You can get the source code used in our experiments, written in ANSI C, as a single zipped archive. The source code of ANF – Approximate Neighborhood Function – by Palmer et al. (see references, and copyright of the authors in their code), is available here. We provide only a patch file, automatically managed by our Makefile, to make the output consistent with the other methods. Extracting distdist.tar.gz, you create a folder called distdist. If you want to use the method ANF, save anf-1.0.tar.gz inside it, and use the make utility to do all the stuff (extract anf-1.0.tar.gz, patch it and compile the source), simply typing make at the shell prompt. Running the executable with the -h option, you can get a brief explanation on how to pass options and the datafile. Note that the option -g exports the results in a multicolumn formatted text file, where each experiment corresponds to a column, with elements equal to the approximate distance distribution values, at increasing distance starting from 0. The first line is an example of command to plot it with gnuplot. Redirecting the standard output to a file, it can be further processed. Indeed, with AnalysisDD.java you can perform also the error analysis. Use:
javac AnalysisDD.java
to obtain the .class file. For error evaluation, just follow the output of:
java AnalysisDD -h