I am writing a column generation algorithm and I am using almost the same structure as cutstock.mod example. However, the problem is this structure even though it seems to be a standard procedure, but its time performance does not seems very good and it takes too much time.
I am wondering if it is possible to define the master and sub problem in the same model and we can just in a loop we hide and un-hide the constraints of the master and sub problem and tell which objective function to maximize?
And we do not use the flow control structure like the cutting stock example. This is a standard structure however, in the large scale models, each time using "new" model and .. and connecting two models when you have too many iterations is not efficient.
Thank you so much in advance
Pinned topic Improvement in the column generation algorithm?
Answered question This question has been answered.
Unanswered question This question has not been answered yet.
Updated on 2013-03-27T08:28:50Z at 2013-03-27T08:28:50Z by SystemAdmin
SystemAdmin 110000D4XK1883 Posts
Re: Improvement in the column generation algorithm?2013-03-27T08:28:50ZThis is the accepted answer. This is the accepted answer.A few months ago there was a related discussion
take a look, may be the proposed there approaches will suffice you