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
|
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Do not modify this file since it was automatically generated from:
%
% whichVector.R
%
% by the Rdoc compiler part of the R.oo package.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\name{whichVector.logical}
\alias{whichVector.logical}
\alias{whichVector.matrix}
\title{Identifies TRUE elements in a logical vector}
\description{
Identifies TRUE elements in a logical vector.
\emph{NOTE: \code{\link[base]{which}}() should be used instead of this method}
unless you are running R (< 2.11.0), for which this method is faster
than \code{\link[base]{which}}() for \code{\link[base]{logical}} \code{\link[base]{vector}}s, especially when there
are no missing values.
}
\usage{
\method{whichVector}{logical}(x, na.rm=TRUE, use.names=TRUE, ...)
}
\arguments{
\item{x}{A \code{\link[base]{logical}} \code{\link[base]{vector}} of length N.}
\item{na.rm}{If \code{\link[base:logical]{TRUE}}, missing values are treated as \code{\link[base:logical]{FALSE}},
otherwise they are returned as \code{\link[base]{NA}}.}
\item{use.names}{If \code{\link[base:logical]{TRUE}}, the names attribute is preserved,
otherwise it is not return.}
\item{...}{Not used.}
}
\value{
Returns an \code{\link[base]{integer}} \code{\link[base]{vector}} of length less or equal to N.
}
\section{Benchmarking}{
In R v2.11.0 \code{\link[base]{which}}() was made approx. 10 times
faster via a native implementation. Because of this, this
method is of little use and approximately 3 times slower.
However, for earlier version of R, this method is still
significantly faster. For example,
simple comparison on R v2.7.1 on Windows XP, show that
this implementation can be more than twice as fast as
\code{\link[base]{which}}(), especially when there are no missing
value (and \code{na.rm=FALSE}) is used.
}
\examples{\dontrun{
# - - - - - - - - - - - - - - - - - - - - - - - - - -
# Simulate two large named logical vectors,
# one with missing values one without
# - - - - - - - - - - - - - - - - - - - - - - - - - -
N <- 1e6
# Vector #1
x <- sample(c(TRUE, FALSE), size=N, replace=TRUE)
names(x) <- seq_along(x)
# Vector #2
y <- x
y[sample(N, size=0.1*N)] <- NA
# - - - - - - - - - - - - - - - - - - - - - - - - - -
# Validate consistency
# - - - - - - - - - - - - - - - - - - - - - - - - - -
stopifnot(identical(which(x), whichVector(x)))
stopifnot(identical(which(y), whichVector(y)))
# - - - - - - - - - - - - - - - - - - - - - - - - - -
# Benchmarking
# - - - - - - - - - - - - - - - - - - - - - - - - - -
# Number of iterations
K <- 5
t1 <- 0
for (kk in 1:K) {
t1 <- t1 + system.time({ idxs1 <- which(x) })
}
t2 <- 0
for (kk in 1:K) {
t2 <- t2 + system.time({ idxs2 <- whichVector(x, na.rm=FALSE) })
}
cat(sprintf("whichVector(x, na.rm=FALSE)/which(x): \%.2f\n", (t2/t1)[3]))
stopifnot(identical(idxs1, idxs2))
t1 <- 0
for (kk in 1:K) {
t1 <- t1 + system.time({ idxs1 <- which(y) })
}
t2 <- 0
for (kk in 1:K) {
t2 <- t2 + system.time({ idxs2 <- whichVector(y) })
}
cat(sprintf("whichVector(y)/which(y): \%.2f\n", (t2/t1)[3]))
stopifnot(identical(idxs1, idxs2))
}}
\author{Henrik Bengtsson}
\seealso{
\code{\link[base]{which}}()
}
\keyword{methods}
\keyword{programming}
\keyword{internal}
|
