The following archive comprise the HCIAC source code along with two example of simple sampling methods and a benchmark of test functions.
The code is written in C++, and has been tested on Linux (Fedora Core 1) with gcc 3.3.2 and on Windows (using Cygwin), with gcc 3.3.1-3.
Download and uncompress this archive, it will create a directory named metah_dreo-al_2004. To test if the archive has been proprely download, check if its md5 sum is equal to the following one :
9c4cec90623af0eadb257b00ad2b7c76
To compile it, just type make hciac or make within the cop/ directory. The binaries are generated in the bin/ directory. Use cop-hciac for the HCIAC metaheuristic.
For a complete help, type cop-hciac --help.
The basic usage is the following : $ ./cop-hciac «problem» «search space» «parameters»
Default output : [optimum vector ...] [optimum value] [evaluation number]
The optimum vector is composed of one or several variables, depending on the problem optimized.
Minimizing the two dimensional B2 function, in [-10,10] :
$ ./cop-hciac B2 -10,-10 10,10
Minimizing the Griewangk problem, with 10 variables, in [-512,512] :
$ ./cop-hciac Griewangk 10:-512 10:512
Minimizing the Shekel problem, with 10 ants and 20 iterations of the Nelder-Mead search :
$ ./cop-hciac Shekel 4:0 4:10 -a 10 -ie 20
Minimizing the Rosenbrock (3) function, using not more than 5000 evaluations, and a stopping criterion sensibility of 1000.
$ ./cop-hciac Rosenbrock 3:-5 3:10 -e 5000 -b 1000
Optimizing a problem using a external command. Here a two dimensional linear function.
$ ./cop-hciac "echo ? ?" 2:-10 2:10