spam-class {spam} | R Documentation |
The spam
class is a representation of sparse matrices.
Objects can be created by calls of the form new("spam", entries, colindices,
rowpointes, dimension)
.
The standard "old Yale sparse format" is used to store sparse matrices.
The matrix
x
is stored in row form. The first element of row i
is
x@rowpointers[i]
. The length of row i
is determined by
x@rowpointers[i+1]-x@rowpointers[i]
. The column indices of x
are stored in
the x@colindices
vector. The column index for element x@entries[k]
is
x@colindices[k]
.
entries
:"numeric"
contains the
nonzero values colindices
:"integer"
ordered indices
of the nonzero values rowpointers
:"integer"
pointer to the beginning
of each row in the arrays entries
and colindices
dimension
:"integer"
~~ signature(x = "spam")
:
transforming a sparse
matrix into a regular matrix.signature(x = "spam")
:
cleaning of a sparse matrix.signature(x = "spam", i,j, value)
:
assigning a
sparse matrix. The negative vectors are not implemented yet. signature(x = "spam", i, j)
:
subsetting a
sparse matrix. The negative vectors are not implemented yet. signature(x, y)
:
matrix multiplication, all combinations of sparse with full
matrices or vectors are implemented.signature(x = "spam")
:
vectorizes the sparse matrix and takes account of the zeros. Hence
the lenght of the result is prod(dim(x))
.signature(x = "spam")
: binds sparse matrices.signature(x = "spam")
:
see chol
for details.signature(x = "spam")
:
see diag
for details.signature(x = "spam")
: truncates or augments
the matrix see dim
for details.signature(x = "spam")
: gives the dimension of the
sparse matrix.signature(x = "spam")
:
see image
for details.signature(x = "spam")
:
see display
for details.signature(x = "spam")
: Is not implemented and
causes an error.signature(x = "spam")
: gives the number of
non-zero elements.signature(x = "spam")
: see lower.tri
for details. signature(x = "spam")
: see Math
for details.signature(x = "spam")
: see Math2
for details. signature(x = "spam")
: calculates the norm of a matrix.signature(x = "spam", y)
: same functionality as
the ordinary plot
. signature(x = "spam")
: see print
for details.signature(x = "spam")
: binds sparse
matrices.signature(a = "spam")
: see solve
for details.signature(object = "spam")
: small summary
statement of the sparse matrix.signature(x = "spam")
:
All functions of the Summary
class (like min
, max
,
range
...) operate on the vector x@entries
and return the
result thereof. See Examples.
signature(x = "spam")
: transpose of a sparse matrix.signature(x = "spam")
: see lower.tri
for details.
The compressed sparse row (CSR) format is often described with the
vectors a
, ia
, ja
. To be a bit more
comprehensive, we have chosen longer slot names.
The slots colindices
and rowpointers
are
tested for proper integer assignments. This is not true for
entries
.
Reinhard Furrer, some of the Fortran code is based on A. George, J. Liu, E. S. Ng, B.W Peyton and Y. Saad (alphabetical)
showMethods("as.spam") smat <- diag.spam(runif(15)) range(smat) cos(smat)