\name{dsBuildRMatrix}
\alias{dsBuildRMatrix}
\title{Builds a matrix summarizing parameters for a database}
\description{
  This applies \code{sumFun} to each link model parameter set in a ASP
  database and returns a big data frame giving the result of the function
  for each object.  The summary function, \code{sumFun}, should return a
  vector of the same length for each object.
}
\usage{
dsBuildRMatrix(FUN, ..., chains, window, linkModelList = listLinkModels(), trimRegExp = NULL, maxparams = 99, silent = TRUE)
}
\arguments{
  \item{FUN}{A summary function for a parameter set.  This function
    should return a labeled vector of statistics for the parameter set.
    It could also return \code{NA} if the parameter set was \code{NA}
    (due to thining).}
  \item{\dots}{Additional arguments passed to \code{FUN}.}
  \item{chains}{A vector of chains to extract data for.  By default it
    uses all chains which have positive length.}
  \item{window}{A vector containing the first and last cycle to use.  Last
  cycle defaults to length of smallest chain in database.  Burn-in
  defaults to 1001 for long series, 501 for short series and 1 for very
  short series.}
  \item{linkModelList}{List of link models to use. (Defaults to all in
    database after filtering.)}
  \item{trimRegExp}{A regular expression for names of link or student
    models which will be trimmed from the list of models to be read.}
  \item{maxparams}{A interger given the maximum number of parameter to
    be read in a parameter set.  This saves reading really big parameter
    into memory and may avoid memory problems on some PCs.
  }
  \item{silent}{Boolean flag, if true print messages about progress.}
}
\details{
  Connects to the current database (the last one defined using
  \code{asp.dbConnect}) and extracts a summary from each parameter set
  and model, putting it into a row of a data frame.  The summary
  function \code{FUN} is assumed to return a numeric vector of a fixed
  length, which is bound into a data frame.
}
\value{
  A \code{data.frame} giving the results for each parameter set.  The
  first three columns provide the name of the task, the name of the
  evidence model and the name of the observable respectively.
}
\note{
  This function does much the same work as \code{daCapo} followed by
  \code{buildRMatrix}.  However, it does not produce the large
  \code{daCapo} object.  Thus it is more efficient with memory, but less
  efficient with time.
}
\author{Russell Almond}
\seealso{\code{\link{buildRMatrix}},\code{\link{meanVar}},
\code{\link{mvr}}, \code{\link{daCapo}}}
\examples{\dontrun{
## Setting maxparams=15 screeens out hyperDirichlet distributions.
meanVarTable <- dsBuildRMatrix(mvr,
                                  c("Difficulty","logDisc","DiffInc"),
                                  maxparams=15)
}
}
\keyword{database}% at least one, from doc/KEYWORDS