Enhancing Quantum Expectation Values via Exponential Error Suppression and CVaR Optimization

Touheed Anwar Atif, Reuben Blake Tate, and Stephan Eidenbenz · 2025

Read Enhancing Quantum Expectation Values via Exponential Error Suppression and CVaR Optimization by Touheed Anwar Atif, Reuben Blake Tate, and Stephan Eidenbenz (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-18513.txt.

Introduction

Touheed Anwar Atif, Reuben Blake Tate, and Stephan Eidenbenz wrote this physics paper, published in 2025. It sharpens the estimates that quantum algorithms produce on noisy hardware, by pairing a purification trick with a risk-aware scoring method from finance.

Quantum expectation values, the average outcomes algorithms need, degrade under noise. Full error correction is costly, so mitigation methods matter. The authors fold in CVaR, a measure that focuses on the worst tail of outcomes, to judge how trustworthy an estimate is. They prove when one CVaR score beats another, and bound how much the purification method helps, both when its extra circuit is perfect and when it adds noise itself. Number examples show how the bounds guide real use. The result is a general way to certify better estimates on today's noisy machines.

Catalog record

CREATORTouheed Anwar Atif, Reuben Blake Tate, and Stephan Eidenbenz
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-18513

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Enhancing Quantum Expectation Values via Exponential Error Suppression and CVaR Optimization Touheed Anwar Atif, Reuben Blake Tate, and Stephan Eidenbenz Abstract Precise quantum expectation values are crucial for quantum algorithm development, but noise in real-world systems can degrade these estimations. While quantum error correction is resource-intensive, error mitigation strategies offer a practical alternative. This paper presents a framework that combines Virtual Channel Purification (VCP) technique with Conditional Value-at-Risk (CVaR) optimization to improve expectation value estimations in noisy quantum circuits. Our contributions are twofold: first, we derive conditions to compare CVaR values from different probability distributions, offering insights into the reliability of quantum estimations under noise. Second, we apply this framework to VCP, providing analytical bounds that establish its effectiveness in improving expectation values, both when the overhead VCP circuit is ideal (error-free) and when it adds additional noise. By introducing CVaR into the analysis of VCP, we offer a general noise-characterization method that guarantees improved expectation values for any quantum observable. We demonstrate the practical utility of our approach with numerical examples, highlighting how our bounds guide VCP implementation in noisy quantum systems. I Introduction The accurate computation of quantum expectation values plays a crucial role in the development and verification of a plethora of quantum algorithms [ 1 , 2 , 3 , 4 , 5 ] . However, real-world quantum systems are often plagued by various types of noise, including gate errors, measurement inaccuracies, and decoherence, which can significantly affect the accuracy of these estimations. Although quantum error correction has been proposed as a means to combat this noise by encoding logical qubits in multiple physical qubits [ 6 ] , the overhead required for error correction is too large for near-term devices [ 7 ] . A near term alternative is to employ error mitigation strategies, which improve the accuracy of quantum computations by obtaining valuable information from noisy quantum circuits without the need to physically recover a noiseless state [ 8 , 9 , 10 ] . In this paper, we introduce a framework that enhances quantum expectation value estimations by combining exponential error suppression techniques [ 11 , 12 , 13 ] with Conditional Value-at-Risk ( CVaR ) optimization [ 14 ] . Specifically, we focus on a recent error suppression method called Virtual Channel Purification (VCP), introduced by Liu et al [ 13 ] , that aims to suppress noise in quantum circuits without requiring detailed knowledge of error models, input/output states, or specific problem contexts. The authors here primarily demonstrated the technique’s effectiveness by looking at the fidelity of quantum states. However, VCP as a virtual technique is most impactful when applied to the expectation values of observables, a perspective that we explore in this work. Conditional Value-at-Risk ( CVaR ), also known as Expected Shortfall, is a risk measure that considers the tail risks of a probability distribution. In the respect of quantum algorithms, the authors in [ 15 ] proposed a modification to the classical optimization algorithm using CVaR and showed performance improvements for VQE and QAOA. In [ 16 ] , the authors demonstrated the robustness of CVaR against noise toward generating better expectation values and provided bounds on noise-free expectation values. Our contributions are twofold. First, we derive conditions on two probability distributions that enable a comparison of the CVaR values produced by samples from each distribution. This comparison is crucial for assessing the reliability of quantum expectation value estimations in the presence of noise, providing insights into which distributions leads to more robust outcomes. Second, we apply this method to the study of the VCP technique. In the idealized regime where the purification circuit, particularly the additional swap network, is error-free, and only the unitary being purified has incoherent noise, we provide analytical bounds that establish the efficacy of the VCP method in enhancing the precision of expectation values. The assumption on incoherent noise is valid in most practical scenarios, particularly with the application of techniques like Pauli-twirling [ 17 , 18 , 13 ] . When noise is present in the swap network, which introduces additional noise the purification circuit, we extend our analysis to provide inner bounds that characterize the region where the technique offers an advantage in terms of error parameters. By introducing CVaR into the analysis of VCP, as our first result for noise-free swap network we provide a provable guarantee for improving expectation values of any quantum observable diagonal in the computational basis (Theorem 3 ). In Theorem 4 , we analyze and compare different orders of VCP circuits in terms of their ability to reduce noise using CVaR . We extend this analysis to a multilayer generalization in Theorem 5 . Next, we consider the case where the swap network of the VCP circuit introduces additional noise. Here, we first provide an exact form for the effective virtual channel when a Clifford is purified employing the VCP circuit that uses a noisy swap circuit (Theorem 7 ). Its generalization to higher orders is described in Theorem 8 . Building on these theorems, our final result in Theorem 9 characterizes an inner bound for the parameter region where the VCP technique remains advantageous despite a noisy swap network. We further demonstrate the practical implications of our findings through numerical examples. The first example (Section IV-A ) examines the region defined by Theorem 9 for depolarizing noise models and provides a characterization in terms of the depolarizing parameters p and q corresponding to the swap network noise and Clifford noise, respectively. The second example (Section IV-B ) explores a scenario involving a collection of IID Clifford gates, highlighting the advantageous region in terms of the number of gates and the associated error parameters. These examples showcase how our bounds can guide the implementation of VCP in realistic quantum systems and improve its performance under noise. The paper is organized as follows: Section II presents the preliminaries on exponential error suppression and sets up the notation. Section III provides the main results focusing on the application of our results to the VCP technique, deriving bounds and characterizing the error parameters. Section IV provides numerical examples to illustrate our findings and discusses their practical implications. Finally, Section V concludes with a discussion on future directions for research and potential applications of the proposed methodology in quantum computing. The proofs are provided in the appendix. II Preliminaries and Notation II-A Virtual Channel Purification (VCP) VCP builds on the concept of exponential error mitigation, specifically the virtual state preparation technique. It uses multiple copies of a noisy quantum channel, rather than a state, to implement a purified channel, with fidelity improving exponentially with the number of copies. The technique combines ideas from the Choi–Jamiołkowski isomorphism and flag fault-tolerance to enable noise reduction without physically purifying the state or channel, relying instead on classical postprocessing and standard quantum circuits. This makes VCP an efficient, hardware-friendly solution for quantum error mitigation, particularly in tasks where accurate expectation values are crucial. Formally, we have the following. Consider a case where the ideal operation is \mathcal{U} , but in practice, the implemented operation is the noisy version \mathcal{U}_{\mathcal{E}}=\mathcal{E}\mathcal{U} , where \mathcal{E}=p_{0}\mathcal{I}+\sum_{i=1}^{4^{N}-1}p_{i}\mathcal{E}_{i}, represents the noise that distorts the ideal operation, N is the

Read the complete text