OpenQEMIST and Rigetti example¶

This notebook shows how OpenQEMIST can be combined with the Rigetti stack to use the Variational Quantum Eigensolver (VQE) as an electronic structure solver, and combine it with a problem-decomposition technique such as Density Matrix Embedding Theory (DMET).

VQE Example¶

This tutorial assumes that the user has correctly set up and configured the OpenQEMIST package. The Variational Quantum Eigensolver (VQE):math:^{1,2} is a hybrid quantum-classical algorithm for simulating quantum systems. We here focus on VQE within the context of solving the molecular electronic structure problem for the ground-state energy of a molecular system. In VQE, we first prepare the trial wavefunction (quantum state) \(\vert \Psi(\vec{\theta}) \rangle = U(\vec{\theta}) \vert 0 \rangle\) based on an ansatz that depends on \(m\) parameters defining \(\vec{\theta}=(\theta_1, \theta_2, \ldots, \theta_m)\). The expectation value of the Hamiltonian (\(\hat{H}\)), \(\langle \Psi(\vec{\theta}) \vert \hat{H} \vert \Psi(\vec{\theta}) \rangle\), will then be simulated.

The expectation value can be minimized based on the variational principle,

begin{equation} E = min_{vec{theta}} frac{langle Psi(vec{theta}) vert hat{H} vert Psi(vec{theta}) rangle}{langle Psi(vec{theta}) vert Psi(vec{theta}) rangle} geq E_{text{gs}}nonumber end{equation}

which ensures that the energy computed will be an upper bound to the true ground-state energy \(E_{\text{gs}}\). This allows us using classical minimizers to find optimal parameters \(\vec{\theta}\) for the ground-state energy \(E_{\text{gs}}\).

VQE can be performed using OpenQEMIST in conjuction with the Rigetti stack for calculating the ground state energy of a molecular system. The unitary coupled-cluster ansatz can be used to prepare the trial wavefunction \(\vert \Psi(\vec{\theta}) \rangle\). In this notebook, we will show you an example using a small molecule, the hydrogen molecule (H:math:_text{2}), for a simulation using VQE.

from openqemist.electronic_structure_solvers import VQESolver, FCISolver
from openqemist.quantum_solvers import RigettiParametricSolver

from pyscf import gto

# Build the molecule
H2 = [['H', [0.0,   0.0,   0.0]], ['H', [0.0,   0.0,   0.74137727]]]
mol = gto.Mole()
mol.atom = H2
mol.basis = "sto-3g"
mol.charge = 0
mol.spin = 0
mol.build()

# Configure the solver object
vqe_solver = VQESolver()
vqe_solver.hardware_backend_type = RigettiParametricSolver
vqe_solver.ansatz_type = RigettiParametricSolver.Ansatze.UCCSD

We can now simulate the molecule and get its energy.

energy_fci = FCISolver().simulate(mol)
energy_vqe = vqe_solver.simulate(mol)

print("\nFCI energy = ", energy_fci)
print("VQE energy = ", energy_vqe)

FCI energy =  -1.1372704220924401
VQE energy =  -1.1372704138513532

It is possible to use different initial parameters for the optimization:

# Using custom initial parameters
# Getting the dimension of the initial parameters vector
num_var_params = vqe_solver.hardware_backend.amplitude_dimension
# Set the intial parameters for the solver
vqe_solver.initial_var_params = [0.01 for i in range(num_var_params)]

vqe_solver.simulate(mol)

VQE : initial variational parameters:
 [0.01, 0.01]

-1.1372665775495083

Using the QVM shot-based simulator¶

To use the QVM, we can use the backend_parameters attribute of the VQESolver object. The VQE object then configures the hardware backend automatically. Because the QVM is slower than the default wavefunction simulator backend, we specify an optimizer function that returns after a few iterations, in the interest of showing the usage of the solver in a reasonable time. See the documentation for more details about using custom optimizers. This interface is what would also be used to target a QPU backend in the future.

def quick_optimizer(backend, amplitudes):
        from scipy.optimize import minimize

        print("Running using custom optimizer.")

        # We use a force the optimizer to return after 2 iterations.
        result = minimize(backend, amplitudes, method='COBYLA',
                options={'disp':True, 'maxiter':2})

        return result.fun

vqe = VQESolver()
vqe.optimizer = quick_optimizer

To use the QVM, we can use the backend_parameters attribute of the VQESolver object. The VQE object then configures the hardware backend automatically. We can then run the simulation with the object. The number of shots can also be set with this parameter.

Note that because we restricted the optimizer to 2 iterations and reduced the number of shots, the resulting energy will not be accurate.

vqe.hardware_backend_type = RigettiParametricSolver
vqe.ansatz_type = RigettiParametricSolver.Ansatze.UCCSD
vqe.backend_parameters = {'backend': '4q-qvm', 'n_shots': 10}

energy = vqe.simulate(mol)
print("Unconverged QMV energy: ", energy)

Unconverged QMV energy:  -1.146711378495224

DMET Example¶

At the current early stage of quantum hardware, the available computational resource is yet very limited. Thus, it is still challenging to perform accurate electronic structure calculations on actual quantum hardware. Simulation on classical computer requires large computational cost as well. Therefore, we need to reduce the problem size while maintaining the accuracy of electronic structure calculation to solve a problem for small sized molecules to perform quantum simulations.

Density Matrix Embedding Theory (DMET):math:^{3,4} is a powerful problem decomposition technique to reduce the problem size, while maintaining the accuracy of the electronic structure calculation. The DMET method decomposes a molecule into fragments, and each fragment is treated as an open quantum system that is entangled with each of the other fragments, all taken together to be that fragment’s surrounding environment (or “bath”). VQE algorithm can be used with DMET using OpenQEMIST in conjuction with the Rigetti stack.

In this notebook, we will show you an example of H\(_\text{4}\) molecule for DMET simulation using VQE as an electronic structure solver.

from openqemist.problem_decomposition import DMETProblemDecomposition
from openqemist.problem_decomposition.electron_localization import meta_lowdin_localization

H4 = [['H', [0.7071067811865476,   0.0,                 0.0]],
      ['H', [0.0,                  0.7071067811865476,  0.0]],
      ['H', [-1.0071067811865476,  0.0,                 0.0]],
      ['H', [0.0,                 -1.0071067811865476,  0.0]]]

mol = gto.Mole()
mol.atom = H4
mol.basis = "minao"
mol.charge = 0
mol.spin = 0
mol.build()

dmet = DMETProblemDecomposition()
dmet.verbose = True
dmet.electron_localization_method = meta_lowdin_localization
# Set the DMET object to use the solver that we configured above
dmet.electronic_structure_solver = vqe_solver
energy_vqe = dmet.simulate(mol, [1,1,1,1])

print("The DMET energy is: ", energy_vqe)

The DMET energy is:  -1.9916111022655567

References¶

Alberto Peruzzo, Jarrod McClean, Peter Shadbolt, Man-Hong Yung, Xiao-Qi Zhou, Peter J. Love, Alán Aspuru-Guzik, and Jeremy L. O’Brien, “A variational eigenvalue solver on a photonic quantum processor”, Nat. Commun., 5, 4213 (2014).
Jarrod R. McClean, Jonathan Romero, Ryan Babbush, and Alán Aspuru-Guzik, “The theory of variational hybrid quantum-classical algorithms”, New J. Phys., 18, 023023 (2016).
Gerald Knizia and Garnet K.-L. Chan, “Density Matrix Embedding: A Simple Alternative to Dynamical Mean-Field Theory”, Phys. Rev. Lett., 109, 186404 (2012).
Sebastian Wouters, Carlos A. Jiménez-Hoyos, Qiming Sun, and Garnet K.-L. Chan, “A Practical Guide to Density Matrix Embedding Theory in Quantum Chemistry”, J. Chem. Theory Comput., 12, pp. 2706–2719 (2016).