See run_varmod.py for details on how to run simulations.

The file artificial.pickle contains the data used in the artificial data
experiments reported in the ICA 2003 paper [1]. The MEG data used in the 
same paper can be obtained from the following webpage:

  http://www.cis.hut.fi/projects/ica/eegmeg/MEG_data.html

The part used in the experiments is a 2500 samples long sequence starting
at time index 8500 and ending at 11000.


Some examples on using the above mentioned datasets  

% python run_varmod.py --sdim=20 --rdim=2 --iters=5000 \
  --prune=20005 --add=20005 srcvar artificial

% python run_varmod.py --prune=500 --varlayer=20 --iters=2000 \
  --sblock=blocks.DynIca --sdim=50 --rdim=5 \
  --prunefreq=100 --add=550 --addfreq=100 srcvar meg


[1] H. Valpola, M. Harva, J. Karhunen, (2003). Hierarchical Models of
    Variance Sources. In Proc. 4th Int. Symp. on Independent Component
    Analysis and Blind Signal Separation (ICA2003), Nara, Japan, pp. 83-88.
    Available online at 
    http://www.cis.hut.fi/projects/bayes/papers/files/Valpola03ICA_Variance.pdf
