1import json
 2from time import perf_counter
 3from canonical_orbit import (all_permutations, product_group, exhaustive,
 4                             canonical, canonical_dfs, orbit)
 5
 6
 7def run():
 8    n, k = 8, 4
 9    blocks = [tuple(range(4)), tuple(range(4, 8))]
10    group = product_group(blocks)
11    masks = exhaustive(n, k)
12    # Exact symmetry: every channel in block 0 has importance 3, and every
13    # channel in block 1 has importance 1. Thus orbit-related masks score alike.
14    score = lambda s: sum(3 if i < 4 else 1 for i in s)
15    t0 = perf_counter()
16    exhaustive_best = max((score(s), s) for s in masks)
17    exhaustive_sec = perf_counter() - t0
18    t0 = perf_counter()
19    reps = canonical_dfs(n, k, group)
20    canonical_sec = perf_counter() - t0
21    canonical_best = max((score(s), s) for s in reps)
22    reps_set = set(reps)
23    covered = set().union(*(orbit(r, group) for r in reps_set))
24    orbit_invariance = all(score(s) == score(canonical(s, group)) for s in masks)
25    return {
26        "n": n, "k": k, "group_size": len(group),
27        "all_masks": len(masks), "canonical_representatives": len(reps),
28        "evaluation_reduction": len(masks) / len(reps),
29        "exhaustive_best_score": exhaustive_best[0],
30        "canonical_best_score": canonical_best[0],
31        "same_best_score": exhaustive_best[0] == canonical_best[0],
32        "orbit_coverage": covered == set(masks),
33        "score_orbit_invariant": orbit_invariance,
34        "exhaustive_scan_sec": exhaustive_sec,
35        "canonical_dfs_sec": canonical_sec,
36    }
37
38if __name__ == '__main__':
39    print(json.dumps(run(), sort_keys=True))