I am looking for an efficient way to create a Hamiltonian in Qiskit.

Following are my desired input and outputs

input: 'ZZI'

output: Z^Z^I, where Z and I are operators.

Here is how I am doing it right now:

from qiskit.circuit.library import TwoLocal
from qiskit.opflow import X, Y, Z, I   
from qiskit.utils import QuantumInstance
from qiskit import *
from qiskit.algorithms.optimizers import COBYLA
from qiskit.algorithms import VQE

weights = [i for i in range(1,4)]

hamiltonian = weights[0]*(Z^Z^I) + weights[1]*(Z^I^Z) + weights[2]*(I^Z^Z)

num_qubits = hamiltonian.num_qubits
ansatz = TwoLocal(num_qubits, ['ry','rz'], 'cx', 'linear', reps=1, insert_barriers=True)

qi = QuantumInstance(Aer.get_backend('statevector_simulator'))
optimizer = COBYLA(maxiter=100)
vqe = VQE(ansatz, optimizer=optimizer, quantum_instance=qi)
result = vqe.compute_minimum_eigenvalue(hamiltonian)

If I have to run the same experiment on a 4 qubit, I need to redo the hamiltonian again. For example it would be:

hamiltonian = weights[0]*(Z^Z^I^I) + weights[1]*(Z^I^Z^I) + weights[2]*(Z^I^I^Z) + weights[3]*(I^Z^Z^I) + weights[4]*(I^Z^I^Z) + + weights[4]*(I^I^Z^Z)

This option is not scalable when I have a 15 qubit circuit. I would really appreciate it if anyone can guide me in a way that I can build a scalable hamiltonian.

I tried the following method and no luck:

from qiskit.quantum_info.operators import Operator, Pauli
hamiltonian = Operator(Pauli(label='ZZI')) + Operator(Pauli(label='ZIZ')) + Operator(Pauli(label='IZZ'))

Any guidance would be really appreciated!


You use PauliSumOp to do (for example) the following:

from qiskit.opflow.primitive_ops import PauliSumOp
paulis  =  ['IZ', 'XZ', 'YZ', 'ZZ', 'XX']
weights  =  [1,2,3,4,5]
pauli_op = [([pauli,weight]) for pauli,weight in zip(paulis,weights)]
hamiltonian = PauliSumOp.from_list([ op for op in pauli_op ])

You can then pass this hamiltonian object into the VQE instance like you did.

  • 1
    $\begingroup$ This worked. Thank you very much! $\endgroup$ Nov 19 at 12:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.