
Dear Matthew,

We did in fact try an implementation in DIP.  As I recall, it was a bit slower than the AMPL implementation.

I will be happy to cooperate with you on this . Please find data and source code for the unit commitment model with switching in DIP.  It has been a while since I worked with this problem so I have probably forgotten something - please let me know.

I modified DecompAlgo.cpp in order to solve the master problem with CPLEX barrier.  The code is not pretty and i'm sure this could have been optimised.

Note that the model presented in the paper is a simplification of what is usually defined as the unit commitment problem in the litterature. See e.g. reference [6] for additional constraints (min. up/down time, ramping constraints, security constraints). These constraints are likely to affect the decomposition (positively or negatively) as they link different time periods.

A few notes on usage:
- "Arcs" in the DIP model represent either generators (having from node = 0) or transmission lines.
- the k-value can be adjusted in the .dat file.

You are welcome to experiment with this and please let me know if you have any questions.  I hope you would like to share any interesting results, that might improve our paper. If you do get publishable results, I would like to know.

Best regards,
Jonas Christoffer Villumsen
