الملاحظات تحرير

Azure Quantum workspace

  • مقالة
  • قراءة خلال 4 دقائق
from azure.quantum import Workspace

Constructor

To create a Workspace object, you must supply the following arguments in order to connect. If you have not already created a workspace, follow the steps in Creating an Azure Quantum workspace guide using the following values:

  • subscription_id: The subscription ID where the workspace is deployed.
  • resource_group: The name of the resource group where the workspace is deployed.
  • name: The name of the workspace.
  • location: The location where the workspace is deployed, for example West US, with either the create or show commands.
  • credential: (Optional) The credential to use to connect to the Azure Quantum and Storage services. Normally one of the credential types from Azure.Identity. Defaults to DefaultAzureCredential, which will attempt multiple forms of authentication.

You can find these values by viewing your Azure Quantum workspace details through the Azure portal.

In case you have not specified any credentials, the first time you run a method which interacts with the Azure service, a window might prompt in your default browser asking for your credentials.

workspace = Workspace (
    subscription_id = "",  # Add your subscription_id
    resource_group = "",   # Add your resource_group
    name = "",             # Add your workspace name
    location= ""        # Add the workspace location, for example, westus
)

Workspace.get_job

Retrieves information about a job.

from azure.quantum import Workspace

workspace = Workspace(...)
job = workspace.get_job("285cfcb4-6822-11ea-a05f-2a16a847b8a3")
print(job.details.status)

Succeeded

Workspace.list_jobs

Returns the list of existing jobs in the workspace.

from azure.quantum import Workspace

workspace = Workspace(...)
jobs = workspace.list_jobs()
for job in jobs:
    print(job.id, job.details.status)

08ea8792-68f2-11ea-acc5-2a16a847b8a3 Succeeded
0ab1863a-68f2-11ea-82b3-2a16a847b8a3 Succeeded
0c5c507e-68f2-11ea-ba75-2a16a847b8a3 Cancelled
f0c8de58-68f1-11ea-a565-2a16a847b8a3 Executing

The Workspace.list_jobs method also allows the user to filter on the creation date, status and name properties of a job when listing. Filters can be combined.

Note

To use the filter feature, you must have version 0.18.2107 or newer of the azure-quantum Python package.

Filtering by creation time on list_jobs

To filter by jobs created after a certain time, set the created_after filter. This parameter accepts a datetime object.

  • A datetime without timezone information is assumed to be in the local time of the user.
from azure.quantum import Workspace
from datetime import datetime, timedelta, timezone

# filter jobs created within the last day (in local time).
creation_time = datetime.now() - timedelta(days=1)

workspace = Workspace(...)
jobs = workspace.list_jobs(created_after=creation_time)

Filtering by job name on list_jobs

To filter by jobs by a certain name, set the name_match filter. This parameter accepts a regex string.

from azure.quantum import Workspace

workspace = Workspace(...)

# filter job name on any regex expression
jobs1 = workspace.list_jobs(name_match="job_.+")
jobs2 = workspace.list_jobs(name_match=".*test.*")
jobs3 = workspace.list_jobs(name_match="regular_string_job_name")

Filtering by job status on list_jobs

To filter by jobs by status, set the status filter. This parameter accepts a JobStatus enumeration.

from azure.quantum import Workspace
from azure.quantum import JobStatus

workspace = Workspace(...)

# list succeeded jobs
jobs1 = workspace.list_jobs(status=JobStatus.SUCCEEDED)

# list pending jobs
jobs2 = workspace.list_jobs(status=JobStatus.EXECUTING)

# list failed jobs
jobs3 = workspace.list_jobs(status=JobStatus.FAILED)

Workspace.cancel_job

Cancels a job that was previously submitted.

from azure.quantum import Workspace

workspace = Workspace(...)
job = workspace.get_job("285cfcb4-6822-11ea-a05f-2a16a847b8a3")

workspace.cancel_job(job)
print(job.details.status)

Succeeded

Workspace.get_targets

Lists instances of all targets available on the Workspace and can be filtered by provider ID or target name. This method returns a list or a single instance. This includes all of the QIO Solvers, since Solver is a subclass of Target. Each target can be used to submit a job as an alternative to creating a Solver instance directly as described in Job management. Workspace.get_targets takes optional keyword parameters. If no keyword parameters are passed, it defaults to using the parameter-free target. Note that there is an exception for the Population Annealing and Substochastic Monte Carlo solvers, as they requrie a timeout parameter to resolve to the parameter-free version. The QMC solver does not have a parameter-free option.

The following example shows how to get all targets associated with your workspace:

from azure.quantum import Workspace

workspace = Workspace(...)
targets = workspace.get_targets()
targets

[<Target name="microsoft.paralleltempering-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.simulatedannealing-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.tabu-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.qmc.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.populationannealing.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.substochasticmontecarlo.cpu", avg. queue time=0 s, Available>,
 <Target name="ionq.qpu", avg. queue time=669 s, Available>,
 <Target name="ionq.simulator", avg. queue time=1 s, Available>,
 <Target name="1qbit.tabu", avg. queue time=0 s, Available>,
 <Target name="1qbit.pathrelinking", avg. queue time=0 s, Available>,
 <Target name="1qbit.pticm", avg. queue time=0 s, Available>,
 <Target name="toshiba.sbm.ising", avg. queue time=5 s, Available>,
 <Target name="quantinuum.hqs-lt-s1", avg. queue time=0 s, Unavailable>,
 <Target name="quantinuum.hqs-lt-s1-apival", avg. queue time=1 s, Available>,
 <Target name="quantinuum.hqs-lt-s1-sim", avg. queue time=6 s, Available>]

To filter by provider, specify the provider_id input argument:

targets = workspace.get_targets(provider_id="microsoft")
targets

[<Target name="microsoft.paralleltempering-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.simulatedannealing-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.tabu-parameterfree.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.qmc.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.populationannealing.cpu", avg. queue time=0 s, Available>,
 <Target name="microsoft.substochasticmontecarlo.cpu", avg. queue time=0 s, Available>]

To get a single target, for instance, the Simulated Annealing solver, specify the name input argument:

solver = workspace.get_targets(name="microsoft.simulatedannealing.cpu")
solver

Since no keyword arguments were given, the workspace defaults to the parameter-free version:

<Target name="microsoft.simulatedannealing-parameterfree.cpu", avg. queue time=0 s, Available>

To specify input arguments, use:

solver = workspace.get_targets(name="microsoft.simulatedannealing.cpu", timeout=100, seed=22)
solver

<Target name="microsoft.simulatedannealing.cpu", avg. queue time=0 s, Available>

This target can then be used to submit a problem and get the resulting job:

from azure.quantum.optimization import Problem, ProblemType

problem = Problem(name="MyOptimizationJob", problem_type=ProblemType.ising)
problem.add_term(c=-9, indices=[0])
problem.add_term(c=-3, indices=[1,0])
problem.add_term(c=5, indices=[2,0])
problem = 
job = solver.submit(problem)
results = job.get_results()
results

{'solutions': [{'configuration': {'0': 1, '1': 1, '2': -1}, 'cost': -17.0}]}