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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
|
\name{is_nonzero}
\alias{is_nonzero}
\alias{is_nonzero,ANY-method}
\alias{is_nonzero,sparseMatrix-method}
\alias{nzcount}
\alias{nzcount,ANY-method}
\alias{nzcount,CsparseMatrix-method}
\alias{nzcount,RsparseMatrix-method}
\alias{nzcount,TsparseMatrix-method}
\alias{nzwhich}
\alias{nzwhich,ANY-method}
\alias{nzwhich,CsparseMatrix-method}
\alias{nzwhich,RsparseMatrix-method}
\alias{nzvals}
\alias{nzvals,ANY-method}
\alias{nzvals,dgCMatrix-method}
\alias{nzvals,lgCMatrix-method}
\alias{nzvals,nMatrix-method}
\alias{nzvals<-}
\alias{nzvals<-,ANY-method}
\alias{sparsity}
\title{is_nonzero() and the nz*() functions}
\description{
A set of functions for direct manipulation of the nonzero elements
of an array-like object.
}
\usage{
is_nonzero(x)
nzcount(x)
nzwhich(x, arr.ind=FALSE)
nzvals(x)
nzvals(x) <- value
sparsity(x)
}
\arguments{
\item{x}{
Typically (but not necessarily) an array-like object that is sparse,
like a \link{SparseArray} derivative, or a dg[C|R]Matrix or lg[C|R]Matrix
object from the \pkg{Matrix} package.
However, \code{x} can also be an ordinary matrix or array, or any
matrix-like or array-like object.
}
\item{arr.ind}{
If \code{arr.ind=FALSE} (the default), the indices of the nonzero
array elements are returned in a numeric vector (a.k.a. \emph{L-index}).
Otherwise, they're returned in an ordinary matrix (a.k.a. \emph{M-index}).
See \code{?\link[S4Arrays]{Lindex}} in the \pkg{S4Arrays} package for
more information about \emph{L-index} and \emph{M-index}, and how to
convert from one to the other.
Note that using \code{arr.ind=TRUE} won't work if \code{nzcount(x)}
is >= \code{.Machine$integer.max} (= 2^31), because, in that case,
the returned \emph{M-index} would need to be a matrix with more rows
than what is supported by base R.
}
\item{value}{
A vector, typically of length \code{nzcount(x)} (or 1) and type
\code{type(x)}.
}
}
\details{
\code{nzcount(x)} and \code{nzwhich(x)} are equivalent to, but
typically more efficient than, \code{sum(is_nonzero(x))} and
\code{which(is_nonzero(x))}, respectively.
\code{nzvals(x)} is equivalent to, but typically more efficient than,
\code{x[nzwhich(x)]} (or \code{x[is_nonzero(x)]}).
\code{nzvals(x) <- value} replaces the values of the nonzero array
elements in \code{x} with the supplied values. It's equivalent to,
but typically more efficient than, \code{x[nzwhich(x)] <- value}.
Note that \code{nzvals(x) <- nzvals(x)} is guaranteed to be a no-op.
}
\value{
\code{is_nonzero()}: An array-like object of \code{type()} \code{"logical"}
and same dimensions as the input object.
\code{nzcount()}: The number of nonzero array elements in \code{x}.
\code{nzwhich()}: The indices of the nonzero array elements in \code{x},
either as an \emph{L-index} (if \code{arr.ind} is \code{FALSE}) or as
an \emph{M-index} (if \code{arr.ind} is \code{TRUE}).
Note that the indices are returned sorted in strictly ascending order.
\code{nzvals()}: A vector of the same \code{type()} as \code{x} and
containing the values of the nonzero array elements in \code{x}.
Note that the returned vector is guaranteed to be \emph{parallel}
to \code{nzwhich(x)}.
\code{sparsity(x)}: The ratio between the number of zero-valued
elements in array-like object \code{x} and its total number of
elements (\code{length(x)} or \code{prod(dim(x))}).
More precisely, \code{sparsity(x)} is \code{1 - nzcount(x)/length(x)}.
}
\seealso{
\itemize{
\item \link{is_nonna} for \code{is_nonna()} and \code{nna*()} functions
\code{nnacount()}, \code{nnawhich()}, etc...
\item \link{SparseArray} objects.
\item S4 classes \linkS4class{dgCMatrix}, \linkS4class{lgCMatrix},
and \linkS4class{ngCMatrix} defined in the \pkg{Matrix} package.
\item Ordinary \link[base]{array} objects in base R.
\item \code{base::\link[base]{which}} in base R.
}
}
\examples{
a <- array(rpois(120, lambda=0.3), dim=c(5, 12, 2))
is_nonzero(a)
## Get the number of nonzero array elements in 'a':
nzcount(a)
## nzwhich() returns the indices of the nonzero array elements in 'a'.
## Either as a "L-index" i.e. an integer (or numeric) vector of
## length 'nzcount(a)' containing "linear indices":
nzidx <- nzwhich(a)
length(nzidx)
head(nzidx)
## Or as an "M-index" i.e. an integer matrix with 'nzcount(a)' rows
## and one column per dimension where the rows represent "array indices"
## (a.k.a. "array coordinates"):
Mnzidx <- nzwhich(a, arr.ind=TRUE)
dim(Mnzidx)
## Each row in the matrix is an n-tuple representing the "array
## coordinates" of a nonzero element in 'a':
head(Mnzidx)
tail(Mnzidx)
## Extract the values of the nonzero array elements in 'a' and return
## them in a vector "parallel" to 'nzwhich(a)':
a_nzvals <- nzvals(a) # equivalent to 'a[nzwhich(a)]'
length(a_nzvals)
head(a_nzvals)
nzvals(a) <- log1p(nzvals(a))
a
## Sanity checks:
stopifnot(
identical(nzidx, which(a != 0)),
identical(Mnzidx, which(a != 0, arr.ind=TRUE, useNames=FALSE)),
identical(nzvals(a), a[nzidx]),
identical(nzvals(a), a[Mnzidx]),
identical(`nzvals<-`(a, nzvals(a)), a)
)
}
\keyword{array}
\keyword{methods}
|
