sort2: Sorting vectors

In grr: Alternative Implementations of Base R Functions

Description

Simplified implementation of sort. For large vectors, typically is about 2x faster for numbers and 20x faster for characters and factors.

Usage

sort2(x)

Arguments

x	a vector of class numeric, integer, character, factor, or logical. Long vectors are not supported.

Examples

chars<-as.character(sample(1e3,1e4,TRUE))
system.time(a<-sort(chars))
system.time(b<-sort2(chars))
identical(a,b)  
 
ints<-as.integer(sample(1e3,1e4,TRUE))
system.time(a<-sort(ints))
system.time(b<-sort2(ints))
identical(a,b)
 
nums<-runif(1e4)
system.time(a<-sort(nums))
system.time(b<-sort2(nums))
identical(a,b)

logs<-as.logical(sample(0:1,1e6,TRUE))
system.time(result<-sort(logs))
system.time(result<-sort2(logs))

facts<-as.factor(as.character(sample(1e3,1e4,TRUE)))
system.time(a<-sort(facts))
system.time(b<-sort2(facts))
identical(a,b)

#How are special values like NA and Inf handled?
#For numerics, values sort intuitively, with the important note that NA and
#NaN will come after all real numbers but before Inf.
sort2(c(1,2,NA,NaN,Inf,-Inf))
#For characters, values sort correctly with NA at the end.
sort2(c('C','B',NA,'A'))
#For factors, values sort correctly with NA at the end
sort2(as.factor(c('C','B',NA,'A')))

## Not run: 
chars<-as.character(sample(1e5,1e6,TRUE))
system.time(a<-sort(chars))
system.time(b<-sort2(chars))

ints<-as.integer(sample(1e5,1e6,TRUE))
system.time(result<-sort(ints))
system.time(result<-sort2(ints))

nums<-runif(1e6)
system.time(result<-sort(nums))
system.time(result<-sort2(nums))

logs<-as.logical(sample(0:1,1e7,TRUE))
system.time(result<-sort(logs))
system.time(result<-sort2(logs))

facts<-as.factor(as.character(sample(1e5,1e6,TRUE)))
system.time(a<-sort(facts))
system.time(b<-sort2(facts))

## End(Not run)

grr documentation built on May 2, 2019, 5:57 a.m.