fp: Implementation of fixed-point algorithm

In wahani/fuTools: fuTools

Description

fp will iteratively evaluate a fixed-point function. It only supplies the user with some generally usefull and standardised output.

Usage

fp(fun, x0, opts = list(tol = 1e-06, maxiter = 100), ...)

Arguments

fun	the function returning the 'fixed-point' of x
x0	starting value for x. Will be coerced to numeric.
opts	list of options used in the algorithm. At this point only tol and maxiter are supported
...	additional arguments passed to fun

Value

Returns a list including:

• x numeric. The solution for x or the value at the last iteration if the maximum number of iterations are reached or fun results in an error.

• iterations numeric. The iteration where the algorithm stopped.

• returnStatus numeric. Code indicating the return status:

  • 0 reached tolerance criterion

  • 1 reached max number of iterations

  • 2 exit with error - last valid x is returned

• errorMessage if fun resulted in an error, the error message

• errorCall if fun resulted in an error, the call

Examples

iterator <- function(x) {
force(x)
function(guess) {
 ret <- mean(c(guess, x/guess))
 if(ret == -Inf) stop("algorithm ended with -Inf!")
 ret
}
}

# Finding the square root of 2:
x2 <- fp(fun = iterator(2), x0 = 1)

print(x2)
summary(x2)

# Finding the square root of -1 - fun will result in an error:
x_1 <- fp(fun = iterator(-1), x0 = 1)

print(x_1)
summary(x_1)

wahani/fuTools documentation built on May 3, 2019, 7:38 p.m.