The code in this directory was used for the experiments in the paper ''Decomposition in conic optimization with partially separable structure,'' by Y. Sun, M. S. Andersen, and L. Vandenberghe (available at www.seas.ucla.edu/~vandenbe/publications). The following software packages are required. - CVXOPT (http://cvxopt.org/download/index.html) - matplotlib (http://matplotlib.org/downloads.html) - psutils (http://code.google.com/p/psutil/) - CVX (http://cvxr.com/cvx/download/) The Python code has been tested for Python 2.6 and 2.7. =========================================================================== LIST OF CODE 1) PAPER EXAMPLES - example_blockarrow.py : primary script for running a small block-arrow example - blockarrow_problem.py : auxillary functions for constructing a block-arrow example - example_snl.py : primary script for running a sensor network localization problem - snl_problem.py : auxillary function for constructing a sensor network localization problem 2) FAST PROX TIMING EXPERIMENT In order to assure that the runtime comparisons between the interior-point methods and the fast prox methods, it is necessary to first generate the problems (via running generate_single_prox.py) before running the timing comparison scripts (prox_speed.m and prox_speed.py) - prox_speed.m : runs timing experiments for CVX and SDPT3 formulations in MATLAB - prox_speed.py : runs fast prox timing experiments in Python - solve_single_prox_via_cvx.m : auxillary function for formulating proximal operator in CVX - solve_single_prox_via_ipm.m : auxillary function for formulations of proximal operators via SDPT3 - generate_single_prox.py : generating random problems (run first) 3) AUXILLARY FUNCTIONS - spingarn.py : implementation of Spingarn's method - proxlib.py : implementation of fast proximal operators - convert.py : routines for constructing a converted problem given an unconverted problem and a sparsity pattern - symbolic.py : routines for constructing a clique tree from a sparsity pattern - misc.py : miscellaneous linear algebra routines - matfile.py : utility for reading/writing MATLAB files in Python