Skip to main content

Supported Solvers

JijZept supports two main types of solvers:

  1. Default Solvers: These solvers are available for immediate use upon entering a JijZept contract.
  2. Third-party Solvers: These solvers require a separate contract in addition to the JijZept agreement.

JijZept Solvers

Solver/AlgorithmclassDescriptionPerformanceType
Simulated AnnealingJijSASamplerSimulated Annealing (SA) for QUBO and the Ising Model. This sampler is designed for verifying and testing models with small instances and is best suited for initial testing and exploration of your models.LowDefault Solver
Simulated Quantum AnnealingJijSQASamplerSimulated Quantum Annealing (SQA) is one of the quantum-inspired algorithms that simulates quantum annealing. It serves as a good intermediate step for simulating quantum annealing. However, due to the use of resources like replica copies for quantum simulation and the lack of parameter tuning in JijZept, it is not recommended for optimization purposes at now. This sampler is designed for verifying and testing models with small instances and is best suited for initial testing and exploration of your models.LowDefault Solver
Local SearchJijSolverLocal search solver that deals directly with constraints (without converting to QUBO). This solver tries to find a feasible solution first and after that seeks for better solutions. Compared to JijSASampler that runs simple SA for QUBO, JijSolver will aggressively adopt heuristics so that Jij's knowledge are put into together.LowDefault Solver
Digital AnnealerJijDA4SamplerDigital Annealer (DA) is a high-performance Ising machine provided by Fujitsu, which achieves rapid optimization calculations through highly parallelized algorithms. It also has the capability to handle constraint conditions, and when used with JijZept, the constraint conditions that DA can handle are automatically converted. This makes it a powerful solver capable of delivering performance even for problems with constraint conditions. For more detail please see link.HighThird-party Solver
D-Wave Leap Hybrid CQM SolverJijLeapHybridCQMSamplerThe LeapHybrid solver is a solver that excels at solving optimization problems with many variables, using a hybrid algorithm that combines quantum annealing machines and classical computers. Its ability to handle Constrained Quadratic Models (CQMs) enables it to address a wide range of problems, including those involving constrained quadratic optimization. For more detail please see link.HighThird-party Solver

Choosing a Solver

When selecting a solver within JijZept, it is important to consider the scope and requirements of your optimization problem. Jij offers JijSASampler, JijSQASampler, and JijSolver as Default Solvers, which are designed for verifying and testing models with small instances. These solvers are best suited for initial testing and exploration of your models.

Please be aware that JijSASampler, JijSQASampler, and JijSolver are intended for trial purposes and not for high-performance calculations on larger datasets. For problems involving larger datasets or requiring better performance, we recommend using Third-party Solvers available in JijZept.

JijZept's true value lies in its ability to automatically convert mathematical models and tune parameters, enabling users to fully harness the power of cutting-edge solvers. By selecting the most suitable high-performance solver for a specific problem, users can effectively leverage JijZept's capabilities. Moreover, the benefit of using JijZept is that it provides a consistent interface, regardless of which solver is chosen, making it convenient and user-friendly for various optimization problems.

With JijZept, you can maximize the potential of advanced solvers, ensuring efficient and accurate results for a wide range of applications.

Third-party Solver Assistance

If you would like to use a Third-party Solver, Jij can facilitate the contract process on your behalf. Simply contact Jij for assistance. If you have any questions or inquiries, please feel free to contact us using this form.