3

I am using the method from this SE answer to calculate the expectation value using qiskit aqua operator logic which works well for a single circuit. Here is a simplified code snippet to demonstrate how it works.

import qiskit
from qiskit import QuantumCircuit, QuantumRegister, ClassicalRegister, Aer
import numpy as np
from qiskit.aqua import QuantumInstance
from qiskit.aqua.operators import PauliExpectation, CircuitSampler, StateFn, CircuitOp, CircuitStateFn

qctl = QuantumRegister(1)
psi  = QuantumCircuit(qctl)
psi  = CircuitStateFn(psi)


qctl = QuantumRegister(2)
op   = QuantumCircuit(qctl)
op.z(0)
op.ry(np.pi/4,0)
op   = CircuitOp(op)


backend = Aer.get_backend('qasm_simulator')
q_instance = QuantumInstance(backend, shots=1024)
measurable_expression = StateFn(op, is_measurement=True).compose(psi)
expectation = PauliExpectation().convert(measurable_expression)
sampler = CircuitSampler(q_instance).convert(expectation)
print('Expectation Value = ', sampler.eval())

However, I am applying it to a VQE application and for each iteration I have $x>1$ circuits. There is no issue when I run on my local machine because there is no queue time, but on the IBMQ machines I would like to submit a batch of the aqua expectation value circuits for each iteration to reduce queue times. For a list of circuits I can submit a batch using the IBMQJobManager() function, however, the method I am using does not store circuits as a list. My question is, can I use the jobmanager with the aqua expectation value to submit a batch of jobs to IBMQ? If not, is there a different way to submit a batch of the aqua expectation value circuits?

Martin Vesely
  • 15,244
  • 4
  • 32
  • 75
thespaceman
  • 597
  • 6
  • 16

1 Answers1

1

There is no need to use IBMQJobManager as QuantumInstance will do the necessary work for you. Just add your CircuitOps to a ListOp then pass it to StateFn constructor:

ops = []

... Construct your first circuit ...


Now, add it to the list:


ops.append(op1)


... Construct your second circuit ...


Now, add it to the list:


ops.append(op2)


measurable_expression = StateFn(ListOp(ops), is_measurement=True).compose(psi)

This way, QuantumInstance will assemble these circuits into a single Qobj and pass it to the backend.

Note that if the number of circuits is larger than backend.configuration().max_experiments, it will be splitted into multiple payloads.

Egretta.Thula
  • 11,986
  • 1
  • 13
  • 34