Exploitable results
The project focused on creating a new set of specific libraries, for combinatorial software and for algorithmic and optimization problems where efficiency is critical. Building on previous projects, the software framework developed is called ABACUS. It includes data structures and functions which are problem-independent and significantly reduces implementation times for commonly used algorithms. It also offers many different strategies for tuning a specific application.
ABACUS is the first framework that integrates column and constraint generation, has flexible data structures for column and constraint management, and supports combinatorial optimization problems. It has a simple hierarchy of software classes, providing an easy-to-use object-oriented interface, with classes specific to the problem derived from a small set of base classes. Data for the problem are input and stored in a class derived from the MASTER base class, where the user also sets up the constraint and variable system. Sub-problem classes derived from the SUB base class represent a node on the decision tree and a number of options are available in the SUB class, to tune the class to the problem. Depending on the application, classes may be derived from CONSTRAINT and VARIABLE base classes with application-specific information. Problem-specific functions are integrated by taking predefined virtual functions and customizing them. The ABACUS library is implemented in C++.
Buscando datos de OpenAIRE...
Se ha producido un error en la búsqueda de datos de OpenAIRE
No hay resultados disponibles