addLoggingWrapper: Return a function which internally stores x or y values.

Description Usage Arguments Value Note Examples

View source: R/addLoggingWrapper.R

Description

Often it is desired and useful to store the optimization path, i.e., the evaluated function values and/or the parameters. Not all optimization algorithms offer such a trace. This wrapper makes a smoof function handle x/y-values itself.

Usage

1
addLoggingWrapper(fn, logg.x = FALSE, logg.y = TRUE)

Arguments

fn

[smoof_function]
Smoof function.

logg.x

[logical(1)]
Should x-values be logged? Default is FALSE.

logg.y

[logical(1)]
Should objective values be logged? Default is TRUE.

Value

[smoof_logging_function]

Note

Logging values, in particular logging x-values, will substantially slow down the evaluation of the function.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
# We first build the smoof function and apply the logging wrapper to it
fn = makeSphereFunction(dimensions = 2L)
fn = addLoggingWrapper(fn, logg.x = TRUE)

# We now apply an optimization algorithm to it and the logging wrapper keeps
# track of the evaluated points.
res = optim(fn, par = c(1, 1), method = "Nelder-Mead")

# Extract the logged values
log.res = getLoggedValues(fn)
print(log.res$pars)
print(log.res$obj.vals)
log.res = getLoggedValues(fn, compact = TRUE)
print(log.res)

smoof documentation built on Feb. 18, 2020, 5:11 p.m.