fix_eigenvalues: Modify network to ensure stationarity.

In EDISON: Network Reconstruction and Changepoint Detection

Description

This function ensures that the eigenvalues of the network structure matrix are smaller or equal to 1, thereby ensuring stationarity of the regression. This is done by removing edges at random until the condition is satisfied.

Usage

fix_eigenvalues(network, q, gauss_weights)

Arguments

network	Original network structure, a matrix of size NumNodes by NumNodes.
q	Number of nodes.
gauss_weights	If TRUE, use Gaussian regression weight, if FALSE conserve original weights.

Value

Returns a network with fewer eigenvalues than the original network, but satisfying the stationarity condition.

Author(s)

Frank Dondelinger

See Also

generateNetwork

EDISON documentation built on May 2, 2019, 2:39 a.m.