Demonstration of robust and efficient quantum property learning with shallow shadows

Hong-Ye Hu, Andi Gu, Swarnadeep Majumder, Hang Ren, Yipei Zhang, Derek S. Wang, Yi‐Zhuang You, Zlatko Minev, Susanne F. Yelin, Alireza Seif

TL;DR

This paper introduces "robust shallow shadows," a new method for efficiently and accurately learning properties of quantum states, even when the quantum computer is noisy. By combining shallow quantum circuits with Bayesian inference to correct for noise, they achieve significant improvements in sample efficiency compared to existing techniques, making quantum state characterization more practical.

What Problem Does This Paper Address?

Characterizing an unknown quantum state is fundamental to quantum information processing, but it's a challenging task. "Classical shadows" offer a way to do this efficiently by using randomized measurements to create a compressed representation of the state, from which various properties can be estimated. However, standard classical shadows struggle with certain types of properties, like those involving non-local interactions or global state features (e.g., fidelity). While using shallow random quantum circuits before measurement can improve efficiency for these harder properties, real-world quantum computers are inherently noisy. This noise corrupts the measurements and leads to inaccurate results, negating the benefits of the shallow circuits. The authors tackle the problem of making classical shadows robust and efficient in the presence of realistic noise.

Methods

The core innovation is the "robust shallow shadows" protocol. It builds upon the idea of shallow random quantum circuits by incorporating two key elements:

  • Shallow Random Circuits: Instead of just random Pauli measurements, they prepend a shallow, randomly chosen quantum circuit (composed of twirled two-qubit gates) to the measurement process. This circuit acts as a "scrambler," making the measurement process more sensitive to a wider range of quantum state properties.
  • Bayesian Inference for Noise Mitigation: Crucially, they acknowledge that the shallow circuits themselves are subject to noise. They employ a Bayesian inference framework to learn a model of this noise from calibration data. This learned noise model is then used in post-processing to correct the measurement results, effectively "un-doing" the noise's effect. This is a significant departure from simply applying theoretical protocols without accounting for experimental imperfections.

They also leverage tensor network methods for efficient classical post-processing of the measurement data.

Datasets

Not discussed in the paper. The experiments use quantum states prepared on a superconducting quantum processor.

Key Insights and Ideas

  • Noise is the Bottleneck: The paper strongly emphasizes that for practical quantum computing, dealing with noise is paramount. Blindly applying theoretical methods designed for ideal systems will lead to biased and unreliable results.
  • Shallow Circuits are a Sweet Spot: While deeper circuits can be more powerful, they also accumulate more noise. Shallow circuits strike a balance, offering improved capability for certain observables without becoming prohibitively noisy.
  • Bayesian Inference is Powerful for Noise Characterization: The authors demonstrate that Bayesian inference provides an efficient and effective way to learn and model the complex noise present in quantum hardware, allowing for accurate error mitigation.
  • Bias-Variance Trade-off is Real: Correcting for noise (reducing bias) often comes at the cost of increased uncertainty (variance) in the estimates. The paper quantifies this trade-off and shows how their method navigates it.

Key Results

  • Accurate Property Recovery: The robust shallow shadows protocol successfully recovers various quantum state properties, including fidelity, local and non-local Pauli observables, and subsystem purities, even on a noisy superconducting quantum processor.
  • Improved Sample Complexity: For nonlocal and low-rank observables, their method reduces the required number of measurements by up to a factor of five compared to standard random Pauli measurements, especially when error mitigation is applied.
  • Noise Mitigation Effectiveness: Experimental results show that their Bayesian noise mitigation significantly improves the accuracy of predictions, bringing them closer to theoretical values and outperforming unmitigated results.
  • Optimal Depth Varies: The optimal depth of the shallow circuit depends on the observable being measured. For fidelity, deeper circuits (up to d=4) are better, while for local Pauli observables, moderate depths (d=2) offer the best trade-off between performance and noise.

Limitations and Future Work

  • Noise Model Assumption: While their Bayesian approach is shown to be robust even with some deviations from the assumed sparse Pauli-Lindblad noise model, it still relies on a specific model structure. Future work could explore more general noise models or model-agnostic approaches.
  • Platform Dependence: The experiments were conducted on an IBM superconducting quantum processor. Validating the protocol on other quantum computing platforms (trapped ions, neutral atoms, photonics) is crucial for demonstrating its generality.
  • Scalability: While the paper demonstrates the protocol on 18 qubits, extending it to larger system sizes is essential for its application to more complex problems.
  • Machine Learning Integration: The authors suggest exploring modern machine learning techniques to further enhance inference and prediction capabilities.
  • Application to Quantum Simulation: Integrating their method with tensor network representations could potentially improve the accuracy of quantum simulations through data post-processing.

Summary updated

Abstract

Extracting information efficiently from quantum systems is crucial for quantum information processing. Classical shadows enable predicting many properties of arbitrary quantum states using few measurements. While random single-qubit measurements are experimentally friendly and suitable for learning low-weight Pauli observables, they perform poorly for nonlocal observables. Introducing a shallow random quantum circuit before measurements improves sample efficiency for high-weight Pauli observables and low-rank properties. However, in practice, these circuits can be noisy and bias the measurement results. Here, we propose the robust shallow shadows, which employs Bayesian inference to learn and mitigate noise in postprocessing. We analyze noise effects on sample complexity and the optimal circuit depth. We provide theoretical guarantees for the success of error mitigation under a wide class of noise processes. Experimental validation on a superconducting quantum processor confirms the advantage of our method, even in the presence of realistic noise, over single-qubit measurements for predicting diverse state properties, such as fidelity and entanglement entropy. Our protocol thus offers a scalable, robust, and sample-efficient method for quantum state characterization on near-term quantum devices.
BibTeX
@inproceedings{hu2025demonstration,
  author = {Hong-Ye Hu and Andi Gu and Swarnadeep Majumder and Hang Ren and Yipei Zhang and Derek S. Wang and Yi‐Zhuang You and Zlatko Minev and Susanne F. Yelin and Alireza Seif},
  title = {{Demonstration of robust and efficient quantum property learning with shallow shadows}},
  booktitle = {Nature Communications},
  year = {2025},
  doi = {10.1038/s41467-025-57349-w},
}
Figure