geographdb-core 0.5.4

Geometric graph database core - 3D spatial indexing for code analysis
Documentation 
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

#!/usr/bin/env python3
"""Plot the Pareto curve from ffn_confidence_routing.csv."""

import sys
from pathlib import Path

import matplotlib.pyplot as plt
import numpy as np


def main() -> None:
    if len(sys.argv) < 2:
        csv_path = Path("target/ffn_confidence_routing.csv")
    else:
        csv_path = Path(sys.argv[1])

    if not csv_path.exists():
        raise SystemExit(f"CSV not found: {csv_path}")

    data = np.genfromtxt(csv_path, delimiter=",", names=True, dtype=None, encoding=None)
    rank = int(data["rank"][0])
    mode = str(data["mode"][0])
    frac = data["fraction_low_rank"]
    speedup = data["estimated_speedup"]
    l2 = data["logit_l2"]
    top1 = data["top1_agreement"]

    fig, axes = plt.subplots(1, 3, figsize=(15, 4.5))

    ax = axes[0]
    ax.plot(frac * 100.0, l2, "o-", color="tab:blue")
    ax.set_xlabel("samples routed through low-rank (%)")
    ax.set_ylabel("mean logit L2 drift")
    ax.set_title(f"rank={rank}, mode={mode}")
    ax.invert_xaxis()
    ax.grid(True, ls="--", alpha=0.4)
    for x, y, t in zip(frac * 100.0, l2, data["threshold"]):
        ax.annotate(f"{t:.2f}", (x, y), textcoords="offset points", xytext=(5, -10), fontsize=7)

    ax = axes[1]
    ax.plot(frac * 100.0, top1 * 100.0, "o-", color="tab:green")
    ax.set_xlabel("samples routed through low-rank (%)")
    ax.set_ylabel("top-1 agreement (%)")
    ax.set_ylim(0, 105)
    ax.set_title("accuracy vs routing fraction")
    ax.invert_xaxis()
    ax.grid(True, ls="--", alpha=0.4)
    for x, y, t in zip(frac * 100.0, top1 * 100.0, data["threshold"]):
        ax.annotate(f"{t:.2f}", (x, y), textcoords="offset points", xytext=(5, -10), fontsize=7)

    ax = axes[2]
    ax.plot(speedup, top1 * 100.0, "o-", color="tab:orange")
    ax.set_xlabel("estimated speedup (×)")
    ax.set_ylabel("top-1 agreement (%)")
    ax.set_ylim(0, 105)
    ax.set_title("Pareto: accuracy vs speedup")
    ax.grid(True, ls="--", alpha=0.4)
    for x, y, t in zip(speedup, top1 * 100.0, data["threshold"]):
        ax.annotate(f"{t:.2f}", (x, y), textcoords="offset points", xytext=(5, -10), fontsize=7)

    out = csv_path.with_suffix(".png")
    fig.tight_layout()
    fig.savefig(out, dpi=150)
    print(f"Saved {out}")


if __name__ == "__main__":
    main()