`` — Figure 4
"""

from __future__ import annotations

import argparse
import json
from collections import defaultdict
from pathlib import Path
from typing import Any

import numpy as np
import pandas as pd

from ahdcma.stats.tests import friedman_test, wilcoxon_pairwise
from ahdcma.viz.ablation import plot_pairwise_pvalue_heatmap
from ahdcma.viz.convergence import plot_convergence
from ahdcma.viz.tables import mean_std_table, wilcoxon_pvalue_table


def _load_results(root: Path) -> list[dict[str, Any]]:
    out: list[dict[str, Any]] = []
    for jf in root.rglob("result.json"):
        try:
            data = json.loads(jf.read_text())
            out.append(data)
        except json.JSONDecodeError:
            continue
    return out


def _pivot(
    results: list[dict[str, Any]],
) -> dict[int, dict[tuple[str, str], list[float]]]:
    """Group final best_f by dim, then by (algo, func)."""
    by_dim: dict[int, dict[tuple[str, str], list[float]]] = defaultdict(lambda: defaultdict(list))
    for r in results:
        by_dim[int(r["dim"])][r["algo"], r["func"]].append(float(r["best_f"]))
    return by_dim


def _convergence_curves(
    results: list[dict[str, Any]], dim: int, func: str
) -> dict[str, list[list[float]]]:
    by_algo: dict[str, list[list[float]]] = defaultdict(list)
    for r in results:
        if int(r["dim"]) != dim or r["func"] != func:
            continue
        by_algo[r["algo"]].append(list(r["best_fitness_curve"]))
    return by_algo


def render_all(
    sweep_root: str | Path = "outputs/runs/cec2022",
    fig_dir: str | Path = "outputs/figures",
    tab_dir: str | Path = "outputs/tables",
    *,
    headline_funcs: list[str] | None = None,
) -> None:
    fig_dir = Path(fig_dir)
    tab_dir = Path(tab_dir)
    fig_dir.mkdir(parents=True, exist_ok=True)
    tab_dir.mkdir(parents=True, exist_ok=True)

    results = _load_results(Path(sweep_root))
    if not results:
        print(f"No result.json files under {sweep_root}; nothing to render.")
        return
    by_dim = _pivot(results)

    headline_funcs = headline_funcs or [
        "F1_zakharov",
        "F5_levy",
        "F9_composition1",
    ]

    summary: dict[str, Any] = {"n_results": len(results), "by_dim": {}}
    for dim, table in by_dim.items():
        algos = sorted({a for a, _ in table})
        funcs = sorted({f for _, f in table})
        # Mean / std table
        mean_df = pd.DataFrame(index=funcs, columns=algos, dtype=object)
        for f in funcs:
            for a in algos:
                arr = table.get((a, f), [])
                if arr:
                    mean_df.at[f, a] = (float(np.mean(arr)), float(np.std(arr)))
                else:
                    mean_df.at[f, a] = np.nan
        mean_std_table(
            mean_df,
            tab_dir / f"cec2022_mean_std_d{dim}.tex",
            caption=f"CEC-2022 mean $\\pm$ std final fitness ({dim}-D, 30 seeds).",
            label=f"tab:cec2022_mean_std_d{dim}",
        )

        # Wilcoxon vs AHD-CMA per function
        if "ahdcma" in algos:
            p_rows: list[dict[str, Any]] = []
            for f in funcs:
                bundle = {a: np.asarray(table[a, f]) for a in algos if (a, f) in table}
                if "ahdcma" not in bundle or len(bundle) < 2:
                    continue
                seeds_min = min(len(v) for v in bundle.values())
                bundle = {k: v[:seeds_min] for k, v in bundle.items()}
                try:
                    df = wilcoxon_pairwise(bundle, baseline_name="ahdcma")
                except ValueError:
                    continue
                row = {"function": f}
                for _, rr in df.iterrows():
                    row[rr["algo"]] = rr["p_value"]
                p_rows.append(row)
            if p_rows:
                p_df = pd.DataFrame(p_rows).set_index("function")
                wilcoxon_pvalue_table(
                    p_df,
                    tab_dir / f"cec2022_wilcoxon_d{dim}.tex",
                    caption=f"Wilcoxon p-values vs AHD-CMA on CEC-2022 ({dim}-D).",
                    label=f"tab:wilcoxon_d{dim}",
                )

                # Per-algo p-value heatmap (function x algo, log10)
                p_for_heat = p_df.dropna(how="all")
                if not p_for_heat.empty:
                    plot_pairwise_pvalue_heatmap(
                        p_for_heat,
                        fig_dir / f"wilcoxon_heatmap_d{dim}",
                        title=f"Wilcoxon p-values vs AHD-CMA, CEC-2022 {dim}-D",
                    )

        # Friedman omnibus across functions
        # Stack into a matrix where each row is a function, each column an algo.
        from numpy.typing import NDArray

        friedman_in: dict[str, NDArray[np.float64]] = {}
        for a in algos:
            cols = []
            for f in funcs:
                arr = table.get((a, f), [])
                if not arr:
                    cols.append(np.nan)
                    continue
                cols.append(float(np.median(arr)))
            friedman_in[a] = np.asarray(cols, dtype=np.float64)
        # Drop functions with any NaN to keep equal-length input
        valid = ~np.any(np.stack([np.isnan(v) for v in friedman_in.values()], axis=0), axis=0)
        friedman_in = {k: v[valid] for k, v in friedman_in.items()}
        if all(arr.size >= 3 for arr in friedman_in.values()):
            stat, p = friedman_test(friedman_in)
        else:
            stat, p = float("nan"), float("nan")

        summary["by_dim"][str(dim)] = {
            "n_funcs": len(funcs),
            "n_algos": len(algos),
            "friedman_statistic": stat,
            "friedman_p_value": p,
        }

        # Convergence figures for the headline functions
        for f in headline_funcs:
            curves = _convergence_curves(results, dim, f)
            if not curves:
                continue
            plot_convergence(
                curves,
                fig_dir / f"convergence_d{dim}_{f}",
                title=f"{f} ({dim}-D)",
                xlabel="Generation",
                ylabel="Best fitness",
                log_y=True,
            )

    summary_path = fig_dir / "cec2022_summary.json"
    summary_path.write_text(json.dumps(summary, indent=2))
    print(json.dumps(summary, indent=2))


def main() -> None:
    p = argparse.ArgumentParser()
    p.add_argument("--sweep-root", default="outputs/runs/cec2022")
    p.add_argument("--fig-dir", default="outputs/figures")
    p.add_argument("--tab-dir", default="outputs/tables")
    args = p.parse_args()
    render_all(args.sweep_root, args.fig_dir, args.tab_dir)


if __name__ == "__main__":
    main()