Overview

This chapter introduces some more advanced features of the modeling language in Mosel. We shall not attempt to cover all its features or give the detailed specification of their formats. These are covered in greater depth in the Mosel Reference Manual.

Almost all large scale LP and MIP problems have a property known as sparsity, that is, each variable appears with a non-zero coefficient in a very small fraction of the total set of constraints. Often this property is reflected in the data tables used in the model in that many values of the tables are zero. When this happens, it is more convenient to provide just the non-zero values of the data table rather than listing all the values, the majority of which are zero. This is also the easiest way to input data into data tables with more than two dimensions. An added advantage is that less memory is used by Mosel.

The main areas covered in this chapter are related to this property:

• dynamic arrays
• sparse data
• conditional generation
• displaying data

We start again with an example problem. The following sections deal with the different topics in more detail.