aries/parse-search.py at main · pgimenes/aries · GitHub

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
import os
import re
import sys, pdb, traceback
import matplotlib.pyplot as plt
import numpy as np
import ast

def excepthook(exc_type, exc_value, exc_traceback):
    traceback.print_exception(exc_type, exc_value, exc_traceback)
    print("\nEntering debugger...")
    pdb.post_mortem(exc_traceback)

# Set the custom exception hook
sys.excepthook = excepthook

tasks = [
    "sorting32",
    "sorting64",
    "sorting128",
    "set_intersection32",
    "set_intersection64",
    "set_intersection128",
    "keyword_counting",
]

for task in tasks:

    trials = []
    for i in range(100):
        fname = f'search/{task}/trial_'+str(i)+'-spec.log'

        if not os.path.exists(fname):
            continue

        with open(fname, 'r') as f:
            content = "\n".join(f.readlines())

            if not os.path.exists(fname):
                continue

            got_branches = int(re.findall(r'got_branches: (\d+)', content)[0])
            got_generate_attempts = int(re.findall(r'got_generate_attempts: (\d+)', content)[0])
            got_aggregate_attempts = int(re.findall(r'got_aggregate_attempts: (\d+)', content)[0])
            got_refine_attempts = int(re.findall(r'got_refine_attempts: (\d+)', content)[0])
            got_post_aggregate_keepbest = ast.literal_eval(re.findall(r'got_post_aggregate_keepbest: (True|False)', content)[0])
            got_post_aggregate_refine = ast.literal_eval(re.findall(r'got_post_aggregate_refine: (True|False)', content)[0])

            config = (
                got_branches,
                got_post_aggregate_keepbest,
                got_post_aggregate_refine,
                got_generate_attempts,
                got_aggregate_attempts,
                got_refine_attempts,
            )

            try:
                query_count = int(re.findall(r'query count: (\d+)+', content)[0])
                score = float(re.findall(r'score: (\d+\.\d+)', content)[0])
                if score >= 1000:
                    continue
                trials.append((query_count, score, config))
            except:
                continue

    # calculate alpha to balance query count and score
    avg_score = np.mean([t[1] for t in trials])
    avg_query_count = np.mean([t[0] for t in trials])
    alpha = min([avg_query_count / (avg_score + avg_query_count), 0.99])
    print(f"task: {task}")
    print(f"alpha: {alpha}")

    for idx, val in enumerate(trials):
        query_count, score, config = val
        cost = alpha * score + (1 - alpha) * query_count
        trials[idx] = (query_count, score, cost, config)

    # create min filter
    best_trials = []
    argmins = []
    for idx in range(len(trials)):
        costs = [t[2] for t in trials[:idx+1]]
        argmin = np.argmin(costs)
        # eliminate the config from 'best_trials'
        best_trials.append(trials[argmin][:-1])
        argmins.append(argmin)
    best_trials = np.array(best_trials)

    # find last effective trial (first index at which the cost == cost[-1])
    last_trial = len(best_trials) - 1
    idx = len(best_trials) - 1
    while idx >= 0:
        if best_trials[idx][2] == best_trials[-1][2]:
            last_trial = idx
        idx -= 1
    i25 = int(0.25 * last_trial)
    i50 = int(0.50 * last_trial)

    # plot in different plots vertically
    fig, axs = plt.subplots(3)
    fig.suptitle('Search Performance')
    axs[0].plot(best_trials[:, 0], label='query count')
    axs[0].legend()
    axs[1].plot(best_trials[:, 1], label='score')
    axs[1].legend()
    axs[2].plot(best_trials[:, 2], label='cost')
    axs[2].legend()

    # create line in each plot at i25, i50, i75
    for ax in axs:
        ax.axvline(x=i25, color='r', linestyle='--')
        ax.axvline(x=i50, color='r', linestyle='--')
        ax.axvline(x=last_trial, color='g', linestyle='--')

    plt.savefig(f'trial-plot-{task}.png')

    # create pareto plot
    fig, ax = plt.subplots()
    fig.suptitle('Search Performance')
    ax.scatter([t[0] for t in trials], [t[1] for t in trials], c=[t[2] for t in trials])
    ax.set_xlabel('query count')
    ax.set_ylabel('score')
    plt.savefig(f'pareto-plot-{task}.png')

    print(f"i25/i50/last_trial/num_trials: {i25}, {i50}, {last_trial}, {len(trials)}")

    for item in [(i25, "25"), (i50, "50"), (last_trial, "last")]:
        idx, name = item
        config = trials[idx][3]

        args = ""
        got_branches, got_post_aggregate_keepbest, got_post_aggregate_refine, got_generate_attempts, got_aggregate_attempts, got_refine_attempts = config
        args += f"""
                                --got_branches {got_branches} \\
                                --got_generate_attempts {got_generate_attempts} \\
                                --got_aggregate_attempts {got_aggregate_attempts} \\
                                --got_refine_attempts {got_refine_attempts} \\
"""
        if got_post_aggregate_keepbest:
            args += "                                --got_post_aggregate_keepbest \\\n"
        if got_post_aggregate_refine:
            args += "                                --got_post_aggregate_refine \\\n"

        print(f"got{name}: {best_trials[idx]}")
        print(args)