RcppAlgos (version 2.9.3)

eulerPhiSieve: Apply Euler's Phi Function to Every Element in a Range

Description

Sieve that generates the number of coprime elements for every number between bound1 and bound2 (if supplied) or all numbers up to bound1. This is equivalent to applying Euler's phi function (often written as \(\phi(x)\)) to every number in a given range.

Usage

eulerPhiSieve(bound1, bound2 = NULL, namedVector = FALSE, nThreads = NULL)

Value

Returns a named/unnamed integer vector if max(bound1, bound2)

\(< 2^{31}\), or a numeric vector otherwise.

Arguments

bound1

Positive integer or numeric value.

bound2

Positive integer or numeric value.

namedVector

Logical flag. If TRUE, a named vector is returned. The default is FALSE.

nThreads

Specific number of threads to be used. The default is NULL.

Author

Joseph Wood

Details

For the simple case (i.e. when bound2 = NULL), this algorithm first generates all primes up to \(n\) via the sieve of Eratosthenes. We use these primes to sieve over the sequence 1:n, dividing each value by \(p\), creating a temporary value that will be subtracted from the original value at each index (i.e. equivalent to multiply each index by \((1 - 1/p)\) but more efficient as we don't have to deal with floating point numbers). The case when is.null(bound2) = FALSE is more complicated but the basic ideas still hold.

This function is very useful when you need to calculate Euler's phi function for many numbers in a range as performing this calculation on the fly can be computationally expensive.

This algorithm benefits greatly from the fast integer division library 'libdivide'. The following is from https://libdivide.com/:

• “libdivide allows you to replace expensive integer divides with comparatively cheap multiplication and bitshifts. Compilers usually do this, but only when the divisor is known at compile time. libdivide allows you to take advantage of it at runtime. The result is that integer division can become faster - a lot faster.”

References

• Euler's totient function

• ridiculousfish (author of libdivide)

• github.com/ridiculousfish/libdivide

• 53-bit significand precision

Examples

## Generate some random data
set.seed(496)
mySamp <- sample(10^6, 5*10^5)

## Generate number of coprime elements for many numbers
system.time(myPhis <- eulerPhiSieve(10^6))

## Now use result in algorithm
for (s in mySamp) {
    sPhi <- myPhis[s]
    ## Continue algorithm
}

## See https://projecteuler.net
system.time(which.max((1:10^6)/eulerPhiSieve(10^6)))

## Generating number of coprime elements
## for every number in a range is no problem
system.time(myPhiRange <- eulerPhiSieve(10^13, 10^13 + 10^6))

## Returning a named vector
eulerPhiSieve(10, 20, namedVector = TRUE)
eulerPhiSieve(10, namedVector = TRUE)

## Using nThreads
system.time(eulerPhiSieve(1e5, 2e5, nThreads = 2))