File size: 7,295 Bytes
7b0b49b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
202
203
204
205
206
from __future__ import annotations

import importlib.util
from pathlib import Path

import numpy as np
import pytest


REPO = Path(__file__).resolve().parents[1]


def _generator(seed: int = 123):
    spec = importlib.util.spec_from_file_location(
        "profilefix_generator", REPO / "generator.py"
    )
    module = importlib.util.module_from_spec(spec)
    assert spec.loader is not None
    spec.loader.exec_module(module)
    module._CHUNK = 128
    return module.Generator(str(REPO), seed=seed)


def _module():
    spec = importlib.util.spec_from_file_location(
        "profilefix_module", REPO / "generator.py"
    )
    module = importlib.util.module_from_spec(spec)
    assert spec.loader is not None
    spec.loader.exec_module(module)
    return module


def test_exact_count_allowed_lengths_and_finite():
    rows = list(_generator().generate(131))
    assert len(rows) == 131
    assert {row.shape for row in rows} == {(4096,)}
    assert all(row.dtype == np.float64 for row in rows)
    assert all(np.isfinite(row).all() and row.std() > 1e-9 for row in rows)


def test_deterministic_across_instances():
    left = list(_generator().generate(130))
    right = list(_generator().generate(130))
    assert all(np.array_equal(a, b) for a, b in zip(left, right, strict=True))


def test_requested_size_does_not_change_prefix():
    short = list(_generator().generate(17))
    long = list(_generator().generate(130))
    assert all(np.array_equal(a, b) for a, b in zip(short, long[:17], strict=True))


def test_nonpositive_request_is_empty():
    assert list(_generator().generate(0)) == []


def test_genuine_short_outputs_are_disabled():
    rows = list(_generator(seed=211).generate(512))
    lengths = np.asarray([row.size for row in rows])
    assert set(lengths) == {4096}


def test_family_weights_are_exact_conservative_uid131_scaling():
    module = _module()
    uid131 = {
        "trend_seasonal_ar": 0.12,
        "regime_shift": 0.12,
        "multiplicative": 0.08,
        "ar2": 0.15,
        "integrated": 0.12,
        "threshold_ar": 0.08,
        "chaotic": 0.04,
        "spectral_gp": 0.07,
        "long_memory": 0.06,
        "ou_stochastic_vol": 0.10,
        "physical_sensors": 0.02,
        "seasonal_counts": 0.02,
        "intermittent": 0.01,
        "pulse_outlier": 0.01,
    }
    assert sum(module._DEFAULT_WEIGHTS.values()) == pytest.approx(1.0)
    assert module._DEFAULT_WEIGHTS["prefix_cumulative_counter"] == 0.025
    for family, weight in uid131.items():
        assert module._DEFAULT_WEIGHTS[family] == pytest.approx(0.975 * weight)


def test_period5_added_without_changing_period96_probability():
    module = _module()
    assert module._SEASONAL_PROBS.sum() == pytest.approx(1.0)
    period_to_probability = dict(
        zip(
            module._SEASONAL_PERIODS.astype(int),
            module._SEASONAL_PROBS,
            strict=True,
        )
    )
    assert period_to_probability[5] > 0.0
    # Raw mass was reallocated only from periods 4 and 7. UID131's raw total
    # was 1.07, so period 96 retains exactly its prior conditional probability.
    assert period_to_probability[96] == pytest.approx(0.07 / 1.07)


def test_first_value_padding_keeps_suffix_and_reads_no_later_value():
    module = _module()
    raw = np.arange(3 * 32, dtype=np.float64).reshape(3, 32)
    lengths = np.asarray([8, 12, 20])
    padded = module._left_pad_first_value(raw, lengths)
    for row, active in enumerate(lengths):
        start = raw.shape[1] - int(active)
        assert np.array_equal(padded[row, start:], raw[row, start:])
        assert np.all(padded[row, :start] == raw[row, start])
    assert np.array_equal(raw, np.arange(3 * 32).reshape(3, 32))


def test_cumulative_primitive_is_prefix_causal():
    module = _module()
    steps = np.random.default_rng(223).normal(size=(3, 512))
    base = np.asarray([1.0e4, 2.0e8, 3.0e12])
    start = np.asarray([7, 31, 63])
    full = module._causal_cumulative_from_steps(steps, base, start)
    prefix = module._causal_cumulative_from_steps(
        steps[:, :257], base, start
    )
    assert np.array_equal(full[:, :257], prefix)


def test_prefix_cumulative_family_is_deterministic_large_and_padded():
    module = _module()
    left = module._prefix_cumulative_counter(
        np.random.default_rng(227), 64, 4096
    )
    right = module._prefix_cumulative_counter(
        np.random.default_rng(227), 64, 4096
    )
    assert np.array_equal(left, right)
    assert np.isfinite(left).all()
    assert (left >= 0.0).all()
    assert float(left.max()) > 1.0e8
    changed = np.diff(left, axis=1) != 0.0
    first_change = changed.argmax(axis=1) + 1
    assert (first_change >= 4096 - 1024).all()
    assert (changed.sum(axis=1) > 8).all()
    assert "prefix_cumulative_counter" in module._FAMILIES
    assert module._DEFAULT_WEIGHTS["prefix_cumulative_counter"] == 0.025


def test_ou_stochastic_vol_is_deterministic_finite_and_active():
    module = _module()
    left = module._ou_stochastic_vol(np.random.default_rng(77), 16, 512)
    right = module._ou_stochastic_vol(np.random.default_rng(77), 16, 512)
    assert np.array_equal(left, right)
    assert np.isfinite(left).all()
    assert (left.std(axis=1) > 1e-9).all()
    assert "ou_stochastic_vol" in module._FAMILIES
    assert module._DEFAULT_WEIGHTS["ou_stochastic_vol"] > 0


def test_physical_sensors_are_deterministic_finite_and_active():
    module = _module()
    left = module._physical_sensors(np.random.default_rng(91), 32, 512)
    right = module._physical_sensors(np.random.default_rng(91), 32, 512)
    assert np.array_equal(left, right)
    assert np.isfinite(left).all()
    assert (left.std(axis=1) > 1e-9).all()
    assert "physical_sensors" in module._FAMILIES
    assert module._DEFAULT_WEIGHTS["physical_sensors"] > 0


def test_seasonal_counts_are_deterministic_integer_and_active():
    module = _module()
    left = module._seasonal_counts(np.random.default_rng(109), 32, 512)
    right = module._seasonal_counts(np.random.default_rng(109), 32, 512)
    assert np.array_equal(left, right)
    assert np.isfinite(left).all()
    assert (left >= 0).all()
    assert np.array_equal(left, np.round(left))
    assert (left.std(axis=1) > 1e-9).all()
    assert "seasonal_counts" in module._FAMILIES
    assert module._DEFAULT_WEIGHTS["seasonal_counts"] > 0


def test_event_family_has_real_flat_runs_and_recovery():
    module = _module()
    left = module._pulse_outlier(np.random.default_rng(131), 32, 512)
    right = module._pulse_outlier(np.random.default_rng(131), 32, 512)
    assert np.array_equal(left, right)
    assert np.isfinite(left).all()
    assert (left.std(axis=1) > 1e-9).all()
    assert np.any(np.diff(left, axis=1) == 0.0)


def test_measurement_artifacts_are_deterministic_finite_and_shape_safe():
    module = _module()
    raw = module._trend_seasonal_ar(np.random.default_rng(149), 64, 512)
    left = module._measurement_artifacts(
        np.random.default_rng(151), raw, preserve_nonnegative=False
    )
    right = module._measurement_artifacts(
        np.random.default_rng(151), raw, preserve_nonnegative=False
    )
    assert np.array_equal(left, right)
    assert left.shape == raw.shape
    assert np.isfinite(left).all()
    assert (left.std(axis=1) > 1e-9).all()