MIPLIB_Problems.tar.gz and Hoffman_Problems.tar.gz contain all the MIPLIB and Hoffman problems used in the experiments, in a format readable by MATLAB.  These are meant to be unpackaged and linked to the same directory in which Run_Experiments.m, described below, resides.  Run_Experiments.m can then read these files and use them in the various algorithms.

******************************************

Run_Experiment.m

Run_Experiment()

This is the main function from which all the experiments can be run, one at a time.  It also specifies the run time limit for all of the algorithms in the global variable max_time.  To change how long the algorithms can run, just alter this global in the function.

Run by entering Run_Experiment at the MATLAB command line.  First you need to choose which algorithm you want to run: LFS, Chubanov, or Motzkin.  See <algorithm>_README.txt for the details of each algorithm.  If you choose Motzkin, you need to specify the under-/over-shooting parameter lambda and the error parameter.

Next you need to choose one of the experiment groups: Telgen, Hoffman, or MIPLIB.  See below for the explanations of each.

Finally, you need to choose the specific experiment you want to run.  Your choice will be limited by the experiment group you chose in the preceding step.

The ouput will tell you how long the particular experiment took and how many iterations/recursions the algorithm went through.  If you chose to run the LFS algorithm on a MIPLIB problem, it will also tell you the total time the algorithm took on all instances of \Delta_A as well as the time it took for the final run of \Delta_A.  See the paper for more details.

******IMPORTANT******:	Prior to running this code, make sure you have downloaded the appropriate algorithm code.  The code is not provided in this tar.gz file, only the problem sets are included.  If you want all the code at once, download All_Project_Code.tar.gz, and then make sure all the directories are linked in MATLAB prior to running the code.

******************************************

Telgen Experiments

There are no problem files for these experiments.  Instead Run_Experiment.m builds the specified version.  See the paper for a full explanation of the problem set.

If you do choose this set, you need to specify a positive integer a, which determines how acute the angle is between the two constraints.  Briefly, the problem instance is in 2 dimensions, with the constraints y >= 0 and x - (2^a)y >= 2^a.

******************************************

Hoffman Experiments

There are 7 problem files for these experiments.  Hoffman.mat contains the original problem described both in our paper and in Hoffman et al.'s paper (see our references).  Each of the files hoffman<n>.mat files contains the maximum subdeterminant of the particular experimant instance.  For example, hoffman5.mat contains the maximum determinant of the 5-dim Hoffman experiment.  This is so that the LFS algorithm does not need to recalculate the maximum subdeterminant every time these experiments are run.

******************************************

MIPLIB Experiments

The rest of the .mat files in this tar.gz file are the MATLAB readable MIPLIB problems used in the experiments.  A list of all the MIPLIB problems is contained in MIPLIB_List.txt, included in the tar.gz file.  This list is used by Run_Experiment.m in order to list the callable MIPLIB experiments.  Make sure this list is in an appropriate directory and the directory is specified in Run_Experiment.m.  There are specific lines in Run_Experiment.m you need to alter if this list or the experiments are not in the same directory.  The lines you need to alter are indicated with appropriate comments.  It is easiest though to just keep the list and all the problems in the same directory as Run_Experiment.m.