File size: 10,073 Bytes
d1cfb66 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 | """Multi-step motif completion horizon experiment."""
from __future__ import annotations
import json
from collections.abc import Sequence
from dataclasses import asdict, dataclass
from pathlib import Path
from control_generator import sample_degree_matched_controls
from motif_completion_experiment import consolidate_motifs
from recurrence_topology_experiment import canonical_signature, training_data
HORIZONS = ("h1", "h2", "full")
@dataclass(frozen=True)
class HorizonCase:
name: str
motif_family: str
expected_class: str
partial_sequence: tuple[str, ...]
candidate_sequence: tuple[str, ...] | None
prediction_available: bool
predicted_full_suffix: tuple[int, ...]
observed_full_suffix: tuple[int, ...] | None
accepted_h1: bool
accepted_h2: bool
accepted_full: bool
@dataclass(frozen=True)
class ExperimentResult:
motifs: dict[str, str]
cases: tuple[HorizonCase, ...]
metrics: dict[str, float]
criteria: dict[str, bool]
outcome: str
def predicted_suffix(partial_sequence: Sequence[str], motif_family: str) -> tuple[int, ...] | None:
motif = consolidate_motifs()[motif_family]
partial_trace = canonical_signature(partial_sequence).trace
if len(partial_trace) >= len(motif.trace):
return None
if motif.trace[: len(partial_trace)] != partial_trace:
return None
return motif.trace[len(partial_trace) :]
def observed_suffix(
partial_sequence: Sequence[str], candidate_sequence: Sequence[str]
) -> tuple[int, ...] | None:
partial_trace = canonical_signature(partial_sequence).trace
candidate_trace = canonical_signature(candidate_sequence).trace
if len(candidate_trace) <= len(partial_trace):
return None
if candidate_trace[: len(partial_trace)] != partial_trace:
return None
return candidate_trace[len(partial_trace) :]
def accepted_at_horizon(
predicted: tuple[int, ...], observed: tuple[int, ...] | None, horizon: str
) -> bool:
if observed is None:
return False
if horizon == "h1":
length = 1
elif horizon == "h2":
length = min(2, len(predicted))
elif horizon == "full":
length = len(predicted)
else:
raise ValueError(horizon)
return len(observed) >= length and observed[:length] == predicted[:length]
def evaluate_case(
name: str,
motif_family: str,
partial_sequence: Sequence[str],
candidate_sequence: Sequence[str] | None,
expected_class: str,
) -> HorizonCase:
predicted = predicted_suffix(partial_sequence, motif_family)
observed = (
None
if candidate_sequence is None
else observed_suffix(partial_sequence, candidate_sequence)
)
predicted_tuple = () if predicted is None else predicted
return HorizonCase(
name=name,
motif_family=motif_family,
expected_class=expected_class,
partial_sequence=tuple(partial_sequence),
candidate_sequence=None if candidate_sequence is None else tuple(candidate_sequence),
prediction_available=predicted is not None,
predicted_full_suffix=predicted_tuple,
observed_full_suffix=observed,
accepted_h1=predicted is not None and accepted_at_horizon(predicted, observed, "h1"),
accepted_h2=predicted is not None and accepted_at_horizon(predicted, observed, "h2"),
accepted_full=predicted is not None and accepted_at_horizon(predicted, observed, "full"),
)
def predefined_cases() -> list[tuple[str, str, tuple[str, ...], tuple[str, ...] | None, str]]:
return [
(
"recurrence_positive_full",
"recurrence_loop",
("X", "Y", "X"),
("X", "Y", "X", "Z", "X"),
"positive",
),
(
"recurrence_prefix_divergent_h1",
"recurrence_loop",
("X", "Y", "X"),
("X", "Y", "X", "Z", "Y"),
"prefix_divergent_control",
),
(
"recurrence_path",
"recurrence_loop",
("P0", "P1", "P2"),
("P0", "P1", "P2", "P3", "P4"),
"path_control",
),
(
"branch_positive_full",
"branch_converge_loop",
("X", "Y", "Z", "X"),
("X", "Y", "Z", "X", "W", "Z", "X"),
"positive",
),
(
"branch_prefix_divergent_h1",
"branch_converge_loop",
("X", "Y", "Z", "X"),
("X", "Y", "Z", "X", "W", "Y", "X"),
"prefix_divergent_control",
),
(
"branch_prefix_divergent_h2",
"branch_converge_loop",
("X", "Y", "Z", "X"),
("X", "Y", "Z", "X", "W", "Z", "Y"),
"prefix_divergent_control",
),
(
"branch_path",
"branch_converge_loop",
("P0", "P1", "P2", "P3"),
("P0", "P1", "P2", "P3", "P4", "P5", "P6"),
"path_control",
),
(
"repeated_positive_full",
"repeated_substructure",
("X", "Y", "X", "Z", "X"),
("X", "Y", "X", "Z", "X", "Y", "X", "Z"),
"positive",
),
(
"repeated_prefix_divergent_h1",
"repeated_substructure",
("X", "Y", "X", "Z", "X"),
("X", "Y", "X", "Z", "X", "Y", "Z", "X"),
"prefix_divergent_control",
),
(
"repeated_path",
"repeated_substructure",
("P0", "P1", "P2", "P3", "P4"),
("P0", "P1", "P2", "P3", "P4", "P5", "P6", "P7"),
"path_control",
),
]
def degree_matched_cases() -> list[tuple[str, str, tuple[str, ...], tuple[str, ...] | None, str]]:
records: list[tuple[str, str, tuple[str, ...], tuple[str, ...] | None, str]] = []
for family, motif in consolidate_motifs().items():
controls = sample_degree_matched_controls(
motif, family=f"horizon_{family}", count=5, seed=404000 + len(records)
)
for control in controls:
split = max(2, len(control.sequence) // 2)
records.append(
(
f"{control.name}_horizon_probe",
family,
control.sequence[:split],
control.sequence,
"degree_matched_non_isomorphic_control",
)
)
return records
def rate(cases: Sequence[HorizonCase], attr: str) -> float:
if not cases:
return 0.0
return sum(bool(getattr(case, attr)) for case in cases) / len(cases)
def run_experiment() -> ExperimentResult:
motifs = consolidate_motifs()
raw_cases = predefined_cases() + degree_matched_cases()
cases = tuple(evaluate_case(*case) for case in raw_cases)
positives = [case for case in cases if case.expected_class == "positive"]
degree_controls = [
case for case in cases if case.expected_class == "degree_matched_non_isomorphic_control"
]
prefix_controls = [case for case in cases if case.expected_class == "prefix_divergent_control"]
path_controls = [case for case in cases if case.expected_class == "path_control"]
compression_ratio = sum(len(sequences) for sequences in training_data().values()) / len(motifs)
metrics = {
"positive_h1_acceptance_rate": round(rate(positives, "accepted_h1"), 6),
"positive_h2_acceptance_rate": round(rate(positives, "accepted_h2"), 6),
"positive_full_acceptance_rate": round(rate(positives, "accepted_full"), 6),
"degree_matched_h1_acceptance_rate": round(rate(degree_controls, "accepted_h1"), 6),
"degree_matched_h2_acceptance_rate": round(rate(degree_controls, "accepted_h2"), 6),
"degree_matched_full_acceptance_rate": round(rate(degree_controls, "accepted_full"), 6),
"prefix_divergent_h1_acceptance_rate": round(rate(prefix_controls, "accepted_h1"), 6),
"prefix_divergent_h2_acceptance_rate": round(rate(prefix_controls, "accepted_h2"), 6),
"prefix_divergent_full_acceptance_rate": round(rate(prefix_controls, "accepted_full"), 6),
"path_prediction_rate": round(
sum(case.prediction_available for case in path_controls) / len(path_controls), 6
),
"compression_ratio": round(compression_ratio, 6),
}
criteria = {
"positive_full_acceptance_rate_eq_1": metrics["positive_full_acceptance_rate"] == 1.0,
"positive_h1_h2_acceptance_rates_eq_1": metrics["positive_h1_acceptance_rate"] == 1.0
and metrics["positive_h2_acceptance_rate"] == 1.0,
"degree_matched_full_acceptance_rate_eq_0": metrics["degree_matched_full_acceptance_rate"]
== 0.0,
"prefix_divergent_full_acceptance_rate_eq_0": metrics[
"prefix_divergent_full_acceptance_rate"
]
== 0.0,
"prefix_divergent_h2_acceptance_rate_le_0_25": metrics[
"prefix_divergent_h2_acceptance_rate"
]
<= 0.25,
"path_prediction_rate_eq_0": metrics["path_prediction_rate"] == 0.0,
"compression_ratio_ge_3": compression_ratio >= 3.0,
}
if all(criteria.values()):
outcome = "supported"
elif criteria["positive_full_acceptance_rate_eq_1"]:
outcome = "partially_supported_control_failures"
else:
outcome = "falsified_positive_prediction_failed"
return ExperimentResult(
motifs={family: motif.key() for family, motif in motifs.items()},
cases=cases,
metrics=metrics,
criteria=criteria,
outcome=outcome,
)
def main() -> None:
result = run_experiment()
output_path = Path("results/multi_step_completion_experiment_results.json")
output_path.parent.mkdir(parents=True, exist_ok=True)
output_path.write_text(json.dumps(asdict(result), indent=2, sort_keys=True) + "\n")
print(output_path)
print(json.dumps({"outcome": result.outcome, "criteria": result.criteria}, indent=2))
if __name__ == "__main__":
main()
|