pyphi.visualize.render.embedding

Global-embedding layout for the hypergraph view.

Positions each MICE (endpoint) by a deterministic embedding of its composition, so spatial proximity reflects compositional similarity rather than purview size. Two methods: PCA of a composition feature vector, and classical (Torgerson) MDS of a purview-overlap distance. Both are deterministic and numpy-only.

Functions

embedding_positions(projection, geometry)	Endpoint and mechanism positions from a global composition embedding.
mds_embed(distance[, n_components])	Classical (Torgerson) MDS of a dissimilarity matrix, sign-fixed; axes with non-positive eigenvalues fall back to an even spread by id.
pca_embed(vectors[, n_components])	First n_components principal components of vectors (rows = items), sign-fixed; zero-variance components fall back to an even spread by id.