Solver Properties¶

The properties that characterize behaviors and supported features of SAPI solvers1 can be queried through SAPI. This chapter defines these properties2.

1

Ocean software also provides classical solvers such as dwave-greedy that you can run locally. See the Ocean documentation for information on parameters of such solvers.

2

Ocean software classes such as the DWaveSampler class may include various other properties that are derived and stored client-side, not read from SAPI in the properties dict; for example, the dwave.cloud.solver.BaseSolver might derive an avg_load property. See the Ocean documentation for those properties.

The examples below use the following setup:

>>> from dwave.system import DWaveSampler, LeapHybridSampler, LeapHybridCQMSampler
...
>>> qpu_2000q = DWaveSampler(solver={'topology__type': 'chimera'})
>>> hybrid_bqm_sampler = LeapHybridSampler()
>>> hybrid_cqm_sampler = LeapHybridCQMSampler()


anneal_offset_ranges¶

Array of ranges of valid anneal offset values, in normalized offset units, for each qubit. The negative values represent the largest number of normalized offset units by which a qubit’s anneal path may be delayed. The positive values represent the largest number of normalized offset units by which a qubit’s anneal path may be advanced.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["anneal_offset_ranges"][50:55]
[[-0.6437569799443247, 0.5093937328282595],
[-0.6409899716199288, 0.5697835920492209],
[-0.6468663649068287, 0.5211452841036136],
[-0.6435316809914349, 0.5011584701861175],
[-0.6500326191157569, 0.5392206819650301]]


anneal_offset_step¶

Quantization step size of anneal offset values in normalized units.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["anneal_offset_step"]
-0.00017565852000668507


anneal_offset_step_phi0¶

Quantization step size in physical units (annealing flux-bias units): $\Phi_0$.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["anneal_offset_step_phi0"]
1.486239425109832e-05


annealing_time_range¶

Range of time, in microseconds with a resolution of 0.01 $\mu s$, possible for one anneal (read). The lower limit in this range is the fastest quench possible for this solver.

Default annealing time is specified by the default_annealing_time property.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["annealing_time_range"]
[1.0, 2000.0]


beta_range¶

Range of values for beta.

Default $\beta$ is specified by the default_beta property.

category¶

Type of solver; for example,

• qpu—quantum computers such as the Advantage.

• hybrid—quantum-classical hybrid; typically one or more classical algorithms run on the problem while outsourcing to a quantum processing unit (QPU) parts of the problem where it benefits most.

• software—emulators such as a sampling emulator.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["category"]
'hybrid'


chip_id¶

Name of the solver.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["chip_id"]


couplers¶

Couplers in the working graph. A coupler contains two elements [q1, q2], where both q1 and q2 appear in the list of working qubits, in the range [0, num_qubits - 1] and in ascending order (i.e., q1 < q2). These are the couplers that can be programmed with nonzero $J$ values.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["couplers"][:5]
[[30, 31], [31, 32], [32, 33], [33, 34], [34, 35]]


default_annealing_time¶

Default time, in microseconds with a resolution of 0.01 $\mu s$, for one anneal (read). You can change the annealing time for a given problem by using the annealing_time or anneal_schedule parameters, but do not exceed the upper limit given by the annealing_time_range property.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["default_annealing_time"]
20.0


default_beta¶

Default value for $\beta$.

default_programming_thermalization¶

Default time, in microseconds with a resolution of 0.01 $\mu s$, that the system waits after programming the QPU for it to return to base temperature. This value contributes to the total qpu_programming_time, which is returned by SAPI with the problem solutions.

You can change this value using the programming_thermalization parameter.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["default_programming_thermalization"]
1000.0


Default time, in microseconds with a resolution of 0.01 $\mu s$, that the system waits after each state is read from the QPU for it to cool back to base temperature. This value contributes to the qpu_delay_time_per_sample field, which is returned by SAPI with the problem solutions.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["default_readout_thermalization"]
0.0


extended_j_range¶

Extended range of values possible for the coupling strengths (quadratic coefficients), $J$, for this solver. Strong negative couplings may be necessary for some embeddings; however, such chains may require additional calibration through the flux_biases parameter to compensate for biases introduced by strong negative couplings. See also per_qubit_coupling_range.

extended $J$ ranges and flux-bias offsets cannot be used with autoscaling (auto_scale) or spin-reversal transforms (num_spin_reversal_transforms).

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["extended_j_range"]
[-2.0, 1.0]


h_gain_schedule_range¶

Range of the time-dependent gain applied to qubit biases for this solver.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["h_gain_schedule_range"]
[-4.0, 4.0]


h_range¶

Range of values possible for the qubit biases (linear coefficients), $h$, for this solver.

The auto_scale parameter, which rescales $h$ and $J$ values in the problem to use as much of the range of $h$ (h_range) and the range of $J$ (j_range) as possible, enables you to submit problems with values outside these ranges and have the system automatically scale them to fit.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["h_range"]
[-2.0, 2.0]


j_range¶

Range of values possible for the coupling strengths (quadratic coefficients), $J$, for this solver.

The auto_scale parameter, which rescales $h$ and $J$ values in the problem to use as much of the range of $h$ (h_range) and the range of $J$ (j_range) as possible, enables you to submit problems with values outside these ranges and have the system automatically scale them to fit.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["j_range"]
[-1.0, 1.0]


max_anneal_schedule_points¶

Maximum number of points permitted in a PWL waveform submitted to change the default anneal schedule.

For reverse annealing, the maximum number of points allowed is one more than the number given in the max_anneal_schedule_points property.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["max_anneal_schedule_points"]
12


max_h_gain_schedule_points¶

Maximum number of points permitted in a PWL waveform submitted to set a time-dependent gain on linear coefficients (qubit biases, see the h parameter) in the Hamiltonian.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["max_h_gain_schedule_points"]
20


maximum_number_of_biases¶

Maximum number of biases, both linear and quadratic in total, accepted by the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["maximum_number_of_biases"]
200000000


maximum_number_of_constraints¶

Maximum number of constraints accepted by the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["maximum_number_of_constraints"]
100000


maximum_number_of_linear_biases_real¶

Maximum number of linear biases of type floating-point/real numbers counted across all constraints and the model’s objective.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["maximum_number_of_linear_biases_real"]
75000000


Maximum number of variables that have interactions (quadratic biases)—counted across all constraints (meaning that the same variable can be counted multiple times if it appears in interactions in more than one constraint)—accepted by the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["maximum_number_of_quadratic_variables"]
15000000


Maximum number of variables that have interactions (quadratic biases) of type floating-point/real numbers—counted across all constraints (meaning that the same variable can be counted multiple times if it appears in interactions in more than one constraint)—accepted by the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["maximum_number_of_quadratic_variables_real"]
0


maximum_number_of_variables¶

Maximum number of problem variables accepted by the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["maximum_number_of_variables"]
1000000


maximum_time_limit_hrs¶

Maximum allowed run time, in hours, that can be specified for the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["maximum_time_limit_hrs"]
24.0


minimum_time_limit¶

Minimum required run time, in seconds, the solver must be allowed to work on the given problem. Specifies the minimum time as a piecewise-linear curve defined by a set of floating-point pairs. The second element is the minimum required time; the first element in each pair is some measure of the problem, dependent on the solver:

• For hybrid BQM solvers, this is the number of variables.

• For hybrid DQM solvers, this is a combination of the numbers of interactions, variables, and cases that reflects the “density” of connectivity between the problem’s variables.

The minimum time for any particular problem is a linear interpolation calculated on two pairs that represent the relevant range for the given measure of the problem. For example, if minimum_time_limit for a hybrid BQM solver were [[1, 0.1], [100, 10.0], [1000, 20.0]], then the minimum time for a 50-variable problem would be 5 seconds, the linear interpolation of the first two pairs that represent problems with between 1 to 100 variables.

For more details, see the Ocean samplers documentation for solver methods that calculate this parameter, and their descriptions.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["minimum_time_limit"]
[[1, 3.0],
[1024, 3.0],
[4096, 10.0],
[10000, 40.0],
[30000, 200.0],
[100000, 600.0],
[1000000, 600.0]]


minimum_time_limit_s¶

Minimum required run time, in seconds, the solver must be allowed to work on any given problem.

For more details, see the Ocean samplers documentation for solver methods that calculate the default runtime of a given problem.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["minimum_time_limit_s"]
5


num_biases_multiplier¶

Multiplier used in the internal calculation of the minimum run time, minimum_time_limit_s, the solver must be allowed to work on the given problem.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["num_biases_multiplier"]
4.65705646e-06


num_constraints_multiplier¶

Multiplier used in the internal calculation of the minimum run time, minimum_time_limit_s, the solver must be allowed to work on the given problem.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["num_constraints_multiplier"]
6.44385747e-09


num_qubits¶

Total number of qubits, both working and nonworking, in the QPU.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["num_qubits"]
5760


Range of values possible for the number of reads that you can request for a problem.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["num_reads_range"]
[1, 10000]


num_variables_multiplier¶

Multiplier used in the internal calculation of the minimum run time, minimum_time_limit_s, the solver must be allowed to work on the given problem.

Example¶

For the solver configured in the example setup above,

>>> hybrid_cqm_sampler.properties["num_variables_multiplier"]
0.000157411458


parameters¶

List of the parameters supported for the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["parameters"]
{'time_limit': 'Maximum requested runtime in seconds.'}


per_group_coupling_range¶

Coupling range permitted per group of qubits for this solver. The couplers of every qubit are divided into two similar sized groups to which apply limitations on the total of all coupling strengths.

Strong negative couplings may be necessary for some embeddings, and can be enabled by using the extended_j_range range of $J$ values. However, the total couplings for a group of qubits must not exceed the range specified by this property. Also, chains may require additional calibration through the flux_biases parameter to compensate for biases introduced by strong negative couplings.

Example¶

>>> from dwave.system import DWaveSampler
[-13.000 10.000]


per_qubit_coupling_range¶

Coupling range permitted per qubit for this solver.

Strong negative couplings may be necessary for some embeddings, and can be enabled by using the extended_j_range range of $J$ values. However, the total couplings for a qubit must not exceed the range specified by this property. Also, chains may require additional calibration through the flux_biases parameter to compensate for biases introduced by strong negative couplings.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["per_qubit_coupling_range"]
[-18.0, 15.0]


problem_run_duration_range¶

Range of time, in microseconds with a resolution of 0.01 $\mu s$, that a submitted problem is allowed to run.

For details about how the run-duration limit is calculated, see Runtime Limit for Your QMIs.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["problem_run_duration_range"]
[0.0, 1000000.0]


problem_timing_data¶

Key-value pairs that contain sample timing data and specify the version 1.0.x problem-timing model used to estimate problem runtimes for a QPU solver. For details about using the timing data to estimate the QPU access times for problem submissions, see the Estimating the QPU Access Time for Problems section.

The key-value pairs are the following:

• decorrelation_max_nominal_anneal_time—Parameter, in microseconds, used in the calculation of inter-sample correlation time. For more information, see the reduce_intersample_correlation solver property.

• decorrelation_time_range—Range of inter-sample correlation time, in microseconds, to add to the per-sample delay time as specified in the qpu_delay_time_per_sample field. Scales linearly with the anneal time from 0 to the amount returned in the decorrelation_max_nominal_anneal_time field. For more details, see the reduce_intersample_correlation solver parameter.

• default_annealing_time—See the default_annealing_time solver property.

• default_programming_thermalization—See the default_programming_thermalization solver property.

• default_readout_thermalization—See the default_readout_thermalization solver property.

• qpu_delay_time_per_sample—Default for the per-sample delay time in microseconds.

• readout_time_model—The readout-time model used to generate the set of data points; these data points define a piecewise-linear curve of typical readout times as a function of the number of qubits in a representative set of problems embedded on a QPU. The data points are provided in the readout_time_model_parameters field. Supported values that specify the scale of the curve are:

• pwl_log_log—the log scale is used for both the horizontal and vertical axes.

• pwl_linear—the linear scale is used for both the horizontal and vertical axes.

• readout_time_model_parameters—A set of $2N$ data points generated by the readout-time model specified in the readout_time_model field.

• reverse_annealing_with_reinit_delay_time_delta—Extra per-sample delay time, in microseconds, when reverse annealing is used and the reinitialize_state parameter is true.

• reverse_annealing_with_reinit_prog_time_delta—Extra programming time, in microseconds, when reverse annealing is used and the reinitialize_state solver parameter is true.

• reverse_annealing_without_reinit_delay_time_delta—Extra per-sample delay time, in microseconds, when reverse annealing is used and the reinitialize_state solver parameter is false.

• reverse_annealing_without_reinit_prog_time_delta—Extra programming time, in microseconds, when reverse annealing is used and the reinitialize_state solver parameter is false.

• typical_programming_time—Typical programming time in microseconds.

• version—Version of the problem-timing model.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["problem_timing_data"]
{'version': '1.0.0', 'typical_programming_time': 14072.88,
'reverse_annealing_with_reinit_prog_time_delta': 0.0,
'reverse_annealing_without_reinit_prog_time_delta': 5.55,
'default_programming_thermalization': 1000.0, 'default_annealing_time': 20.0,
2.711975459489206, 3.1639057672764026, 3.750276915153992, 1.539131714800995,
1.8726623164229292, 2.125631787097315, 2.332672340068556, 2.371606651233025,
2.3716219271760215], 'qpu_delay_time_per_sample': 20.54,
'reverse_annealing_with_reinit_delay_time_delta': -4.5,
'reverse_annealing_without_reinit_delay_time_delta': -1.5,
'decorrelation_time_range': [500.0, 10000.0]}


programming_thermalization_range¶

Range of time, in microseconds with a resolution of 0.01 $\mu s$, possible for the system to wait after programming the QPU for it to cool back to base temperature. This value contributes to the total qpu_programming_time, which is returned by SAPI with the problem solutions.

You can change this value using the programming_thermalization parameter, but be aware that values lower than the default accelerate solving at the expense of solution quality.

Default value for a solver is given in the default_programming_thermalization property.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["programming_thermalization_range"]
[0.0, 10000.0]


quota_conversion_rate¶

Rate at which user or project quota is consumed for the solver as a ratio to QPU solver usage. Different solver types may consume quota at different rates.

Time is deducted from your quota according to:

$\frac{num\_seconds}{quota\_conversion\_rate}$

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["quota_conversion_rate"]
1


qubits¶

Indices of the working qubits in the working graph.

Note

In a D-Wave QPU, the set of qubits and couplers that are available for computation is known as the working graph. The yield of a working graph is typically less than the total number of qubits and couplers that are fabricated and physically present in the QPU.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["qubits"][:5]
[30, 31, 32, 33, 34]


Range of time, in microseconds with a resolution of 0.01 $\mu s$, possible for the system to wait after each state is read from the QPU for it to cool back to base temperature. This value contributes to the qpu_delay_time_per_sample field, which is returned by SAPI with the problem solutions.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["readout_thermalization_range"]
[0.0, 10000.0]


supported_problem_types¶

Indicates what problem types are supported for the solver.

QPU and QPU-like solvers support the following energy-minimization problem types:

• qubo—Quadratic unconstrained binary optimization (QUBO) problems; use $0/1$-valued variables.

• ising—Ising model problems; use $-1/1$-valued variables.

Hybrid solvers support the following energy-minimization problem types:

• bqm—binary quadratic model (BQM) problems; use $0/1$-valued variables and $-1/1$-valued variables; constraints are typically represented as penalty models.

• cqm—constrained quadratic model (CQM) problems; use both binary-valued variables (both $0/1$-valued variables and $-1/1$-valued variables) and integer-valued variables; support constraints natively.

• dqm—discrete quadratic model (DQM) problems; use variables that can represent a set of values such as {red, green, blue, yellow} or {3.2, 67}; constraints are typically represented as penalty models.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["supported_problem_types"]
['ising', 'qubo']


tags¶

May hold attributes about a solver that you can use to have a client program choose one solver over another.

Example¶

For the solver configured in the example setup above,

>>> qpu_2000q.properties["tags"]
['lower_noise']


topology¶

Indicates the topology type (chimera or pegasus) and shape of the QPU graph.

Example¶

The topology seen in this example is a C16 Chimera graph, meaning that the D-Wave 2000Q QPU has 16 x 16 blocks of Chimera unit cells, and each unit cell has K4,4 connectivity.

For the solver configured in the example setup above,

>>> qpu_2000q.properties["topology"]
{'type': 'chimera', 'shape': [16, 16, 4]}


version¶

Version number of the solver.

Example¶

For the solver configured in the example setup above,

>>> hybrid_bqm_sampler.properties["version"]
'2.0'


vfyc¶

Flag indicating whether this solver is a VFYC solver.

Example¶

For the solver configured in the example setup above,

>>> qpu_advantage.properties["vfyc"]
False