Skip to content

Latest commit

 

History

History
86 lines (59 loc) · 2.92 KB

README.md

File metadata and controls

86 lines (59 loc) · 2.92 KB
Raw

Build Status

dimod-mock

dimod-mock is a package for mocking dimod-based samplers, including the structured ones. It lets you easily test your routines before trying them out on sophisticated samplers or real hardware.

Installation

To install run

pip install dimodmock

Usage

Mocking simple samplers

Suppose you created some instance of your sampler (derived from dimod.Sampler). This is how you create its mock

import dimodmock

sampler = ... # Create your sampler here
mock = dimodmock.SamplerMock.from_sampler(sampler)

The mock object created this way has the following properties identical to the source sampler

  • properties
  • parameters

The following methods are implemented:

  • sample
  • sample_qubo
  • sample_ising

and have the same interface as the source sampler. In particular, call to those methods will fail if unknown parameters are passed as keyword arguments. The returned samples are choosen randomly at uniform. By default, the returned samplesets have one sample. See below for overriding this behaviour.

Mocking structured samplers

You can also mock structured samplers. In this case use dimodmock.StructuredMock.from_sampler. In addition to the characteristics described above, the StructuredMock has also the following properties, identical to the source sampler:

  • edgelist
  • nodelist

Also, sample, sample_qubo, sample_ising methods respect the samplers structure, so trying to solve mismatching instance will fail.

Support for num_reads

By default, samplesets returned by all mocks contain only one sample. However, if the source sampler supports num_reads property (which is true in particular for DWaveSampler), it will be respected by the mock and the correct number of samples will be returned.

Advanced usage

It is also possible to create mocks by specifying their properties in their initializer. For SimpleMock you need to specify properties and parameters. For StructuredMock you also need to specify edgelist and nodelist.

For instance, suppose you want to create a mock of a structured sampler defined on full graph of 3 nodes, having a max_num_reads property and num_reads parameter. Here is how to do it:

from dimodmock import StructuredSampler

mock = StructuredSampler(
    nodelist=[0, 1, 2], 
    edgelist=[(0, 1), (1, 2), (2, 0)], 
    properties={"max_num_reads": 10},
    parameters={"num_reads": ["max_num_reads"]},
)

Caveats

Currently mocks don't validate keyword arguments passed to sample* methods. So, in the example above, mock will happily accept num_reads=100. In the future, additional validation might be implemented.

Reporting issues

Please report any issues you encounter using Github's issue tracker.