Decision Optimization

Hi,

have you had a look at the cutstock.py example ?

Column generation is a solution process that begins with a small, manageable part of a problem (specifically, a few of the variables), solves that part, analyzes the partial solution to determine the next part of the problem (specifically, one or more variables) to add to the model, and then solves the new, enlarged model. Column generation repeats the process until a satisfactory solution to the whole problem is achieved.

Hi,Sorry for the delayed response. 

    new_pattern_id = max(pt.id for pt in master_model.patterns) + 1
    new_pattern = TPattern(new_pattern_id, 1)
    master_model.patterns.append(new_pattern)
    for used, item in zip(item_usages, master_model.items):
        master_model.pattern_item_filled[new_pattern, item] = used

    # --- add one decision variable, linked to the new pattern.
    new_pattern_cut_var = master_model.continuous_var(lb=0, ub=master_model.MAX_CUT,
                                                      name='cut_{0}'.format(new_pattern_id))
    master_model.cut_vars[new_pattern] = new_pattern_cut_var

    # update constraints
    for item, ct in zip(master_model.items, master_model.item_fill_cts):
        # update fill constraint by changing lhs
        ctlhs = ct.lhs
        filled = master_model.pattern_item_filled[new_pattern, item]
        if filled:
            ctlhs += new_pattern_cut_var * filled

I think the above code snippet, is where a new variable is initialized corresponding to the new pattern. And then, all the corresponding constraints are retrieved. 
Finally, you add the variable to each constraint using : ctlhs += new_pattern_cut_var * filled

I think the above is a good way. I will try using this and inform of the performance improvement. 

Thanks!

Column generation is a solution process that begins with a small, manageable part of a problem (specifically, a few of the variables), solves that part, analyzes the partial solution to determine the next part of the problem (specifically, one or more variables) to add to the model, and then solves the new, enlarged model. Column generation repeats the process until a satisfactory solution to the whole problem is achieved.