New Hybrid Solver: Constrained Quadratic Model

Release date: 2021-10-05

This release introduces a constrained quadratic model (CQM) solver for problems with binary and integer variables and one or more constraints. In contrast to previous hybrid solvers, which required that you represent any problem constraints as penalty models in your objective, the CQM solver natively
supports equality and inequality constraints. To enable formulating such quadratic models with convenient notation, the Ocean™ SDK now supports symbolic math.

The first online CQM solver is hybrid_constrained_quadratic_model_v1. It accepts problems with up to 5,000 variables, each of which can represent binary or integer values, 750 million total linear plus quadratic biases (values assigned to both nodes and edges of the graph representing your problem), and 100,000 constraints.

Submit problems to the hybrid CQM solver as you would submit any BQM-formulated problem; in Ocean software use dwave-system tool’s LeapHybridCQMSampler. Example CQM problems are available in the documentation (see examples on bin packing and stock selling, for instance) and in our collection of code examples on GitHub.

For details on hybrid solvers, see Using Leap’s Hybrid Solvers. For performance information on the new hybrid CQM solver, see our white paper.

Note: Contact D-Wave at sales@dwavesys.com if your application requires scale or performance that exceeds the currently advertised capabilities of Leap’s generally available hybrid solvers.