Parametric states and parametric estimators

This tutorial aims to introduce parametric states and parametric estimators, which are also important building blocks of variational algorithms.

Parametric state

In QURI Parts, there are 2 types of parametric states:

ParametricCircuitQuantumState: A state with parametric circuit acting on a zero state.
ParametricQuantumStateVector: A state with parametric circuit acting on a fixed state vector.

Here, parametric circuit can be either an UnboundParametricQuantumCircuit or a LinearMappedUnboundParametricQuantumCircuit. The parametric state can be created with the quantum_state function. Let's first create a parametric circuit.

import numpy as np
from quri_parts.circuit import LinearMappedUnboundParametricQuantumCircuit, CONST

linear_param_circuit = LinearMappedUnboundParametricQuantumCircuit(2)
theta, phi = linear_param_circuit.add_parameters("theta", "phi")

linear_param_circuit.add_H_gate(0)
linear_param_circuit.add_CNOT_gate(0, 1)
linear_param_circuit.add_ParametricRX_gate(0, {theta: 1/2, phi: 1/3, CONST: np.pi/2})
linear_param_circuit.add_ParametricRY_gate(0, {theta: -1/2, phi: 1/3})
linear_param_circuit.add_ParametricRZ_gate(1, {theta: 1/3, phi: -1/2, CONST: -np.pi/2})

Now, we create parametric state with it.

from quri_parts.core.state import quantum_state

# ParametricCircuitQuantumState
param_circuit_state = quantum_state(n_qubits=2, circuit=linear_param_circuit)
print(param_circuit_state)

# ParametricQuantumStateVector
param_state_vector = quantum_state(n_qubits=2, circuit=linear_param_circuit, vector=np.array([0.5, -0.5, 0.5j, -0.5j]))
print(param_state_vector)

#output
    ParametricCircuitQuantumState(n_qubits=2, circuit=<quri_parts.circuit.circuit_linear_mapped.ImmutableLinearMappedUnboundParametricQuantumCircuit object at 0x11b932fa0>)
    ParametricQuantumStateVector(n_qubits=2, circuit=<quri_parts.circuit.circuit_linear_mapped.ImmutableLinearMappedUnboundParametricQuantumCircuit object at 0x11ba8c7c0>, vector=[ 0.5+0.j  -0.5+0.j   0. +0.5j -0. -0.5j])

We can assign concrete values to the circuit inside the parametric states with the .bind_parameters state. After binding the parameters, ParametricCircuitQuantumState becomes a GeneralCircuitQuantumState and ParametricQuantumStateVector becomes a QuantumStateVector.

params = [0.1, 0.2]

# ParametricCircuitQuantumState -> GeneralCircuitQuantumState
circuit_state = param_circuit_state.bind_parameters(params)

# ParametricQuantumStateVector -> QuantumStateVector
vector_state = param_state_vector.bind_parameters(params)

Parametric estimators

In addition to the QuantumEstimators and ConcurrentQuantumEstimators introduced in the estimator tutorial, QURI Parts also provide estimators for parametric states. They are:

ParametricQuantumEstimator: Estimate an operator's expectation value for a parametric state with a set of parameters.
ConcurrentParametricQuantumEstimator: Estimate an operator's expectation value for a parametric state with multiple sets of parameters.

Note that different from ConcurrentQuantumEstimator, a ConcurrentParametricQuantumEstimator only performs estimation for a parametric state at a time. The concurrent estimation is done across multiple sets of parameters.

Parametric estimate with a `QuantumEstimator`

A GeneralCircuitQuantumState can be created from a circuit obtained by binding values to a parametric circuit and a QuantumEstimator can be used to estimate an expectation value of an operator for the state:

from quri_parts.core.state import quantum_state
from quri_parts.qulacs.estimator import create_qulacs_vector_estimator

circuit = linear_param_circuit.bind_parameters([0.2, 0.3])
circuit_state = quantum_state(2, circuit=circuit)

from quri_parts.core.operator import Operator, pauli_label
op = Operator({
    pauli_label("X0 Y1"): 0.5 + 0.5j,
    pauli_label("Z0 X1"): 0.2,
})


estimator = create_qulacs_vector_estimator()

estimate = estimator(op, circuit_state)
print(f"Estimated expectation value: {estimate.value}")

However, it is more straightforward to use ParametricCircuitQuantumState. It has a few benefits:

It makes it clear that the state is parametric and possible to treat parameter-related problems in terms of quantum state (e.g. gradient of an expectation value for the state with respect to its parameters).
It may improve performance for some circuit simulators (e.g. Qulacs).

With this in mind, we introduce solid implementations of these parametric estimators provided by the quri_parts.qulacs package.

The Qulacs parametric estimators

The quri_parts.qulacs package provides both vector parametric estimators and density matrix parametric estimators. Here we demonstrate the Qulacs vector parametric estimators for performing noiseless estimation for parametric states. They are generated by the

create_qulacs_vector_parametric_estimator
create_qulacs_vector_concurrent_parametric_estimator

functions.

from quri_parts.qulacs.estimator import (
    create_qulacs_vector_parametric_estimator,
    create_qulacs_vector_concurrent_parametric_estimator,
)

parametric_estimator = create_qulacs_vector_parametric_estimator()
concurrent_parametric_estimator = create_qulacs_vector_concurrent_parametric_estimator()

Next, let's create an operator for us to estimate.

from quri_parts.core.operator import Operator, pauli_label

operator = Operator(
    {
        pauli_label("X0 Y1"): 0.5 + 0.5j,
        pauli_label("Z0 X1"): 0.2,
    }
)

Estimate for parametric circuit state

The Qulacs parametric estimator can be used to estimate the expectation values for parametric circuit state.

# Estimate for parametric circuit state
print("Estimate with a set of circuit parameter:")
print(parametric_estimator(operator, param_circuit_state, [0.1, 0.2]))

# Concurrent estimate for parametric circuit state
print("")
print("Concurrent Estimate with two sets of circuit parameters:")
print(concurrent_parametric_estimator(operator, param_circuit_state, [[0.1, 0.2], [0.3, 0.4]]))

#output
    Estimate with a set of circuit parameter:
    _Estimate(value=(-0.6962182648863564-0.4982686669563893j), error=0.0)

    Concurrent Estimate with two sets of circuit parameters:
    [_Estimate(value=(-0.6962182648863564-0.4982686669563893j), error=0.0), _Estimate(value=(-0.6896044045783039-0.49823172433086493j), error=0.0)]

Estimate for parametric vector state

The Qulacs parametric estimator can also be used to estimate the expectation values for parametric state vectors.

# Estimate for parametric state vector
print("Estimate with a set of circuit parameter:")
print(parametric_estimator(operator, param_state_vector, [0.1, 0.2]))

# Concurrent estimate for parametric state vector
print("")
print("Concurrent Estimate with two sets of circuit parameters:")
print(concurrent_parametric_estimator(operator, param_state_vector, [[0.1, 0.2], [0.3, 0.4]]))

#output
    Estimate with a set of circuit parameter:
    _Estimate(value=(-0.02316087896430696+6.461699812418398e-05j), error=0.0)

    Concurrent Estimate with two sets of circuit parameters:
    [_Estimate(value=(-0.02316087896430696+6.461699812418398e-05j), error=0.0), _Estimate(value=(-0.055857039676488814-0.00023256584020162063j), error=0.0)]

Parametric states and parametric estimators