fast-dm - a fast method for diffusion model analysis


INTRODUCTION
------------

Stochastic Diffusion Models are used in cognitive science to analyse
cognitive processes in fast binary decisions. It is assumed that the 
information supporting one and refuting the other decision can be
described by a Wiener Diffusion process with constant drift over time.
In the model, the decision process is terminated when the process 
exits from a given interval.

The diffusion-model data analysis  was introduced in psychology by
Roger Ratcliff (1978). Ratcliff's diffusion model is specified by the
following parameters: The distance of thresholds (a), the starting
point (z), the drift rate (v), a non-decisional part of the response
time (t0), and three so-called inter-trial variability parameters
concerning the starting point (sz), the drift (eta), and the non-
decisional response-time constant (st).

Fast-dm estimates values for this parameters from response-time
distributions for both alternative decisions. Fast-dm is based on
a fast numerical approach for solving the Partial Differential Equation 
defining the above sketched diffusion model (cf. Voss & Voss, 2006).
For the parameter estimation, the Kolmogorov-Smirnov method introduced
by Voss, Rothermund, & Voss (2004) is used for the optimisation 
criterion.

Fast-dm comes with NO WARRANTY, to the extent permitted by law.  You
may redistribute copies of fast-dm  under the terms of the GNU General
Public License.  For more information about these matters, read the
file COPYING of the source code distribution.

Please mail any suggestions and bug reports to <voss@seehuhn.de>.


REFERENCES
----------
Ratcliff, R., 1978. A theory of memory retrieval. Psychological Review,
  85, 59-108.

Ratcliff, R., Rouder, J. N., 1998. Modelling response times for two-choice
  decisions. Psychological Science 9 (5), 347--356.

Voss, A., Rothermund, K., Voss, J., 2004. Interpreting the parameters of the
  diffusion model: An empirical validation. Memory & Cognition 32, 1206--1220.

Voss, A., Voss, J., (2006). A Fast Numerical Algorithm for the Estimation
  of Diffusion-Model Parameters. Manuscript submitted for publication.

Wagenmakers, E.-J., van der Maas, H. L. J., & Grasman, R. P. P. P., In Press.
  An ez-diffusion model for response time and accuracy. 
  Psychonomic Bulletin & Review.


INSTALLATION
------------

On Unix systems:
Just type the following commands

    ./configure
    make

On Windows systems:
As usual, on a Windows system things are a little more complicated. The
following steps may help you if you are using Microsoft's Visual Studio 7:
(1) Create a new project. Choose a "win32 Console project", and make sure 
    that the "empty project" check-box is selected.
(2) Copy all "fast-dm" files in the new project directory.
(3) Add the source code to the project. You need cdf.c, dataset.c, erf.c, 
    experiment.c, EZ-diff.c, file.c, main.c, pde.c, phi.c, simplex.c, 
    xmalloc.c, and fast-dm.h
(4) In the project settings, select "compile as C-code" (and not C++).
(5) Compile the project.


INTERNALS
---------

The program is split into several source files.  The common header
file for all of these is "fast-dm.h".

    main.c - main program for the fast-dm project
    EZ-diff.c - estimate some parameters using the EZ-model
    simplex.c - the downhill simplex method of Nelder and Mead
    cdf.c - compute the CDF for the diffusion model
    phi.c - the CDF and inverse CDF of the standard normal distribution
    pde.c - numerically solve the Fokker-Planck equation
    readdata.c - read the data file into memory
    file.c - read control and data files
    erf.c - Gaussian error function (only needed on MS Windows)

    plot-cdf.c - plot cumulative distribution functions
