Povratne informacije Uredi

Apply solvers to solve optimization problems

  • Članak
  • 2 min. za čitanje

Once you have expressed an optimization Problem, you're ready to solve it by applying a solver.

Configure and run the solver

In this example, you'll use a parameter-free version of parallel tempering. For more information, see Optimization providers on Azure Quantum.

solver = ParallelTempering(workspace, timeout=100)

For arguments, the solver takes a previously created Workspace object and a timeout value, which is the maximum amount of time (in seconds) to run the solver. Detailed documentation on parameters is available in the reference for each solver.

Note

For more information on creating and managing workspaces, see Manage quantum workspaces.

Solvers provide an optimize method that expects a Problem object. The optimize method uploads the problem definition, submits a job to solve the problem, and polls the status until the job has completed running. Once the job has completed, it returns a JobOutput object which contains the results. For more information, see Understand Solver results.

result = solver.optimize(problem)
print(result)

This method will submit the problem to Azure Quantum for optimization and synchronously wait for it to be solved. You'll see output like the following in your terminal:

> {'solutions':[{'configuration': {'0': 1, '1': 1, '2': -1, '3': 1}, 'cost': -32.0}]}

See Solve long running problems for solving problems asynchronously.

Using CPU vs FPGA solvers

By default, the CPU version of a solver is used. In order to specify a different version, such as FPGA, specify a platform parameter:

solver =  SimulatedAnnealing(workspace, timeout=100, platform=HardwarePlatform.FPGA)

For more information about FPGA, see FPGA vs. CPU

Returning multiple solutions

Both FPGA and CPU solvers in the Microsoft QIO provider support the option to return more than one solution during a single run. Use set_number_of_solutions to set this option.

# return a maximum of 5 solutions
solver.set_number_of_solutions(5)

This option has the following behavior and requirements:

  • Value must be between 1 and 10 (default = 1).
  • Solvers are not guaranteed to return the exact number of solutions specified by the parameter, as solvers may not find that many solutions.
  • Solvers are guaranteed to return the best solution they find in index 0. They will return the rest of the solutions (if any) in the same list and in sorted order.
  • A build of the azure-quantum Python package with version 0.17.2106 or higher is required to use the feature. For more information, see Update the QDK

Next steps