File size: 4,670 Bytes
9fa2182
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import multiprocessing as mp
import os
import tempfile
import time
from pathlib import Path

import numpy as np
import pandas as pd

from .data import generate_data

EMPTY_DF = lambda: pd.DataFrame(
    {
        "Equation": [],
        "Loss": [],
        "Complexity": [],
    }
)


def process(
    file_input,
    force_run,
    test_equation,
    num_points,
    noise_level,
    data_seed,
    niterations,
    maxsize,
    binary_operators,
    unary_operators,
    plot_update_delay,
    parsimony,
    populations,
    population_size,
    ncycles_per_iteration,
    elementwise_loss,
    adaptive_parsimony_scaling,
    optimizer_algorithm,
    optimizer_iterations,
    batching,
    batch_size,
):
    """Load data, then spawn a process to run the greet function."""
    if file_input is not None:
        # Look at some statistics of the file:
        df = pd.read_csv(file_input)
        if len(df) == 0:
            return (
                EMPTY_DF(),
                "The file is empty!",
            )
        if len(df.columns) == 1:
            return (
                EMPTY_DF(),
                "The file has only one column!",
            )
        if len(df) > 10_000 and not force_run:
            return (
                EMPTY_DF(),
                "You have uploaded a file with more than 10,000 rows. "
                "This will take very long to run. "
                "Please upload a subsample of the data, "
                "or check the box 'Ignore Warnings'.",
            )

        col_to_fit = df.columns[-1]
        y = np.array(df[col_to_fit])
        X = df.drop([col_to_fit], axis=1)
    else:
        X, y = generate_data(test_equation, num_points, noise_level, data_seed)

    with tempfile.TemporaryDirectory() as tmpdirname:
        base = Path(tmpdirname)
        equation_file = base / "hall_of_fame.csv"
        equation_file_bkup = base / "hall_of_fame.csv.bkup"
        process = mp.Process(
            target=pysr_fit,
            kwargs=dict(
                X=X,
                y=y,
                niterations=niterations,
                maxsize=maxsize,
                binary_operators=binary_operators,
                unary_operators=unary_operators,
                equation_file=equation_file,
                parsimony=parsimony,
                populations=populations,
                population_size=population_size,
                ncycles_per_iteration=ncycles_per_iteration,
                elementwise_loss=elementwise_loss,
                adaptive_parsimony_scaling=adaptive_parsimony_scaling,
                optimizer_algorithm=optimizer_algorithm,
                optimizer_iterations=optimizer_iterations,
                batching=batching,
                batch_size=batch_size,
            ),
        )
        process.start()
        last_yield_time = None
        while process.is_alive():
            if equation_file_bkup.exists():
                try:
                    # First, copy the file to a the copy file
                    equation_file_copy = base / "hall_of_fame_copy.csv"
                    os.system(f"cp {equation_file_bkup} {equation_file_copy}")
                    equations = pd.read_csv(equation_file_copy)
                    # Ensure it is pareto dominated, with more complex expressions
                    # having higher loss. Otherwise remove those rows.
                    # TODO: Not sure why this occurs; could be the result of a late copy?
                    equations.sort_values("Complexity", ascending=True, inplace=True)
                    equations.reset_index(inplace=True)
                    bad_idx = []
                    min_loss = None
                    for i in equations.index:
                        if min_loss is None or equations.loc[i, "Loss"] < min_loss:
                            min_loss = float(equations.loc[i, "Loss"])
                        else:
                            bad_idx.append(i)
                    equations.drop(index=bad_idx, inplace=True)

                    while (
                        last_yield_time is not None
                        and time.time() - last_yield_time < plot_update_delay
                    ):
                        time.sleep(0.1)

                    yield equations[["Complexity", "Loss", "Equation"]]

                    last_yield_time = time.time()
                except pd.errors.EmptyDataError:
                    pass

        process.join()


def pysr_fit(
    *,
    X,
    y,
    **pysr_kwargs,
):
    import pysr

    model = pysr.PySRRegressor(
        progress=False,
        timeout_in_seconds=1000,
        **pysr_kwargs,
    )
    model.fit(X, y)