First, learn about Dimensions and Hierarchies in the Optimization Engine at Main Concepts for Optimization Problems.

Hierarchies

The problem description starts with enumerating the hierarchies of dimensions used in the problem. Dimensions are sorted from the finest grain to the larger grain.

For instance, in the following example:

• Each customer belongs to a customer_group and an unrelated customer_country.

• Each product belongs to a product_family which itself belongs to a product_group.

• shipping_mode is a dimension without any hierarchical relation to any other dimension.

hierarchies:
- [customer, customer_group]
- [customer, customer_country]
- [product, product_family, product_group]
- [shipping_mode]

All the dimensions, in their correct hierarchies, are to be defined. They are used in the spaces definitions.

Note: Maybe, from the end-user point of view, the customer group is hierarchical with the country: “wholesales/UK” is not the same group of customers as “wholesales/CZ” for instance. But if the customer group has the same name in both countries, like “wholesales”, it is not a hierarchy, it is a cartesian product. A hierarchy would be with the values “UK” and “CZ” for the countries and “wholesales-UK”, “wholesales-CZ”, “retail-CZ“ for the customer groups.