In column-wise modeling, newly constructed variables are inserted into existing modeling objects. The term column-wise comes from linear programming, where the constraints are typically represented as a matrix. Adding new variables to the optimization problem corresponds to adding columns to the constraint matrix.

This is the procedure for column-wise modeling:

Start with the set of modeling objects of these classes

• IObjective,
• IRange, and
• ILPMatrix

For each of these objects call IMPModeler.ColumnArray, with the object as an argument, along with the other arguments needed to install an array of new variables in the existing modeling objects. See the documentation of IMPModeler.ColumnArray for details of these parameters.

The column-array objects can then be linked to an aggregate column-array object with the method ColumnArray.And, which will include all the individual objects. If the new variables are to be installed in only one modeling object, there is no need to use the method and.

The column-array object constructed this way is now ready to be used to create an array of new variables and install them in the existing modeling objects, as defined by the column-array object. This is done by passing the column-array object as an argument to the constructor methods for arrays of variables, for example IMPModeler.NumVarArray or IMPModeler.IntVarArray. The newly created variables will immediately be part of the existing modeling objects that have been used for constructing the column-array object.