To run the experiments, run ./create_probs.sh to create the problems (it uses random_cone_problem_fixed_part.py to create the problems from the command line). Then, call ./solve_probs.sh which uses test_fixed_part.py to load the data files and solve them.

===============================================================================

A little more cleaned up now. Try running the examples in dist_ecos subfolder:

python simple_example.py
python general_example.py
python compare_simple_general.py

The first script demos simple consensus, the second general consensus, and the third compares the convergence of the two methods.

Its designed to be run either serially or in parallel, to make it easier to debug.

Dependencies

You'll need:

• QCML
• ECOS
• NUMPY
• SCIPY
• NetworkX
• PyMetis

PyMetis is a little tricky to install. On Mac OSX:

brew install --with-icu --build-from-source boost
brew install metis

Then, download pymetis and:

cd pymetis && sudo python setup.py install

You may also have to modify line 525 in aksetup_helper.py

"boost_%s" -> "boost_%s-mt"