ergmmDegroot: Calculates latent space diffusion using the degroot function

In jcfisher/latentnetDiffusion: Runs latent space diffusion model (ALPHA TESTING VERSION)

Description

Calculates latent space diffusion using the degroot function

Usage

ergmmDegroot(ergmm, Y, draws = 5, simulate = FALSE, iterations = 5, ...)

Arguments

ergmm:	the object to get the posterior from
draws:	the number of the draws from the posterior to take
simulate:	logical indictor of whether to use the predicted probabilities or the posterior predictive distribution
iterations:	the number of iterations to run the simulation
...:	additional parameters passed to degroot (if simulate is FALSE) or degrootList (if simulate is TRUE)

Details

This function calculates latent space diffusion, as described in Fisher (2015), for a given number of draws from the posterior distribution of a latent space model. If simulate is FALSE (the default), the function uses predicted probabilities of a tie to simulate diffusion. Otherwise, if simulated is TRUE, the function simulates (iterations) networks drawn from the first (draws) draws from the posterior predictive distribution, and then runs the diffusion process over those simulated networks.

More specifically, when simulate is FALSE, the function gets the predicted probability from the first (draws) draws from the posterior distribution. Holding each of those draws constant, it calculates the degroot function (iterations) times on each draw, returning a list of the output of the degroot function on each draw

When simulate is TRUE, the function simulates (iterations) networks from each of the first (draws) draws from the posterior, creating the posterior predictive distribution. Then, for each of the draws, the function runs degrootList, meaning that it calculates the weighted averaging over the first network, followed by the second network, and so on until (iterations) is reached. A list is returned.

Value

A (draws) length list of the simulation output for each draw from the posterior.

jcfisher/latentnetDiffusion documentation built on May 20, 2019, 5:26 p.m.