randomize_asymmetric: Maximise pairwise asymmetry

In franzikoch/Feedbackloops: Feedback loop analysis for competition networks

Maximise pairwise asymmetry

Description

During this randomisation procedure, all interaction strengths are reordered to make the 2-link loops in the randomised system as asymmetric as possible. Diagonal values and network topology are preserved.

Usage

randomize_asymmetric(it, ij_col, ji_col)

Arguments

it	interaction table (created by interaction_strengths())
ij_col	column of a_ij values to randomise (choose scaled or unscaled)
ji_col	column of a_ji values to randomise (scaled or unscaled)

Details

To do this, all links are ordered by size. Then, the very strongest link is paired with the weakest one, the second strongest with the second weakest etc. The location of pairwise interactions in the network is chosen randomly. Also, links are randomised between the ij and ji column, so that the strong links can appear both above and below the diagonal.

The function returns a new interaction table that contains two new columns $a_ij_asym and $a_ji_asym that contains the same values as the original columns but in a different order.

To get a Jacobian matrix with maximised pairwise asymmetry, use assemble_jacobian() and specify the new columns.

Value

Interaction table with two additional columns $a_ij_asym and $a_ji_asym, containing randomised interaction strengths

franzikoch/Feedbackloops documentation built on July 1, 2023, 12:42 p.m.