set_abs {cheapr}

Math operations by reference - Experimental

Description

These functions transform your variable by reference, with no copies being made. It is advisable to only use these if you know what you are doing.

Usage

set_abs(x)

set_floor(x)

set_ceiling(x)

set_trunc(x)

set_exp(x)

set_sqrt(x)

set_change_sign(x)

set_round(x, digits = 0)

set_log(x, base = exp(1))

set_pow(x, y)

set_add(x, y)

set_subtract(x, y)

set_multiply(x, y)

set_divide(x, y)

Arguments

x

A numeric vector.

digits

Number of digits to round to.

base

Logarithm base.

y

A numeric vector.

Details

These functions are particularly useful for situations where you have made a copy and then wish to perform further operations without creating more copies.
NA and NaN values are ignored though in some instances NaN values may be replaced with NA. These functions will not work on any classed objects, meaning they only work on standard integer and numeric vectors and matrices.

When a copy has to be made

A copy is only made in certain instances, e.g. when passing an integer vector to set_log(). A warning will always be thrown in this instance alerting the user to assign the output to an object because x has not been updated by reference.
To ensure consistent and expected outputs, always assign the output to the same object,
e.g. x <- set_log(x) (do this)
set_log(x) (don't do this)
x2 <- set_log(x) (Don't do this either)

No copy is made here unless x is an integer vector.

Value

The exact same object with no copy made, just transformed.

Examples

library(cheapr)
library(bench)

x <- rnorm(2e05)
options(cheapr.cores = 2)
mark(
  base = exp(log(abs(x))),
  cheapr = set_exp(set_log(set_abs(x)))
)
options(cheapr.cores = 1)