Optimization using a multi-dimensional data model
View Patent ↗In one embodiment, a method for optimization using multi-dimensional data includes receiving a problem instance to be solved using an optimization engine. The problem instance is specified in a multi-dimensional format associated with a multi-dimensional data model that includes multiple data dimensions each including a hierarchy of members. The problem instance includes an objective function, an optimization operator, and one or more constraints. The optimization engine is unable to solve the problem instance including the objective function, the optimization operator, and the one or more constraints in the multi-dimensional format. The method also includes, using a transformation module, transforming the problem instance including the objective function, the optimization operator, and the one or more constraints into a format appropriate for the optimization engine and communicating the transformed problem instance including the objective function, the optimization operator, and the one or more constraints to the optimization engine to be solved.
1. A system for optimization using multi-dimensional data, comprising:
a server operable to:
using a multidimensional data model, organize data stored at one or more data storage locations, the multi-dimensional data model including a plurality of data dimensions each including a hierarchy of members;
receive input from a user specifying a problem instance to be solved using an optimization engine, the problem instance specified by the user in a multidimensional format and comprising an objective function, an optimization operator, and one or constraints, the optimization engine being unable to solve the problem instance comprising the objective function, the optimization operator, and the one or more constraints in the multi-dimensional format; and
communicate the problem instance comprising the objective function, the optimization operator, and the one or more constraints in the multi-dimensional format to a transformation module, the transformation module is operable to:
receive the problem instance comprising the objective function, the optimization operator, and the one or more constraints in the multi-dimensional format;
transform the problem instance comprising the objective function, the optimization operator, and one or more constraints into a format appropriate for the optimization engine;
communicate the transformed problem instance comprising the objective function, the optimization operator, and one or more constraints to the optimization engine to be solved;
receive a solution associated with the problem instance from the optimization engine;
transform the solution into the multi-dimensional format; and
communicate the transformed solution to the server, wherein the server displays the transformed solution to the user on a display.
2. The system of claim 1 , further comprising a business repository operable to store the multi-dimensional data model, the server further operable to communicate with the business repository to access data specified using the multidimensional format.
3. The system of claim 1 , wherein transforming the problem instance comprises:
parsing the received problem instance to identify pre-defined multidimensional syntax; and
translating the multi-dimensional syntax to a syntax appropriate for the optimization engine.
4. The system of claim 1 , wherein transforming the problem instance comprises generating multiple problem constraints in a format appropriate for the optimization engine from a single problem constraint included in the received problem instance, the single problem constraint identifying a member in each data dimension to which the constraint is applicable.
5. The system of claim 1 , wherein transforming the problem instance comprises importing data applicable to the problem instance from one or more data storage locations, the imported data being included in the transformed problem instance in a format appropriate for the optimization engine.
6. The system of claim 1 , wherein:
the objective function comprises a data measure or a combination of data measures to be optimized; and
the problem instance further comprises:
a problem domain that includes all data in the multi-dimensional data model that is located hierarchically below one or more specified intersections in the multi-dimensional data model, each intersection identified by specifying a member in each data dimension; and
an evaluation level specified by identifying a particular level in the hierarchy of each data dimension.
7. The system of claim 6 , wherein one or more data measures in the objective function have an associated data value in a data storage location for each of one or more intersections in the problem domain.
8. The system of claim 6 , wherein the objective function further comprises an aggregation domain for each data measure.
9. The system of claim 1 , wherein the server is further operable to replicate a single constraint in the multi-dimensional format into multiple constraints in the multi-dimensional format, the single constraint including one or more coverage sets identifying multiple members of one or more data dimensions to which the constraint applies.
10. A method for optimization using multi-dimensional data, comprising:
receiving a problem instance to be solved using an optimization engine, the problem instance specified in a multi-dimensional format associated with a multidimensional data model that includes plurality of data dimensions each including a hierarchy of members, the problem instance comprising an objective function, an optimization operator, and one or more constraints, the optimization engine being unable to solve the problem instance comprising the objective function, the optimization operator, and the one or more constraints in the multi-dimensional format;
using a transformation module, transforming the problem instance comprising the objective function, the optimization operator, and the one or more constraints into a format appropriate for the optimization engine;
communicating the transformed problem instance comprising the objective function, the optimization operator, and the one or more constraints to the optimization engine to be solved;
receiving a solution associated with the problem instance from the optimization engine;
using the transformation module, transforming the solution into the multidimensional format; and
communication the transformed solution to a server.
11. The method of claim 10 , wherein transforming the problem instance comprises:
parsing the received problem instance to identify pre-defined multidimensional syntax; and
translating the multi-dimensional syntax to a syntax appropriate for the optimization engine.
12. The method of claim 10 , wherein transforming the problem instance comprises generating multiple problem constraints in a format appropriate for the optimization engine from a single problem constraint included in the specified problem instance, the single problem constraint identifying a member in each data dimension to which the constraint is applicable.
13. The method of claim 10 , wherein transforming the problem instance comprises importing data applicable to the problem instance from one or more data storage locations, the imported data being included in the transformed problem instance in a format appropriate for the optimization engine.
14. The method of claim 10 , wherein:
the objective function comprises a data measure or a combination of data measures to be optimized; and
the problem instance further comprises:
a problem domain that includes all data in the multi-dimensional data model that is located hierarchically below one or more specified intersections in the multi-dimensional data model, each intersection identified by specifying a member in each data dimension; and
an evaluation level specified by identifying a particular level in the hierarchy of each data dimension.
15. The method of claim 14 , wherein one or more data measures in the objective function have an associated data value in a data storage location for each of one or more intersections in the problem domain.
16. The method of claim 14 , wherein the objective function further comprises an aggregation domain for each data measure.
17. The method of claim 10 , further comprising automatically replicating a single constraint in the multi-dimensional format into multiple constraints in the multi-dimensional format, the single constraint including one or more coverage sets identifying multiple members of one or more data dimensions to which the constraint applies.