httk.analyse.generic.lower_hull¶
Generic lower convex-hull analysis for finite point-and-value collections.
Classes¶
The lower convex hull of scalar values over arbitrary finite coordinates. |
Module Contents¶
- class httk.analyse.generic.lower_hull.LowerConvexHull(points, values, *, tolerance=1e-08)¶
The lower convex hull of scalar values over arbitrary finite coordinates.
A point belongs to the hull when its value is no greater than the cheapest leave-one-out convex mixture at the same coordinates, within
tolerance. The input order is retained throughout, including for tied duplicate points.Coordinates and values are converted through
numpy.float64and exposed as ordinaryfloatvalues. Every coordinate equality is retained in the mixture LP, together with the affinesum(weights) == 1equality.- Parameters:
points (collections.abc.Sequence[collections.abc.Sequence[float]]) – Coordinate rows for the input points.
values (collections.abc.Sequence[float]) – Scalar values corresponding to
points.tolerance (float) – Maximum value excess treated as on the lower hull.
- Raises:
ValueError – If the points, values, or tolerance are invalid.
- property points: tuple[tuple[float, Ellipsis], Ellipsis]¶
Return input coordinates in their original order as float tuples.
- property hull_indices: tuple[int, Ellipsis]¶
Return indices of points on the lower hull in input order.
- property value_above_hull: tuple[float, Ellipsis]¶
Return non-negative leave-one-out value excesses for every input point.
- property supported_segments: tuple[tuple[int, int], Ellipsis]¶
Return midpoint-supported pairs of distinct lower-hull points.
- decomposition(index)¶
Return lower-hull mixture
(index, weight)pairs, orNoneon hull.
- is_on_hull(index)¶
Return whether
indexbelongs tohull_indices.