# [R-pkgs] "The Matrix" is approaching version 1.0-0

Martin Maechler maechler at stat.math.ethz.ch
Fri Mar 28 15:18:37 CET 2008

```A new version of "the Matrix"
(well, actually the R package named "Matrix") has become
available on the CRAN mirrors.

As some of you have noticed, the version numbers (current is version
0.999375-8) are converging to one, and we feel that we have solved
enough of the many (mostly small) problems to announce that release
1.0-0 is imminent.

In the DESCRIPTION of the package we say
--------------------------------------------------------------
Title: A Matrix package for R
Author: Douglas Bates <....> and Martin Maechler <....>
Maintainer: Doug and Martin <Matrix-authors at R-project.org>
Description: Classes and methods for dense and sparse matrices and
operations on them using Lapack, CSparse and CHOLMOD
--------------------------------------------------------------

The Matrix package provides efficient methods for several formal
(i.e. S4) classes of matrices where each of the actual classes are
some combination of the following three categories

1) dense or sparse
2) symmmetric, triangular, diagonal or "general" (or "permutation")
3) numeric ("d"ouble), logical (TRUE/FALSE/NA) or "patter[n]" (0/1) matrices

Interfaces to many efficient algorithms from Lapack (for "dense") and from
CHOLMOD / Csparse (for "sparse") are all implemented via method definitions
for the customary linear algebra functions
%*%, t(), crossprod(), tcrossprod(), chol(), qr(), solve(),
colSums(), rowSums(), kronecker(), determinant(), ...
and for various formal groups of generics, such as
"Math" (sin,exp,gamma,..)  "Arith" (+,-,*,...),
"Logic" (>, <=, ..),  "Summary" (sum, max, ...)  etc;
is.na()

Furthermore, 'indexing' : "A[...]" and "A[..] <- value" of
all(!) kinds of S/R indexing and some new generic functions such as

lu()     {LU decomposition}
Schur(), BunchKaufman(),
norm(), rcond()        {Matrix norms and condition numbers}
expm()   {Matrix exponential},
band(), triu(), tril()  {extract band-diagonal or triangular sub-matrices}
symmpart(), skewpart()   { (x + t(x))/2  and  (x - t(x)) / 2 }

are provided. Further, an extension to the xtabs function
xtabs(*, sparse=TRUE)
for large sparse, two-way contingency tables
and coercion of one *factor* (possibly crossed with one <numeric>)
to the corresponding (potentially huge) sparse model matrix for sparse
least squares and related computations.

Further to the above, "Matrix" objects are also constructed by
Matrix(), spMatrix(), bdiag() {block-diagonal}, Diagonal() and many
as(., "....Matrix") possibilities.

The Matrix package also provides a C level API (header files of
exported C functions providing low-level functionality) to many of its
internal algorithms that other packages can link to.  Currently, the
'lme4' package makes heavy use of these exported C functions.

---------------------------------------------------------------------------

One of the things we plan to improve considerably is the documentation
for the package.  Currently there are four vignettes but all but the

Comparisons: Comparisons of Least Squares calculation speeds

are really not complete in one way or another.

---------------------------------------------------------------------------

We would appreciate current users of the Matrix package (and also
generally interested useRs) exploring the package's capabilities and
giving us feedback about problems that they might encounter or missing
features, inefficiencies and maybe even "infelicities" (a.k.a. bugs).
Fixing problems before the release of the 1.0-0 version of "The Matrix",
rather than after its release, is our preferred approach.

Douglas Bates and Martin Maechler
Matrix-authors <at> r-project <.> org

```