centrality_diffusion: Diffusion Centrality

View source: R/centrality.R

centrality_diffusionR Documentation

Diffusion Centrality

Description

Diffusion centrality has two forms, chosen by diffusion_method. The "kandhway_kuri" form (Kandhway and Kuri 2014) adds the degrees of a node's neighbors to its own degree, both scaled by \lambda: DC(v) = \lambda k_v + \lambda \sum_{u \in N(v)} k_u. The "power_series" form sums the rows of the first n powers of the weight matrix:

DC(v) = \sum_{w} \left( W + W^2 + \cdots + W^n \right)_{vw},

and min-max scales these sums to the range from 0 to 1.

Usage

centrality_diffusion(x, mode = "all", lambda = 1, ...)

Arguments

x

Network input accepted by centrality.

mode

For directed networks: "all" (default), "out" or "in".

lambda

Scale factor \lambda of the "kandhway_kuri" form. Default 1.

...

Further arguments to centrality. The measure uses diffusion_method ("kandhway_kuri" or "power_series", default NULL, which picks by input type) and loops (default TRUE, FALSE for tna input).

Details

The default is "kandhway_kuri", and "power_series" for tna input. The "kandhway_kuri" form uses binary degrees, so edge weights are ignored, and mode sets both the degrees and the neighbor set. On a directed network mode = "all" uses total degrees and the undirected neighbor set. lambda multiplies every score. The "power_series" form uses the edge weights, or ones with weighted = FALSE, and ignores mode and lambda. With loops = FALSE the diagonal of W is set to zero. The "power_series" values match tna::centralities(measures = "Diffusion") of tna 1.5.8 or later, which scales in the same way, and are NaN when the sums do not vary across the nodes.

Value

A named numeric vector with one score per node, in input node order.

See Also

centrality_expected, centrality_diffusion_centrality, centrality.

Examples

centrality_diffusion(regulation_net)

cograph documentation built on Nov. 19, 2026, 5:07 p.m.