a helper function, which extracts the lower and upper bounds of a function would be nice :)

See here or here for some important ones.

Until now we force the user to pass a named list. Maybe better allow (numeric) unnamed vectors as well?

Actually I have a function for this already in another repo. Just need to copy it and add some sanity checks.

persp3d or plot3d via rgl package or both

d> load_all()
Loading smoof
Re-compiling smoof
"C:/PROGRA1/R/R-311.2/bin/x64/R" --no-site-file --no-environ --no-save --no-restore CMD INSTALL
"C:\Users\t_tob_000\smoof" --library="C:\Users\T_TOB_~1\AppData\Local\Temp\RtmpwNST2A\devtools_install_257032d326ae"
--no-R --no-data --no-help --no-demo --no-inst --no-docs --no-exec --no-multiarch --no-test-load --preclean

• installing source package 'smoof' ...
  ** libs
  Warnung: Ausführung von Kommando 'make -f "C:/PROGRA1/R/R-311.2/etc/x64/Makeconf" -f "C:/PROGRA1/R/R-311.2/share/make/winshlib.mk" SHLIB="smoof.dll" WIN=64 TCLBIN=64 OBJECTS="bbob_r_to_c_interface.o benchmarks.o benchmarksdeclare.o benchmarkshelper.o cec09.o cec09_r_to_c_interface.o" shlib-clean' ergab Status 127
  Warnung: Ausführung von Kommando 'make -f "C:/PROGRA1/R/R-311.2/etc/x64/Makeconf" -f "C:/PROGRA1/R/R-311.2/share/make/winshlib.mk" SHLIB="smoof.dll" WIN=64 TCLBIN=64 OBJECTS="bbob_r_to_c_interface.o benchmarks.o benchmarksdeclare.o benchmarkshelper.o cec09.o cec09_r_to_c_interface.o"' ergab Status 127
  ERROR: compilation failed for package 'smoof'
• removing 'C:/Users/T_TOB_~1/AppData/Local/Temp/RtmpwNST2A/devtools_install_257032d326ae/smoof'

d> session_info()
Session info ----------------------------------------------------------------------------------------------------------------
setting value
version R version 3.1.2 (2014-10-31)
system x86_64, mingw32
ui RStudio (0.98.1102)
language (EN)
collate German_Germany.1252
tz Europe/Berlin
date 2015-09-30

Packages --------------------------------------------------------------------------------------------------------------------
package * version date source
colorspace 1.2-6 2015-03-11 CRAN (R 3.1.3)
curl 0.9.3 2015-08-25 CRAN (R 3.1.3)
devtools * 1.9.1 2015-09-11 CRAN (R 3.1.3)
digest 0.6.8 2014-12-31 CRAN (R 3.1.3)
ggplot2 1.0.1 2015-03-17 CRAN (R 3.1.3)
gtable 0.1.2 2012-12-05 CRAN (R 3.1.3)
httr 1.0.0 2015-06-25 CRAN (R 3.1.3)
magrittr 1.5 2014-11-22 CRAN (R 3.1.3)
MASS 7.3-35 2014-09-30 CRAN (R 3.1.2)
memoise 0.2.1 2014-04-22 CRAN (R 3.1.3)
munsell 0.4.2 2013-07-11 CRAN (R 3.1.3)
plyr 1.8.3 2015-06-12 CRAN (R 3.1.3)
proto 0.3-10 2012-12-22 CRAN (R 3.1.3)
R6 2.1.1 2015-08-19 CRAN (R 3.1.3)
Rcpp 0.12.1 2015-09-10 CRAN (R 3.1.3)
reshape2 1.4.1 2014-12-06 CRAN (R 3.1.3)
roxygen2 4.1.1 2015-04-15 CRAN (R 3.1.3)
scales 0.3.0 2015-08-25 CRAN (R 3.1.3)
stringi 0.5-5 2015-06-29 CRAN (R 3.1.3)
stringr 1.0.0 2015-04-30 CRAN (R 3.1.3)

Add a helper to build a scalarized single-objective smoof function from a multi-objective function.

Some functions, e.g. the Branin test function, have multiple global optima, but at the moment we allow only single values to be passed. Think about the best way to do this. Since, in the general case, we want mixed objective functions (functions with both numeric and discrete parameters) we need to pass a list of lists or better a data frame (each row a value) and also allow to pass a matrix for numeric only functions.

Some people might prefer base R plots. Should we offer them as well?

Since the autoplot function delegates to different plotting functions with different parameters we currently have no documentation for these parameters. Find a way to explain them.

Hallo Jakob,

sorry for bothering you with all that Windows stuff.
I made a dt check on the commit 66c4c36 with my Windows 8 system and got the attached log file. Do you have any glue why the x in line 55 converted?

All best,
Dirk

Log-File:

* using log directory 'C:/Users/Dirk/Documents/Repositories/smoof/smoof.Rcheck'
* using R version 3.1.3 (2015-03-09)
* using platform: i386-w64-mingw32 (32-bit)
* using session charset: ISO8859-1
* checking for file 'smoof/DESCRIPTION' ... OK
* checking extension type ... Package
* this is package 'smoof' version '1.0'
* checking package namespace information ... OK
* checking package dependencies ... OK
* checking if this is a source package ... OK
* checking if there is a namespace ... OK
* checking for executable files ... OK
* checking for hidden files and directories ... OK
* checking for portable file names ... OK
* checking whether package 'smoof' can be installed ... [13s] OK
* checking installed package size ... OK
* checking package directory ... OK
* checking DESCRIPTION meta-information ... OK
* checking top-level files ... OK
* checking for left-over files ... OK
* checking index information ... OK
* checking package subdirectories ... OK
* checking R files for non-ASCII characters ... OK
* checking R files for syntax errors ... OK
* loading checks for arch 'i386'
** checking whether the package can be loaded ... OK
** checking whether the package can be loaded with stated dependencies ... OK
** checking whether the package can be unloaded cleanly ... OK
** checking whether the namespace can be loaded with stated dependencies ... OK
** checking whether the namespace can be unloaded cleanly ... OK
** checking loading without being on the library search path ... OK
* loading checks for arch 'x64'
** checking whether the package can be loaded ... OK
** checking whether the package can be loaded with stated dependencies ... OK
** checking whether the package can be unloaded cleanly ... OK
** checking whether the namespace can be loaded with stated dependencies ... OK
** checking whether the namespace can be unloaded cleanly ... OK
** checking loading without being on the library search path ... OK
* checking dependencies in R code ... WARNING
'::' or ':::' import not declared from: 'rPython'
* checking S3 generic/method consistency ... OK
* checking replacement functions ... OK
* checking foreign function calls ... OK
* checking R code for possible problems ... OK
* checking Rd files ... WARNING
plot3D.Rd: non-ASCII input and no declared encoding
problem found in 'plot3D.Rd'
* checking Rd metadata ... OK
* checking Rd line widths ... OK
* checking Rd cross-references ... OK
* checking for missing documentation entries ... OK
* checking for code/documentation mismatches ... OK
* checking Rd \usage sections ... WARNING
Undocumented arguments in documentation object 'plot3D'
  'x'
Documented arguments not in \usage in documentation object 'plot3D':
  'xÂ'

Functions with \usage entries need to have the appropriate \alias
entries, and all their arguments documented.
The \usage entries must correspond to syntactically valid R code.
See chapter 'Writing R documentation files' in the 'Writing R
Extensions' manual.
* checking Rd contents ... OK
* checking for unstated dependencies in examples ... OK
* checking line endings in C/C++/Fortran sources/headers ... OK
* checking compiled code ... OK
* checking examples ...
** running examples for arch 'i386' ... ERROR
Running examples in 'smoof-Ex.R' failed
The error most likely occurred in:

> base::assign(".ptime", proc.time(), pos = "CheckExEnv")
> ### Name: makeFunnelFunction
> ### Title: Generator for funnel-like functions.
> ### Aliases: makeFunnelFunction
> 
> ### ** Examples
> 
> # Unfortunately
> fn = makeFunnelFunction(n.peaks = 10L, dimension = 2L, topology = "funnel", seed = 123)
Error: could not find function "makeFunnelFunction"
Execution halted
** running examples for arch 'x64' ... ERROR
Running examples in 'smoof-Ex.R' failed
The error most likely occurred in:

> base::assign(".ptime", proc.time(), pos = "CheckExEnv")
> ### Name: makeFunnelFunction
> ### Title: Generator for funnel-like functions.
> ### Aliases: makeFunnelFunction
> 
> ### ** Examples
> 
> # Unfortunately
> fn = makeFunnelFunction(n.peaks = 10L, dimension = 2L, topology = "funnel", seed = 123)
Error: could not find function "makeFunnelFunction"
Execution halted

We now not only offer the means to work with objecitve functions, but also offer a lot of ploting functions and a load of optimization test functions.

How do I install rPython with R 3.1.3 on Windows OS?
An install.packages("rPython") from the austrian repo throws me the warning: "package rPython not available". The installation guide from rForge gives the same one.

Benchmarking functions for optimization algorithms can be categorized in unimodal/multimodal, convex/nonconvex. Introduce a tagging system, i.e., the possibility to assign different tags to functions.

UF1, ..., UF10

There are two stupid warnings. No time for this at the moment.

It would be great, if we could build up noisy functions out of a deterministic component and a noisy component. But we have to be aware of the different possible constellations:

• we know the true function and the noisy component
• we only know the noisy function
• our function is deterministic

We need to improve code coverage.

For some function (e. g. the BBOB functions, since they are implemented in C), it might be beneficial to pass a matrix and enable the smoof function to be vectorized. We thus should introduce an logical option vectorized to makeObjeciveFunction.
Furthermore there should be a possibility to deactivate the simple signature.

Plot only the feasible region. Until now only box-constraints are considered.

At least for 2D objective functions with one discrete/logical and one numeric parameter via facets. Possible would also be a 3d plots for two discrete/logical parameters and one numeric one.

Similar to the already implemented countingWrapper. Should offer the possibilty to log at least y values and optionally x-values. Keep in mind: we want arbitrary functions, thus we cannot generally store the x values in a matrix but in a list of lists, since we want support for mixed functions as well.

Provide documentation for the package as a whole.

Hmm, the result may be a data.frame/matrix of non-dominated points or a function?

Add a function to quickly visualize the Pareto-optimal front by building a grid in the equidistant 2-dimensional grid, computing the objective value and filter the non-dominated points.

We want contour line plots, no image maps, because they look really ugly with ggplot2. Maybe add a logical argument imagemap with default value FALSE?

Check all the (at the moment 68) single-objective test functions thoroughly. Sometimes, there exist inconsistent definitions in particular of the box constraints in the literature.
We should add at least one reference to each function and the formula which was used should be printed to the documentation.

Mention implemented functions. Maybe best is to write a R function, which extracts all generators, prints the name in a markdown list and replaces a placeholder in the README.md file.

Reimplement all the functions in C.

Suggestion:
add a makeFunction(fun.name, dimension, ...) Function

Why?
I would like to do something like

lapply(as.list(filterFunctionsByTags(tags = c("multimodal", "continuous", "scalable"))),
  function(x) makeFunction(fun.name = x))

Currently I have to first lookup the function names with filterFunctionsByTags(tags = c("multimodal", "continuous", "scalable")) and then search for the respective make***Function to create the function. Or is there another possibility to use the resuls of filterFunctionsByTags to set multiple functions?

All functions need type and tags (see #30)

Until now we force the user to pass a named list. Maybe better allow (numeric) unnamed vectors as well?

Hi,

getTags(makeEggholderFunction())
# [1] "continuous"     "differentiable" "non-separable"  "scalable"       "multimodal"    

says that the Function is scalable, but I can't set the number of dimensions.

According to https://sites.google.com/site/gotestfunctions/multimodal-function-list/-23-giunta and https://sites.google.com/site/gotestfunctions/multimodal-function-list/-23-giunta the global optimum for the giunta function is at 0.06447... but

getGlobalOptimum(makeGiuntaFunction())
# $param
#         x1        x2
#1 0.4583428 0.4583428
#
# $value
# [1] 0.06463

Single- and multi-objective target functions have some things in common: name, fn, noisy. To avoid copy and paste there should be a non-exported internal makeBasicObjectiveFunction function.

Add functions isScalable, isMultimodal, ... and in general hasTags.
They should work on both smoof_generator and smoof_function objects.

At the moment we need to instantiate the functions to access their tags. We should do the following instead:

1. add tags to the generator object
2. access the generators tags within the generator to set the tags of the function
3. simplify filterFunctionsByTags

Sometimes we may want out test set to consists only multimodal functions. Add a possibility to filter functions by tag.

We should check for the implemented single-objective optimization test functions, whether the function value at the global.opt.params value(s) corresponds to the global.opt.value. This is another way to check the implementations.

Hallo Jakob,

The Rastrigin function can be defined with e.g. two dimensions: makeRastriginFunction(2L), which give a parameter set with two numerics. However, the function take a vector of length two, not two numerics. Would it be consistent to define a parameter vector or multiple arguments in the function?

All best
Dirk