find_zero: Find when an increasing function observed with noise crosses...

In brb225/FragilityTools: Toolbox For Patient Count Fragility Measures

Description

Find the smallest positive integer x-value larger than in initial value for which an approximately monotonic function f is nonnegative. This is an internal function used in general.fi.samplesize. The argument is increased exponentially until it becomes nonnegative, then a binary search is performed.

Usage

find_zero(
  f,
  x.init = 10L,
  fz.verbose = FALSE,
  D = 1,
  gamma = 0.6,
  burnin_dur = 16,
  eps = 0.1,
  proj = function(a) a,
  limits = c(1, 9999999),
  init.step = TRUE
)

Arguments

f	a function with one argument
x.init	the initial value towards finding the zero of f
fz.verbose	a logical value for whether to print status updates while running
D	an internal argument controlling the step size in Ruppert-Polyak averaging
gamma	the power of n^-1 in gradient descent
burnin_dur	the number of iterations of SGD to burn in before there's checks for the average to converge
eps	a parameter to control the error. The smaller is it, the more precise the output but the longer the function will take to run.
proj	a function to project/map the argument of the function onto the functions domain
limits	a length two numeric vector with the first entry being the lower bound and the second being an upper bound.

Value

the argument of f which satisfies the desired conditions.

Examples

FragilityTools:::find_zero(function(x) x - 100 + rnorm(1), fz.verbose = FALSE)

brb225/FragilityTools documentation built on Jan. 21, 2022, 1:26 a.m.