optiwindnet.mesh

Module Contents

optiwindnet.mesh.make_planar_embedding(L: networkx.Graph, offset_scale: float = 0.0001, max_tri_AR: float = _MAX_TRIANGLE_ASPECT_RATIO) → tuple[networkx.PlanarEmbedding, networkx.Graph]

Triangulate a location and produce graphs P and A for it.

P is the planar embedding mesh and A is the available-edges graph. TODO: change the name of this function.

Parameters:

• L – locations graph

• offset_scale – Fraction of the diagonal of the site’s bbox to use as spacing between border and nodes in concavities (only where nodes are the border).

• max_tri_AR – maximum aspect ratio allowed for triangles on the hull of A as the algorithm removes flat triangles (higher values keep more).

Returns:

P - the planar embedding graph - and A - the available-edges graph.

optiwindnet.mesh.delaunay(L: networkx.Graph, bind2root: bool = False) → networkx.Graph

DEPRECATED. Create the extended-Delaunay-based available-edges graph A.

Parameters:

• L – location

• bind2root – assign edge attribute ‘root’ (used by legacy heuristics)

Returns:

A - available-edges graph

optiwindnet.mesh.planar_flipped_by_routeset(edges_G: set[tuple[int, int]], *, planar: networkx.PlanarEmbedding, VertexC: optiwindnet.geometric.CoordPairs, ST: int, diagonals: bidict.bidict | None = None) → networkx.PlanarEmbedding

Ajust planar to include the edges actually used by a routeset.

Copies planar and flips the edges to their diagonal if the latter is an edge in edges_G. Ideally, the returned PlanarEmbedding includes all edges_G (an expected discrepancy are gates).

edges_G is the set of routeset edges in prime-id form (i.e. clones already mapped through fnT), each as a normalized (u, v) pair with u < v. Gate edges have u < 0. ST is T + B (the boundary above which constraint vertices live).

If diagonals is provided, some diagonal gates may become planar’s edges if they are not crossing any edge in edges_G. Otherwise gates are ignored.

Important: the routeset must be free of edge×edge crossings.