The MLPQ System

Abstract

MLPQ, short for Management of Linear Programming Queries, is a constraint-relational database system that is continuously built and updated at the University of Nebraska-Lincoln since 1995. The initial goal of the MLPQ system was to seamlessly extend traditional relational database systems. For example, MLPQ extends the minimum and maximum aggregation operators, which relational database systems apply only to a finite set of constant values, to operators that find the minimum or the maximum of a linear function over k attribute variables, which range over a k-dimensional polyhedral space represented by rational linear inequality constraints in the database. Of course, that is exactly linear programming, something that goes way beyond the normal capabilities of relational databases and was thought before to be a “separate issue.” This separation was unfortunate because raw operations research data needs to be often reformulated before it becomes suitable as input to a linear program solver. MLPQ allows such reformulations to be done by simple database queries.