DstarM (version 0.4.0)

Voss.density: Calculate model density for a given set of parameters

Description

Calculate model density for a given set of parameters

Usage

Voss.density(t, pars, boundary, DstarM = TRUE, prec = 3)
LBA.density(t, pars, boundary, DstarM = TRUE, ...)
Wiener.density(t, pars, boundary, DstarM)

Value

A numeric vector of length length(t) containing a density.

Arguments

t

Time grid for density to be calculated on.

pars

Parameter vector where (if DstarM == TRUE) the first index contains the boundary parameter, the second contains the drift speed, the third contains the relative starting point, the fourth contains a proportion of the maximum size of the variance on the relative starting point, the fifth contains the standard deviation of the drift speed. if DstarM == FALSE then third index of pars contains the Ter, the fifth the drift speed, the the sixth contains a proportion of the maximum size of the variance on the relative starting point, the fifth contains the standard deviation of the drift speed, and the seventh contains a proportion of the maximum variance of the Ter.

boundary

For which response option will the density be calculated? Either 'upper' or 'lower'.

DstarM

Logical, see pars.

prec

Precision with which the density is calculated. Corresponds roughly to the number of decimals accurately calculated.

...

Other arguments, see dLBA

Details

These functions are examples of what fun.density should look like. Voss.density is an adaptation of ddiffusion, LBA.density is an adaptation of dLBA, and wiener.density is an adaptation of dwiener. To improve speed one can remove error handling. Normally error handling is useful, however because differential evolution can result in an incredible number of function evaluations (more than 10.000) it is recommended to omit error handling in custom density functions. estDstarM will apply some internal error checks (see testFun) on the density functions before starting differential evolution. A version of ddifusion without error handling can be found in the source code (commented out to pass R check). Note that for in Voss.density if DstarM == FALSE nondecision parameters are implemented manually and might differ from from how they are implemented in other packages. The parameter t0 specifies the mean of a uniform distribution and st0 specifies the relative size of this uniform distribution. To obtain the lower and upper range of the uniform distribution calculate a = t0 - t0*st0, and b = t0 + t0*st0.

Examples

t = seq(0, .75, .01)
V.pars = c(1, 2, .5, .5, .5)
L.pars = c(1, .5, 2, 1, 1, 1)
W.pars = V.pars[1:3]
V1 = Voss.density(t = t, pars = V.pars, boundary = 'upper', DstarM = TRUE)
V2 = Voss.density(t = t, pars = V.pars, boundary = 'lower', DstarM = TRUE)
L1 = LBA.density(t = t, pars = L.pars, boundary = 'upper', DstarM = TRUE)
L2 = LBA.density(t = t, pars = L.pars, boundary = 'lower', DstarM = TRUE)
W1 = Wiener.density(t = t, pars = W.pars, boundary = 'upper', DstarM = TRUE)
W2 = Wiener.density(t = t, pars = W.pars, boundary = 'lower', DstarM = TRUE)
densities = cbind(V1, V2, L1, L2, W1, W2)
matplot(t, densities, type = 'b', ylab = 'Density', lty = 1, las = 1, bty = 'n',
        col = rep(1:3, each = 2), pch = c(0, 15, 1, 16, 2, 17), cex = .8,
        main = 'Model densities')
legend('topright', legend = c('Voss', 'LBA', 'RWiener'), lty = 1,
       pch = 15:17, col = 1:3, bty = 'n')