recording_function: Recording functions
In olafmersmann/soobench: Single Objective Optimization Benchmark Functions

Description Usage Arguments See Also Examples

Return a new function which is identical to the soofunction passed in except that evaluated parameter settings and function values are recorded.

1	recording_function(fn, record_pars = TRUE, record_values = TRUE, predicate)

`fn`	A test function (class `soo_function` or `wrapped_soo_function`).
`record_pars`	[`boolean(1)`] If `TRUE`, parameter values (x) will be recorded.
`record_values`	[`boolean(1)`] If `TRUE`, function values (y) will be recorded.
`predicate`	[`function(par, value, time)`] Predicate function that returns `TRUE` if recoding should take place for the given parameter setting `par`, function value `value` and evaluation `time`. Note that `time` is measured in function evaluations.

recorded_values to retrieve the recorded parameter and function values.

fn <- recording_function(generate_sphere_function(2))
X <- random_parameters(10, fn)
y <- fn(X)
rv <- recorded_values(fn)
all.equal(rv$par, X)
all.equal(rv$value, y)

## With a predicate
pv <- function(par, value, time)
  time %% 3 == 0
fn <- recording_function(generate_sphere_function(2), predicate=pv)
X <- random_parameters(10, fn)
y <- fn(X)
rv <- recorded_values(fn)
all(rv$time %% 3 == 0)
all.equal(rv$par, X[, rv$time])
all.equal(rv$value, y[rv$time])