PaperlayerView exact revision on arXiv

Source-linked AI summary

QuantumNAS: Noise-Adaptive Search for Robust Quantum Circuits

Hanrui Wang, Yongshan Ding, Jiaqi Gu, Zirui Li, Yujun Lin, David Z. Pan, Frederic T. Chong, Song Han

arXiv:2107.10845v5quant-phcs.ARcs.LG

TL;DR

QuantumNAS addresses the challenge of making variational quantum circuits resilient to noise while searching circuit structures and qubit mappings efficiently. It combines SuperCircuit-based search, noise-adaptive evolutionary co-search, and iterative pruning, achieving higher robustness and performance than baselines across QML and VQE experiments.

  • Problem

    Quantum noise can offset the capacity gains of larger parameterized circuits, while circuit and qubit-mapping search must handle large design spaces and sampling variance.

  • Method

    QuantumNAS co-searches variational circuit architectures and qubit mappings using a SuperCircuit that trains sampled SubCircuits, followed by noise-aware evolutionary search and iterative gate pruning.

  • Results

    QuantumNAS consistently outperforms baselines on QML tasks, exceeding 85% 4-class and 95% 2-class accuracy, while pruning improves accuracy and reduces circuit size.

  • Takeaways & Limitations

    Noise information in the loop supports robust circuit and mapping search across devices, while pruning removes redundant gates without sacrificing noise-free accuracy.

Abstract

from arXiv · show

Quantum noise is the key challenge in Noisy Intermediate-Scale Quantum (NISQ) computers. Previous work for mitigating noise has primarily focused on gate-level or pulse-level noise-adaptive compilation. However, limited research efforts have explored a higher level of optimization by making the quantum circuits themselves resilient to noise. We propose QuantumNAS, a comprehensive framework for noise-adaptive co-search of the variational circuit and qubit mapping. Variational quantum circuits are a promising approach for constructing QML and quantum simulation. However, finding the best variational circuit and its optimal parameters is challenging due to the large design space and parameter training cost. We propose to decouple the circuit search and parameter training by introducing a novel SuperCircuit. The SuperCircuit is constructed with multiple layers of pre-defined parameterized gates and trained by iteratively sampling and updating the parameter subsets (SubCircuits) of it. It provides an accurate estimation of SubCircuits performance trained from scratch. Then we perform an evolutionary co-search of SubCircuit and its qubit mapping. The SubCircuit performance is estimated with parameters inherited from SuperCircuit and simulated with real device noise models. Finally, we perform iterative gate pruning and finetuning to remove redundant gates. Extensively evaluated with 12 QML and VQE benchmarks on 14 quantum computers, QuantumNAS significantly outperforms baselines. For QML, QuantumNAS is the first to demonstrate over 95% 2-class, 85% 4-class, and 32% 10-class classification accuracy on real QC. It also achieves the lowest eigenvalue for VQE tasks on H2, H2O, LiH, CH4, BeH2 compared with UCCSD. We also open-source TorchQuantum (https://github.com/mit-han-lab/torchquantum) for fast training of parameterized quantum circuits to facilitate future research.

I. INTRODUCTION

QuantumNAS addresses NISQ-era noise by jointly searching variational circuits and qubit mappings, rather than optimizing only compilation. Its SuperCircuit separates parameter training from circuit search, enabling noise-aware evolutionary exploration and pruning for more robust circuits.

  • NISQ devices have error rates of 10^-3 to 10^-2, making quantum-error reduction necessary to bridge algorithmic demands and available hardware.
  • Existing noise-adaptive methods mainly optimize compilation components such as qubit mapping, scheduling, and crosstalk, leaving circuit-level co-design less explored.
  • QuantumNAS targets a task and device by jointly searching a robust variational circuit and its corresponding qubit mapping with noise information in the loop.
  • On MNIST-4, increasing parameters improves noise-free accuracy but can reduce measured accuracy because additional gates accumulate noise; circuit choice also creates large variance.
  • A SuperCircuit shares gates across many SubCircuits, allowing one training process to estimate SubCircuit performance before evolutionary co-search evaluates circuit–mapping pairs under device noise.
  • Iterative pruning and finetuning remove redundant gates while preserving noise-free performance, yielding fewer noise sources and improved measured performance.

II. BACKGROUND AND MOTIVATION

Quantum circuits use parameterized gates for QML and VQE, but selecting trainable structures and parameters is costly in noisy NISQ hardware. QuantumNAS addresses this with SuperCircuit-based estimation, evolutionary circuit–mapping search, and stability-oriented sampling.

  • Motivation: Quantum circuits must balance expressive capacity against noise, because additional parameters introduce more gates and can offset accuracy gains.In the MNIST motivation, measured accuracy peaks at 45 parameters as accumulated noise offsets the capacity benefit.
  • Variational Quantum Circuits: Variational circuits use parameterized gates whose trainable parameters are optimized for computational tasks such as QML and VQE.QML circuits encode inputs, process them with trainable gates, and measure qubits; VQE circuits prepare states for molecular-energy estimation.
  • Noise-Adaptive Search: Evolutionary search mutates and recombines SubCircuit–qubit-mapping pairs, while later pruning removes redundant gates before deployment.The broader workflow evaluates candidates, searches for the best pair, trains the selected SubCircuit, prunes gates, finetunes, and compiles for real devices.
  • SuperCircuit Construction and Training: SuperCircuit represents a large gate design space and trains sampled SubCircuits by updating only their selected parameters.A SubCircuit can vary in its number of blocks and gates, while the SuperCircuit contains the full gate set.
  • Efficient Search: Inherited SuperCircuit parameters provide relative SubCircuit performance estimates without fully training every candidate, reducing search cost.Relative performance is sufficient for derivative-free evolutionary search, and the reported reduction is around N_device × N_search, or 16,000 in the stated setting.
  • SuperCircuit Construction and Training: Front sampling and restricted sampling address sampling variance by improving parameter sharing and limiting changes between successive SubCircuits.Restricted sampling constrains the maximum number of differing layers; the illustrated example limits differences to 3 layers instead of 5.

C. Noise-Adaptive Evolutionary Co-Search

QuantumNAS jointly searches circuit structures and qubit mappings using evolutionary optimization with noise-aware performance estimation. The estimator uses inherited SuperCircuit parameters and backend noise approximations to rank candidates efficiently.

  • The genetic search encodes circuit width, depth, and logical-to-physical qubit mapping in a joint candidate representation.
  • Each iteration evaluates candidate pairs, selects high-performing parents, and generates new populations through mutation and crossover.
  • The performance estimator inherits SuperCircuit parameters and evaluates candidates using noise-aware simulation or real quantum hardware.
  • The estimator switches between a more accurate slower method and a faster less accurate method, using the former for circuits with ≤10 qubits.
  • The estimator’s two approximations retain useful relative-ranking reliability, with average Spearman correlation 0.75 across five tasks and two design spaces.

D. Iterative Quantum Pruning

QuantumNAS applies iterative fine-grained pruning to remove redundant parameters and gates after circuit search. It finetunes the remaining parameters and preserves noise-free performance while reducing compiled noise sources.

  • Iterative pruning targets redundant parameters and gates left by evolutionary search or unnecessary among similarly performing parameter sets.
  • The method ranks normalized rotation angles and removes those closest to 0°, then finetunes the remaining parameters to recover accuracy.
  • Pruning ratios increase iteratively until the desired final ratio is reached, using polynomial pruning-ratio decay.
  • The final ratio preserves noise-free simulation performance relative to the unpruned circuit, while fewer compiled gates can increase accuracy by up to 9%.

E. TorchQuantum Library

TorchQuantum is a PyTorch-based library for constructing, training, simulating, and deploying parameterized quantum circuits. Its tensorized processing and native automatic differentiation support substantially faster training than PennyLane in the described benchmark.

  • TorchQuantum implements quantum states and gate matrices with native torch.Tensor objects for differentiable circuit simulation.
  • The library supports common quantum gates, PyTorch-style APIs, IBM Qiskit conversion, ready-to-use templates, and parameter-shift gradients.
  • TorchQuantum uses tensorized CPU/GPU batch processing, avoiding PennyLane’s linear training-speed reduction with batch size.
  • 246 to 104 times faster training was measured for 10-qubit circuits containing 100 RX and 100 CRY gates versus PennyLane.
  • The evaluation covers 6 QML and 6 VQE tasks across IBM quantum devices, with reported experiments using device-specific configurations and baselines.
  • QuantumNAS achieves the highest accuracy on real QC devices in Figure 13, with pruning improving accuracy by 2% on average.
  • On four 5-qubit real QC devices, QuantumNAS designs outperform baseline designs with higher measured accuracy.

B. Experimental Results

QuantumNAS consistently improves noisy-device performance across QML and VQE evaluations, including device-specific and larger-scale settings. Its circuits remain effective when searched or trained with real quantum hardware.

  • QML results: Over 85% 4-class and 95% 2-class accuracy was achieved on IBMQ-Yorktown across five QML tasks, generally outperforming baselines.Pruning improved average accuracy by 2% for 4-class and 1% for 2-class tasks.
  • QML results: 3% accuracy improvement accompanied pruning that removed 14 parameters and reduced circuit depth by 11.The pruned circuit retained similar noise-free accuracy while using fewer gates and experiencing less noise.
  • Device variation: 5% better average accuracy was obtained on IBMQ-Santiago than baselines despite its lower noise.Performance remained much higher than baselines three weeks after search, supporting device-specific noise adaptation.
  • Device variation: Best performance occurred when circuits were searched and run on the same device, demonstrating the value of device-specific circuits.QuantumNAS used noise models tailored to each target device.
  • Scalability: QuantumNAS achieved over 5% better accuracy for a 21-qubit model and supports training, search, and evaluation on quantum machines.Real-QC training with parameter shift produced accuracy comparable to classical-simulator training, while runtime increased approximately linearly with qubit count.
  • VQE results: QuantumNAS consistently produced lower H2 expectation values than baselines, while pruning 50% of parameters steadily reduced eigenvalues.The theoretical optimal value for H2 is -1.85, and UCCSD was far from optimal under hardware noise.

C. Performance Analysis

Performance analyses show that co-search, evolutionary sampling, and pruning jointly balance circuit capacity against noise. The strongest results come from sufficiently broad design spaces and task- and device-aware choices.

  • Search components: 9% average accuracy gain came from co-designing circuit structure and qubit mapping, exceeding either noise-adaptive search alone.Circuit search alone improved accuracy more than mapping search alone because its design space is larger.
  • Search components: 12% higher final accuracy was achieved with front and restricted sampling than with random sampling.These strategies control differences between consecutive samples, stabilizing SuperCircuit training and improving relative-performance estimates.
  • Hardware effects: Lower device error rates improve performance under the same topology, while topology and mapping also affect results.These observations motivate hardware-aware design and mapping choices.
  • Design space: Larger design spaces can outperform shallow circuits because QuantumNAS balances learning capacity against noise.Small U3+CU3 spaces had consistently worse accuracy despite lower circuit noise.
  • Search components: Evolutionary search found lower-loss SubCircuit–mapping pairs than random search, whose best performance quickly saturated.The lower-loss pairs delivered higher accuracy.
  • Pruning: A pruning sweet spot maximizes accuracy by balancing gate-error reduction against reduced circuit capacity.Circuits with more parameters generally tolerate larger pruning ratios.

VI. CONCLUSION

QuantumNAS combines noise-adaptive variational-circuit and qubit-mapping co-search with SuperCircuit-based exploration and iterative pruning. Experiments support improved robustness across QML and VQE tasks, while the released TorchQuantum library supports further research.

  • Conclusion: QuantumNAS is a noise-adaptive co-search framework for robust variational circuits and qubit mappings.It uses SuperCircuit-based search to explore a broad design space and iterative pruning to remove redundant gates.
  • Conclusion: Experiments on QML and VQE tasks demonstrate higher robustness and performance than baseline designs.The conclusion positions the framework as applicable across both machine-learning and quantum-simulation workloads.
  • Conclusion: TorchQuantum is open-source infrastructure for fast training of parameterized quantum circuits and future variational-algorithm research.Source data is also planned for release to facilitate further research.
  • Outlook: Future directions include noise-adaptive quantum feature-map design, barren-plateau mitigation, and chemistry applications of ansatz search.These directions are presented as opportunities for further theoretical and experimental exploration.

APPENDIX

QuantumNAS provides a noise-adaptive circuit-and-mapping search pipeline, supported by TorchQuantum and artifact materials for training, evaluation, and reproduction. The appendix documents benchmarks, metrics, runtime conditions, and open-source availability.

  • QuantumNAS searches for a robust parameterized-circuit architecture and corresponding qubit mapping with noise information in the loop.
  • TorchQuantum supports parameterized quantum-circuit research, while the QuantumNAS pipeline covers SuperCircuit training, search, SubCircuit training, pruning, and real-hardware evaluation.
  • Benchmarks: The artifact includes QNN image classification, vowel recognition, and VQE benchmarks using automatically searched and human-designed models.
  • Requirements and availability: The artifact is publicly available under the MIT License and uses Python, PyTorch, Qiskit, Intel CPUs, Nvidia GPUs, and remote IBMQ access.
  • Runtime conditions: Real-machine noise changes constantly, and one 4-qubit circuit search and test takes approximately 5 to 10 CPU hours or 5 to 10 times less on one Nvidia GPU.
  • Metrics: Classification uses top1 accuracy, whereas VQE uses molecule ground-state expectation value, with higher classification accuracy and lower VQE energy preferred.

C. Description

The artifact is available through the TorchQuantum repository and requires hardware, software, and benchmark datasets for reproducing QuantumNAS experiments.

  • Access and hardware: The artifact is available at the TorchQuantum GitHub repository, with remote IBMQ access required and a GPU highly recommended for faster training.
  • Software: The software dependencies include Python, PyTorch, Qiskit, and the packages listed in requirements.txt.
  • Datasets and benchmarks: Datasets include Vowel recognition, MNIST, and FashionMNIST, while VQE benchmarks cover various molecules.
  • Models and baselines: QNN models are searched SubCircuits, with human design, random search, and noise-unaware searched models used as baselines.

D. Installation

Installation and execution proceed through the TorchQuantum package and documented scripts for QuantumNAS and human-baseline workflows.

  • Installation: The repository can be installed locally with `pip install --editable .`, with detailed instructions provided in README.md.
  • TorchQuantum example: The artifact provides examples for constructing, training, and deploying a simple MNIST QNN through `./example1/1_train_qnn.sh`.
  • QuantumNAS pipeline: The QuantumNAS workflow runs SuperCircuit training, co-search, SubCircuit training, pruning, and evaluation through five documented scripts.
  • Baselines: Separate scripts train and evaluate human baseline designs, and additional scripts cover datasets, design spaces, quantum machines, and other baselines.

F. Evaluation and expected results

Evaluation checks circuit training behavior and compares QuantumNAS with baselines on real quantum machines, while the artifact’s reproducibility is limited by retired hardware.

  • Training functionality: Successful circuit training should produce decreasing classification loss and increasing training accuracy.
  • Expected results: QuantumNAS is expected to achieve better classification accuracy and lower molecule ground-state energy expectation values than baseline models.
  • General framework: TorchQuantum supports constructing, training, and deploying customized quantum-circuit architectures from scratch.
  • Reproducibility boundary: Several IBMQ machines used for the paper’s real-device results, including Yorktown, Athens, and Melbourne, have already retired.
Loading 2107.10845v5…