LpX

LpX is an open-source domain-specific language and runtime designed for modeling and solving linear programming problems. It provides a high-level syntax that mirrors mathematical notation and a flexible back end that translates models into formats suitable for established solvers. LpX can connect to a range of back-end engines, including GLPK, CBC, CPLEX, and Gurobi, enabling users to choose a solver based on licensing, performance, or problem size. It supports continuous variables, linear constraints, and objective functions, and can handle parameterized models for scenario analysis.

Key features include declarative model syntax, automatic variable typing, bounds, and linear constraints, and an objective definition. It includes facilities for parametric optimization, multiple objectives, sensitivity analysis, and scenario management. Users can write models concisely and leverage incremental solving where supported by the solver. LpX emphasizes interoperability, offering data bindings to CSV, JSON, and SQL sources, and providing import/export of models to standard formats such as MPS or LP.

Implementation and architecture: The compiler front end parses the LpX language into an intermediate representation, which the runtime uses to issue solver calls. A solver interface layer abstracts communication with back-end engines. The project is maintained by a community of contributors and released under an open-source license.

Usage and reception: LpX has been adopted in academic teaching, operations research practice, and small to mid-sized optimization projects. It is commonly compared with AMPL, Pyomo, and JuMP, with attention to its readable syntax and solver-agnostic approach. Limitations include maturity and performance on very large problems, which depend on the underlying solver.

History: The project began in the late 2010s as an education-focused modeling tool and evolved into a general-purpose LP framework with an emphasis on accessibility and interoperability.