At the moment, CeFFC has been used only on linux machines with NVIDIA GPUs.
As stated in the manuscript, this tool uses Serbain Giuriou basic cuda kmeans, which we modified to implement kmeans++ and kmeans
with point retraction. In addition, we use the maximal clique enumeration algorithm proposed and implemented by   
David Eppstein, Maarten Löffler, and Darren Strash. Listing all maximal cliques in large sparse 
real-world graphs. ACM J. Exp. Alg., 18: 3–1, 2013.

1. Compile CeFFC:

nvcc -Xcompiler -fopenmp -arch=sm_35 -rdc=true *.cu -w -std=c++14 -O3 bin/*.o -o CeFFC 

2. To execute you need to have a tractography represented in files file.bundles and file.bundledata. You need to specify an 
output directory name and CeFFC will create it.  CeFFC uses number of clusters for central and extremes fiber points of 200 and 300
correspondly (see the manuscript). In addition, the fiber point positions considered are 0,3,10,17,20. dRmax and dMmax by default are set to 6.

The execution parameters is given as follows:

./CeFFC path_subject_bundles_file output_dir s_min s_max s_step dRmax dMmax

The output in output_dir

a) path_subjects_bundles_file specify the absolute or relative path with the input bundles file.
b) output_dir is the name of the directory you want to write the output.
c) s_min s_max s_step is to save final clusters into files of range of sizes given by the ranges [s_min,s_min+s_step] up to a range of 
[s_min+s_step, s_max]. These final clusters will be saved into the output_dir/bundlesPerSize directory.
d) dRmax is the maximum distance considered in the Step 3 of the algorithm.
e) dMmax is the maximum distance considered in the Step 4 of the algorithm.


In output_dir/bundleCentroids directory the centroids of the final clusters will be saved in a bundles file.
In output_dir/bundlesFinalClusters directory the final clusters will be saved in only one file. CeFFC also saves the 
stats.log file containing timing info, total number of clusters, total number of fibers in all clusters and coverage. In addition
it will contain the number of fibers in the clusters by size range based on s_min s_max s_step provided as parameters.

Example of execution:

./CeFFC ./subs/sub0.bundles output0 100 2000 100 6 6

Note:
Do no overwrite the output directory. Delete it first, if you want to use the same direactory as output or give another name for an
output directory.