Variables and Parameters

Variables

This module offers tools to define different types of variables:

• Binary Variables

• Spin Variables

• Continuous Variables with the following customizable parameters:

  • Domain: Specifies the variable type, one of REAL, INTEGER, POSITIVE_INTEGER, BINARY, or SPIN.

  • Bounds: Defines the allowed value range of the variable.

  • Encoding: Specifies how the variable is represented using binary encodings, one of: Bitwise, DomainWall, or OneHot.

  • Precision: Applicable to Domain.REAL domain; defines the resolution (e.g., floating-point precision).

An example of creating different types of variables:

from qilisdk.core.variables import BinaryVariable, Bitwise, Domain, SpinVariable, Variable

x = Variable("x", domain=Domain.REAL, bounds=(1, 2), encoding=Bitwise, precision=1e-1)
s = SpinVariable("s")
b = BinaryVariable("b")

Continuous variables support indexing, where each index refers to a component of the binary-encoded form of the variable. For example:

print(x.to_binary())

Output:

(0.1) * x(0) + (0.2) * x(1) + (0.4) * x(2) + (0.30000000000000004) * x(3) + (1.0)

To index the first binary variable from the binary representation of x you can write: x[0]. Each binary variable configuration generates a float within the bounds, based on the defined precision. For instance:

x.evaluate([0, 1, 0, 0])

Output:

1.2

Parameters

Many components in QiliSDK expose symbolic parameters that can be optimized or re-bound at runtime. The Parameter class represents a scalar symbol with optional bounds, and Parameterizable provides a uniform API (get_parameter_names(), set_parameter_values()…) implemented by circuits, schedules, models, and more.

from qilisdk.core import Parameter

theta = Parameter("theta", value=0.5, bounds=(0.0, 1.0))
print(theta.value)     # 0.5
theta.set_value(0.75)
print(theta.bounds)    # (0.0, 1.0)

Parameters behave like symbolic variables in algebraic expressions, so you can combine them with other variables and evaluate terms without having to pass the parameter explicitly - its stored value is used automatically.

Objects that inherit from Parameterizable collect all the Parameter instances they encounter. For example:

from qilisdk.digital import Circuit, RX

circuit = Circuit(nqubits=1)
circuit.add(RX(0, theta=theta))

print(circuit.get_parameter_names())   # ['RX(0)_theta_0']
print(circuit.get_parameter_values())  # [0.75]
circuit.set_parameters({"RX(0)_theta_0": 0.9})

All parameter helper methods accept an optional where predicate to filter the exposed parameters. This is useful when you want to update only a subset, such as trainable parameters.

trainable_values = circuit.get_parameter_values(where=lambda p: p.is_trainable)
circuit.set_parameter_values([0.3], where=lambda p: p.is_trainable)

Note

Parameterizable objects expose parameter state through get_parameters(), get_parameter_names(), get_parameter_values(), and the corresponding set_* methods. Accessing .parameters directly is not part of the public API.