Matrix (version 0.999375-20)

spqr: SuiteSparse QR factorization of Sparse Matrix

Description

SuiteSparse QR ("SPQR") implements a multifrontal sparse QR factorization method.

Usage

spqr(x, ...)
## S3 method for class 'dgCMatrix':
spqr(x, econ = 0,
     ordering = c("default", "fixed", "natural", "amd", "colamd",
                  "metis", "best", "bestamd"),
     tol = -2)

Arguments

x

a sparse matrix, i.e., an Robject inheriting from class sparseMatrix.

econ

integer for economy, i.e. specifying the number of rows of $R$ or columns of $Q$ to return. The default is m <- nrow(x).

ordering

a string specifying the ordering strategy.

tol

a number specifying the tolerance to be applied. If tol <= -2<="" code="">, a default tolerance is used; if $-2 < tol < 0$, then no tolerance is used, and if $tol >= 0$ it is used as tolerance, namely columns with 2-norm

...

further arguments passed to methods.

Value

• an object of class "SPQR", notably with slots Q and R containing the factorizing matrices.

References

http://www.cise.ufl.edu/research/sparse/SPQR/ has links to publications and a user manual.

See Also

qr for long matrices, uses a simpler algorithm from library CSparse.

Examples

data(KNex); mm <- KNex$mm
if(.Platform$OS.type != "windows") # C code seg.faults on some Windows versions
str(mm.QR <- spqr(mm))