dapp.simulate: Simulate from Dynamic Admixture of Poisson Process

In neuromplex: Neural Multiplexing Analysis

Description

Simulate spike trains from DAPP model to binned spiking data

Usage

 
dapp.simulate(horizon = 1000, bin.width = 25, lengthScale,
     lsPrior = rep(1/length(lengthScale),length(lengthScale)),
     hyper = list(prec = c(1,1), sig0 = 1.87, w=c(1,1)), nsamp = 1e3)

Arguments

horizon	time horizon of the response period (in ms)
bin.width	width of the time bins (in ms) to be used to aggregate spike counts
lengthScale	an array giving the length scale parameter values to be used for Gaussian process prior. Defaults to sort(0.16 * resp.horiz / c(4, 3, 2, 1, 0.5, 0.1)) where resp.horiz is the time horizon of the response period.
lsPrior	an array of the same length as lengthScale giving the prior probabilities of the length scale values.
hyper	a list of hyper parameters with the following iterms. 'prec': a 2-vector giving the shape and rate parameters of the gamma distribution on the Dirichlet precision parameter. 'sig0': a scalaer giving the scale of the (centered) logistic distribution used in transforming the Gaussian random curves into curves restricted between 0 and 1.
nsamp	number of priors draws to be made

Details

Primarily intended to be used internally by the summary.dapp and plot.dapp functions. Could also be use to draw directly from the model.

Value

Returns a list of class "dapp" containting the following items.

lsProb	draws of length scale
alpha.pred	prior predictive draws of alpha
prec	draws of precision

Examples

prior <- dapp.simulate(1000, 25)

neuromplex documentation built on April 22, 2021, 5:11 p.m.