This project extends the R statistical programming environment.  Package bigmemory supports the creation, storage, access, and manipulation of massive matrices.  These matrices are allocated to shared memory and may use memory-mapped files.  Packages biganalytics, bigtabulate, synchronicity, and bigalgebra (please see the bigalgebra page for 32-bit/64-bit library information) provide advanced functionality.  We provide a short overview with examples in the Documentation area.

NEWS (July 2011).  Version 4.2.11 has fixed a few minor problems and has been uploaded to CRAN.  We note some problem with newer macos and package bigtabulate and will try to track that down.  There are also some more obscure bugs (for example, with read.big.matrix in selected cases) that we will try to address.  We encourage emails if you need advice or want to offer feedback.

The Bigmemory Project was awarded the 2010 John M. Chambers Statistical Software Award by the ASA Sections on Statistical Computing and Statistical Graphics.

Obtaining bigmemory and other packages from R-Forge: we recommend commands like the following (noting that these should be considered development versions, with the CRAN versions as "stable":

install.packages("bigmemory", repos="")

OLDER NEWS (May 4, 2010).  After fixing some obscure warnings, we uploaded bigmemory version 4.2.1 to CRAN; it will hopefully be available soon.  This new version is minimalist.  The sister packages will follow over the next few days.  Mutex support (locking with shared memory) has been abstracted to package synchronicity; the basic tools for exploratory data analysis as well as more advanced analytics are being relocated to package biganalytics; new package bigalgebra is in the pipeline.  Package bigtabulate provides table-, split-, and tapply-like functionality; while it is certainly of interest to users, it also provides a self-contained example for package developers interested in bigmemory and the process of building tools for use with both big.matrix and matrix objects.