1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
|
\name{extract_sparse_array}
\alias{extract_sparse_array}
\alias{extract_sparse_array,ANY-method}
\title{extract_sparse_array}
\description{
\code{extract_sparse_array()} is an internal generic function that
is the workhorse behind the default \code{\link{read_block_as_sparse}()}
method. It is not intended to be used directly by the end user.
It is similar to the \code{\link[S4Arrays]{extract_array}()} internal
generic function defined in the \pkg{S4Arrays} package, with the major
difference that, in the case of \code{extract_sparse_array()}, the
extracted array data is returned as a \link{SparseArray} object instead
of an ordinay array.
}
\usage{
extract_sparse_array(x, index)
\S4method{extract_sparse_array}{ANY}(x, index)
}
\arguments{
\item{x}{
An array-like object for which \code{\link[S4Arrays]{is_sparse}(x)}
is \code{TRUE}.
}
\item{index}{
An unnamed list of integer vectors, one per dimension in \code{x}.
Each vector is called a \emph{subscript} and can only contain
positive integers that are valid 1-based indices along the corresponding
dimension in \code{x}.
Empty or missing subscripts are allowed. They must be represented
by list elements set to \code{integer(0)} or \code{NULL}, respectively.
The subscripts cannot contain NAs or non-positive values.
Individual subscripts are NOT allowed to contain duplicated indices.
This is an important difference with \code{\link[S4Arrays]{extract_array}}.
}
}
\details{
\code{extract_sparse_array()} should \emph{always} be called on
an array-like object \code{x} for which \code{is_sparse(x)} is
\code{TRUE}. Also it should \emph{never} be called with duplicated
indices in the individual list elements of the \code{index} argument.
For maximum efficiency, \code{extract_sparse_array()} methods should:
\enumerate{
\item NOT check that \code{\link[S4Arrays]{is_sparse}(x)} is \code{TRUE}.
\item NOT check that the individual list elements in \code{index}
contain no duplicated indices.
\item NOT try to do anything with the dimnames on \code{x}.
\item always operate natively on the sparse representation of the
data in \code{x}, that is, they should never \emph{expand}
it into a dense representation (e.g. with \emph{as.array()}).
}
Like for \code{\link[S4Arrays]{extract_array}()},
\code{extract_sparse_array()} methods need to support empty or
missing subscripts. For example, if \code{x} is an M x N matrix-like
object for which \code{is_sparse(x)} is \code{TRUE}, then
\code{extract_sparse_array(x, list(NULL, integer(0)))} must
return an M x 0 \link{SparseArray} derivative, and
\code{extract_sparse_array(x, list(integer(0), integer(0)))}
a 0 x 0 \link{SparseArray} derivative.
}
\value{
A \link{SparseArray} derivative (\link{COO_SparseArray} or
\link{SVT_SparseArray}) of the same \code{type()} as \code{x}.
For example, if \code{x} is an object representing an M x N sparse
matrix of complex numbers (i.e. \code{type(x) == "complex"}), then
\code{extract_sparse_array(x, list(NULL, 2L))} must return the 2nd column
in \code{x} as an M x 1 \link{SparseArray} derivative of \code{type()}
\code{"complex"}.
}
\seealso{
\itemize{
\item \code{\link[S4Arrays]{is_sparse}} in the \pkg{S4Arrays} package
to check whether an object uses a sparse representation of the
data or not.
\item \link{SparseArray} objects.
\item \code{S4Arrays::\link[S4Arrays]{type}} in the \pkg{S4Arrays}
package to get the type of the elements of an array-like object.
\item \code{\link{read_block_as_sparse}} to read array blocks as
SparseArray objects.
\item \code{\link[S4Arrays]{extract_array}} in the \pkg{S4Arrays} package.
\item \linkS4class{dgCMatrix} objects implemented in the \pkg{Matrix}
package.
}
}
\examples{
extract_sparse_array
showMethods("extract_sparse_array")
## --- On a dgCMatrix object ---
m <- matrix(0L, nrow=6, ncol=4)
m[c(1:2, 8, 10, 15:17, 24)] <- (1:8)*10L
dgcm <- as(m, "dgCMatrix")
dgcm
extract_sparse_array(dgcm, list(3:6, NULL))
extract_sparse_array(dgcm, list(3:6, 2L))
extract_sparse_array(dgcm, list(3:6, integer(0)))
## --- On a SparseArray object ---
a <- array(0L, dim=5:3, dimnames=list(letters[1:5], NULL, LETTERS[1:3]))
a[c(1:2, 8, 10, 15:17, 20, 24, 40, 56:60)] <- (1:15)*10L
svt <- as(a, "SparseArray")
svt
extract_sparse_array(svt, list(NULL, 4:2, 1L))
extract_sparse_array(svt, list(NULL, 4:2, 2:3))
extract_sparse_array(svt, list(NULL, 4:2, integer(0)))
}
\keyword{internal}
\keyword{array}
\keyword{methods}
|
