log_like_forces function

create an initial configuration for the latent positions.

This performs an iterative relaxation type algorithm to approximately find the positions of the nodes in the latent space that maximises the log-likelihood.

log_like_forces(net, D, X, B, m ,steps)

Arguments

• net: network object on which to perform layout.
• D: dimension of the latent space.
• X: the initial guess for X
• B: the intercept term.
• m: usually N will suffice.
• steps: maximum number of iteration steps.

Details

Usually only used internally in vblpcmstart()

Returns

Matrix of latent positions X

Author(s)

Michael Salter-Townshend

See Also

igraph::layout.fruchterman.reingold

Examples

data(sampson)
N=network.size(samplike)
X=matrix(runif(N*2,-2,2),ncol=2)
XX=vblpcmcovs(N,"plain",as.sociomatrix(samplike))
out<-log_like_forces(samplike, 2, X, 0, m=N, 1e3)
plot(samplike,coord=out$X)