makeR stands for making/generating/simulating responses from
a LBA model. make_r and make.r use C++ function. These
make r, _r, .r functions are essentially rLBA,
including rlba_norm. They uses a LBA model with parameters, b, A,
mean_v, sd_v and t0 (no st0) to generate choice RT random deviates.
maker(drifts, n, b, A, n_v, t0, st0 = 0, seed = NULL,
return_ttf = FALSE)a n x n_v drift rate matrix. It can be a vector with 2 or more elements. n is the numbers of observation. n_v is the numbers of response/accumulator.
numbers of observation/model simulations. This must be a scalar.
decision threshold, a vector or a scalar.
start point upper bound, a vector of a scalar.
numbers of response/accumulator, an integer. Note n_v must match
the length/size of drifts vector.
nondecision time, a vector or a scalar.
nondecision time variation, a vector of a scalar. It is the upper bound of a uniform distribution for t0 variability.
an integer specifying if and how the random number generator should be initialized.
a boolean switch indicating if return RTs for all
accumulators. When return_ttf is TRUE, a n_v x n ttf matrix is
returned.
make_r gives either a time-to-finish (ttf) matrix or a n x 2
matrix, storing RTs (first column) and responses (second column). n
equals to number of model simulations. ttf is a n_v x n matrix with RTs from
all accumulators.
make_v draws drift rate from normal or truncated normal distribution.
Each trial is stored as a row and each column is a drift rate for an
accumulator. You need to transpose drift rates generated by make_v for
makeR.
make.r is a wrapper function of make_r. You may
need to use ":::" to call make.r, because of S3 method naming convention. If
you call make_r directly, beware it returns C index and is only a
numeric matrix. It does not carry a string vector for the column names, RTs
and responses. See timing test to see why it might be a good idea not to
return it as a data frame. rlbaCnorm is R version of correlated LBA
model.
rlba_norm adds checks and calls make_v and make_r.
rlba_norm is only slightly quicker than make_r.
n1PDFfixedt0 is defective density function for the fisrt node LBA
model. Defective means its probability does not normally normalize to 1.
Only the probabilities from all nodes/accumulators add to 1.
n1PDFfixedt0 is equation (3) on page 159 in Brown and
Heathcote (2008). This equation assumes st0 is 0.
fptcdf and fptpdf are distribution and density functions with
four parameters A, b, v and sv, assuming t0 is zero. fptcdf and
fptpdf are respectively equation (1) and equation (2) on page 159 in
Brown and Heathcote (2008).