sciencehub-team/constellarationPublic

def hypervolume(gradient_scale_length, aspect_ratio, reference=(1.0, 20.0)) -> float:
    """Area dominated by the points in the plane of (−L∇B, A), up to the reference point.

    The MHD-stable score of constellaration 0.2.6 for feasible points; equal
    to pymoo's Hypervolume to 1e-12 on all 58 leaderboard submissions.
    """
    x = -np.asarray(gradient_scale_length, dtype=float)
    y = np.asarray(aspect_ratio, dtype=float)
    keep = (x < reference[0]) & (y < reference[1])
    x, y = x[keep], y[keep]
    order = np.argsort(x, kind="stable")
    x, y = x[order], y[order]
    edges = np.append(x, reference[0])
    lowest = np.minimum.accumulate(y) if len(y) else y
    return float(np.sum(np.diff(edges) * (reference[1] - lowest)))

def load_entries(path: str = ENTRIES) -> pd.DataFrame:
    """The leaderboard as recorded on 25 October 2026: one row per single-objective
    submission and one per point of each MHD-stable submission."""
    return pd.read_parquet(path)

def mhd_stable_scores(entries: pd.DataFrame, tolerance: float = TOLERANCE, top: int = 10) -> pd.DataFrame:
    """The `top` MHD-stable submissions by hypervolume when only points within `tolerance` count.

    All submissions are ranked; a resubmission with the same user and
    recorded score counts once. "recorded rank" is the order of the
    recorded scores.
    """
    points = entries[entries["problem"] == "mhd_stable"]
    rows = []
    for (user, submitted), group in points.groupby(["user", "submitted"]):
        kept = group[group["feasibility"] <= tolerance]
        rows.append({
            "user": user,
            "submitted": submitted[:10],
            "points": len(group),
            "recorded score": group["entry_score"].iloc[0],
            "score at this tolerance": hypervolume(kept["gradient_scale_length"], kept["aspect_ratio"]),
            "points in the last tenth of the band": float(np.mean((group["feasibility"] > 0.9 * TOLERANCE) & (group["feasibility"] <= TOLERANCE))),
        })
    table = pd.DataFrame(rows).sort_values("recorded score", ascending=False).drop_duplicates(["user", "recorded score"])
    table.insert(0, "recorded rank", range(1, len(table) + 1))
    table.insert(0, "rank at this tolerance", table["score at this tolerance"].rank(ascending=False, method="min").astype(int))
    return table.sort_values(["rank at this tolerance", "recorded rank"]).head(top).reset_index(drop=True)