Simulating noisy variational quantum eigensolver with local noise models

Jinfeng Zeng, Zipeng Wu, Chenfeng Cao, Chao Zhang, Shi-Yao Hou, Pengxiang Xu, Bei ZengQuantum Eng 2021 · DOI

TL;DR

This paper investigates the impact of realistic noise models on the Variational Quantum Eigensolver (VQE) algorithm, a promising candidate for near-term quantum computers. By simulating VQE with amplitude damping, dephasing, and depolarizing noise, the authors demonstrate that noise degrades accuracy but that VQE can still outperform classical mean-field solutions, suggesting quantum entanglement can persist in noisy NISQ devices.

What Problem Does This Paper Address?

Near-term quantum computers, often referred to as Noisy Intermediate-Scale Quantum (NISQ) devices, are characterized by inherent noise that significantly impacts the performance of quantum algorithms. The Variational Quantum Eigensolver (VQE) is a hybrid quantum-classical algorithm designed to leverage these NISQ devices for tasks like finding the ground state energy of molecules or materials. However, a critical challenge for VQE's practical application is understanding how physical noise, such as amplitude damping, dephasing, and depolarizing errors, affects its accuracy and its ability to preserve quantum phenomena like entanglement. This paper aims to systematically study these noise effects through numerical simulations and experimental validation, providing insights into VQE's robustness and its potential for demonstrating quantum advantage.

Methods

The authors employ a two-pronged approach: numerical simulation and experimental validation.

  • Numerical Simulation: They simulate the VQE algorithm using various local noise models: amplitude damping, dephasing, and depolarizing noise. These models are applied to single-qubit and two-qubit gates within a hardware-efficient ansatz circuit. The simulations explore the impact of noise probability and circuit depth on the accuracy of the calculated ground state energy. They also compare the VQE results with classical mean-field solutions to assess the preservation of quantum entanglement.
  • Experimental Validation: The researchers implement VQE on real IBM quantum computers via the IBM Quantum Experience. They build a noise model that incorporates parameters derived from the calibration data of these devices, including thermal relaxation times (T1, T2) and gate error rates. This allows for a direct comparison between their simulation results and experimental outcomes.

Datasets

Not discussed in the paper. The paper focuses on simulating and experimenting with quantum algorithms on specific spin models rather than using pre-existing datasets.

Key Insights and Ideas

  • Noise Degradation is Inevitable but Quantifiable: As expected, increasing noise probability leads to a deviation of the VQE's calculated ground state energy from the exact value. This deviation generally accumulates with increasing circuit depth.
  • Depolarizing Noise is Particularly Damaging: Among the studied noise models, depolarizing noise shows a more pronounced negative impact on VQE accuracy, especially for certain spin models.
  • VQE Can Preserve Quantumness: Crucially, the paper demonstrates that even in the presence of realistic NISQ noise, VQE can still outperform classical mean-field solutions. This suggests that quantum entanglement, a key resource for quantum advantage, can be maintained in noisy quantum circuits at levels relevant for NISQ devices.
  • Simulation Matches Experiment: The numerical simulations, incorporating realistic noise models derived from IBMQ device calibration data, show good agreement with experimental results, validating the simulation approach.
  • Noise Accumulation Explained: The paper provides an analytical explanation for the accumulation of depolarizing noise with circuit depth, which is consistent with their numerical findings.

Key Results

  • The ground state energy calculated by noisy VQE deviates from the exact ground state energy as noise probability increases.
  • For the transverse field Ising model, the probability of depolarizing noise that allows VQE to surpass the mean-field solution varies with the number of qubits, ranging from approximately 0.024 to 0.041 for 2 to 8 qubits.
  • For the Heisenberg and transverse field Heisenberg models, VQE can tolerate significantly higher depolarizing noise probabilities (up to ~0.206 for 2 qubits) before falling below the mean-field solution.
  • Experimental results on IBMQ devices showed a relative deviation of less than 10% between the simulated noisy VQE and the experimental VQE results.

Limitations and Future Work

  • Limited Scope of Noise Models: While the paper studies three common local noise models, real quantum devices can exhibit more complex noise behaviors.
  • Specific Spin Models: The study focuses on one-dimensional spin models. Extending the analysis to more complex Hamiltonians and higher dimensions would be valuable.
  • Hardware-Efficient Ansatz: The use of a hardware-efficient ansatz is practical for NISQ devices, but exploring the impact of noise on other types of ansatze (e.g., physically motivated ones) could offer different insights.

The authors suggest that their work sheds light on practical research for noisy VQE and that a deeper understanding of noise effects will aid in developing error mitigation techniques for near-term quantum computers. Future work could involve exploring the noise effect on other variational quantum algorithms and investigating more sophisticated noise models.

Summary updated

BibTeX
@article{zeng2021simulating,
  author = {Jinfeng Zeng and Zipeng Wu and Chenfeng Cao and Chao Zhang and Shi-Yao Hou and Pengxiang Xu and Bei Zeng},
  title = {{Simulating noisy variational quantum eigensolver with local noise models}},
  journal = {Quantum Eng.},
  year = {2021},
  doi = {10.1002/que2.77},
}
Figure