Multiscale Modeling -- A Bayesian Perspective
-
R code
This package for R also requires the packages nlme and akima,
which can be found on CRAN.
Installation from source also requires a fortran compiler (such as g95).
To install, download this Gzip file of
code, then do one of the following:
- From a unix prompt, with root access, enter
R CMD INSTALL mrm_1.0.tar.gz
Then start up R and enter
library(mrm)
- From a unix prompt, without root access, first create a
directory for R libraries such as ~/Rlibs, then enter
R CMD INSTALL --library=~/Rlibs mrm_1.0.tar.gz
Then start up R and enter
library(mrm, lib.loc="~/Rlibs")
- From within R, use the install.packages function
Note: we have noticed that the latest versions of R on Max OSX
with PowerPC chips (e.g., G4, G5) have trouble with compiling
packages from source because of links to the fortran compiler. One
way to fix this is to first install gfortran (and adding
/usr/local/bin to your path if necessary), then install R from
its source code and to add
"F77=gfortran" to the configuration (e.g., before running
"make", run "./configure F77=gfortran
--with-blas='-framework vecLib' --with-lapack --with-aqua
--enable-R-framework"). This may also require you to install
nlme and/or akima from source, i.e., install.packages("akima",
type="source").
The contents of this package can be viewed from
within R with help(package=mrm).
Code examples (included in the package): Convolutions GPs Hidden Resolution Models
These examples can be accessed in R with the
example command.
- Errata: let us know if you find any corrections
marf at vt.edu
herbie at soe.ucsc.edu