⚛️ quantum physics

Benchmarking non-Clifford gates using only Pauli twirling group

This paper introduces Pauli Transfer Character Benchmarking, a protocol that enables the robust estimation of non-Clifford gate fidelities using only local Pauli operations, thereby overcoming the limitations of existing randomized benchmarking methods that struggle with non-Clifford gates.

Original authors: Han Ye, Guoding Liu, Xiongfeng Ma

Published 2026-02-02

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Han Ye, Guoding Liu, Xiongfeng Ma

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to tune a very delicate, high-tech musical instrument (a quantum computer). You want to know if a specific note (a quantum gate) is being played perfectly. However, there's a problem: every time you try to test the note, your own hands are shaking (preparation errors), and your ears are slightly muffled (measurement errors). These "shakes and muffs" make it impossible to tell if the instrument is out of tune or if you are just doing a bad job testing it.

For the easy notes (called Clifford gates), scientists have a clever trick called "Randomized Benchmarking." They play a long, random sequence of easy notes before and after the test note. This "twirls" the noise, smoothing it out so the shakes and muffs cancel themselves out, revealing the true quality of the test note.

But here is the catch: This trick works great for easy notes, but it completely fails for the "hard" notes (called non-Clifford gates). These hard notes are essential for the computer to do complex math, but they are too complicated for the old "twirling" trick to handle without using massive, error-prone equipment.

The New Solution: "Pauli Transfer Character Benchmarking" (PTCB)

The authors of this paper, Han Ye, Guoding Liu, and Xiongfeng Ma, have invented a new way to test these hard notes using only the simplest tools available (local Pauli operations). They call their method Pauli Transfer Character Benchmarking (PTCB).

Here is how their new method works, using a simple analogy:

1. The Problem with the "Hard" Notes

Think of the hard notes as a secret code that gets scrambled by noise. The old method tried to scramble the noise directly, but it couldn't do that without breaking the code.

2. The Magic Mirror Trick

The authors' solution is like using a magic mirror.

Instead of trying to fix the noise directly, they place a special "mirror" (a Clifford gate) in front of the hard note.
This mirror reflects the hard note in a specific way that turns the "scrambled" parts of the signal into a straight line (a diagonal line in math terms).
Crucially, they use a virtual mirror pair: they imagine a mirror and its reflection (a gate and its inverse) working together. Because they are virtual, they don't actually need to build a complex machine to create them; they just arrange the simple "Pauli" tools (the basic building blocks) to act like the mirror pair.

3. The "Twirl" Without the Mess

By using this virtual mirror setup, they can "twirl" the noise just like the old method, but they only use the simple, high-quality tools that the computer already has. This allows them to isolate the specific "fingerprint" of the hard note's performance, ignoring the shaking hands and muffled ears.

4. The Result

They tested this idea on a specific hard note called the Toffoli gate (a three-qubit gate often used in complex calculations).

They simulated a noisy environment where the computer was making mistakes.
They ran their new PTCB protocol.
The Result: The method successfully estimated the "fidelity" (how good the note is) without being fooled by the preparation or measurement errors. It proved that you can test these complex, hard-to-reach notes using only simple, local tools.

Why This Matters (According to the Paper)

The paper claims this is a breakthrough because:

It solves a dead end: Previously, people thought you couldn't test these hard notes without using complex, error-prone multi-qubit tools. This paper shows you can do it with simple tools.
It's robust: It ignores the "shaking hands" (SPAM errors) that usually ruin these tests.
It's practical: It relies on tools (Pauli gates) that current quantum computers already do very well.

In short, the authors found a way to use a "virtual mirror" to make the noise disappear, allowing us to finally get a clear look at how well quantum computers are playing their most difficult notes, using only the simplest instruments available.

Technical Summary: Benchmarking Non-Clifford Gates Using Only the Pauli Twirling Group

Problem Statement
Accurate benchmarking of quantum processes is essential for developing fault-tolerant quantum computation. However, standard techniques like Randomized Benchmarking (RB) are often confounded by State Preparation and Measurement (SPAM) errors. While RB addresses this by employing group twirling to regularize noise channels, practical implementations favor "local" twirling gates (e.g., Pauli or single-qubit Clifford gates) due to their high fidelity and experimental feasibility. A significant gap exists in this framework: while local twirling effectively benchmarks Clifford gates, it faces fundamental limitations when applied to generic multi-qubit non-Clifford gates. Existing methods for non-Clifford gates typically require complex multi-qubit twirling groups, which violate the gate-independent noise assumption and incur substantial compilation overhead. Furthermore, it has been shown that local twirling gates cannot effectively twirl the noise channels of certain multi-qubit controlled-phase gates. Consequently, there is a lack of a robust, SPAM-error-free protocol for characterizing non-Clifford gates using only local operations.

Methodology: Pauli Transfer Character Benchmarking (PTCB)
To address this, the authors propose the Pauli Transfer Character Benchmarking (PTCB) protocol. The core innovation lies in estimating the product of two symmetric elements of the Pauli Transfer Matrix (PTM), specifically $\tilde{U}_{PQ} \tilde{U}_{QP}$ , for a noisy channel $\tilde{U}$ , using exclusively the Pauli twirling group and Pauli-basis SPAM.

The protocol operates on the following principles:

Virtual Conjugation: Instead of directly twirling the noise channel, the protocol constructs a "twirled" channel by combining the target gate $U$ with a Clifford gate $C$ . By selecting $C$ such that $C P C^\dagger = \pm Q$ (where $P$ and $Q$ are Pauli operators), the off-diagonal PTM element $\tilde{U}_{PQ}$ is effectively moved to a diagonal position.
Virtual Clifford Pair: To ensure all operations remain local (Pauli), the protocol employs a virtual conjugated Clifford pair ( $C$ and $C^\dagger$ ). In the circuit implementation, the sequence involves the target gate $\tilde{U}$ , followed by $C^\dagger$ , a Pauli operation, and $C$ . Crucially, the authors demonstrate that the sequence $C^\dagger P_2 P_1^\dagger C$ can be compiled into a single Pauli gate. This allows the protocol to access off-diagonal PTM elements without physically implementing the Clifford gates, thereby avoiding the noise and overhead associated with multi-qubit Clifford compilation.
SPAM Robustness: The protocol utilizes group twirling over the Pauli group. By averaging over random Pauli sequences and applying specific projectors (weighted sums of Pauli gates), the method isolates the term $(\tilde{U}_{PQ} \tilde{U}_{QP})^m$ . The survival probability is then fitted exponentially to extract the product of the PTM elements, rendering the estimation robust to SPAM errors.
Fidelity Estimation: For unitary gates satisfying $U^2 = I$ (self-adjoint), the protocol estimates the process fidelity. The authors derive a lower bound for the fidelity based on the geometric mean of the estimated PTM products. They note that for noise channels sufficiently close to the identity, this lower bound is a tight approximation. The protocol is also extended to non-self-adjoint gates by incorporating the inverse gate $U^\dagger$ into the circuit structure.

Key Contributions

PTCB Protocol: Introduction of a novel benchmarking method that estimates PTM element products using only local Pauli operations, bypassing the need for multi-qubit twirling groups.
Non-Clifford Benchmarking: Development of a specific fidelity benchmarking scheme for non-Clifford gates (specifically those satisfying $U^2=I$ ) that is robust to SPAM errors.
Virtual Implementation: The use of a virtual conjugated Clifford pair to access off-diagonal PTM elements while maintaining the locality of all physical operations.
Theoretical Framework: A demonstration that local twirling can be extended to characterize non-Clifford gates by constructing the twirled channel as a combination of the target gate and a Clifford gate, rather than twirling the noise directly.

Results
The authors validate the protocol through numerical simulations using the Toffoli gate as a concrete example. The noise model includes local dephasing, local amplitude damping, and unitary noise (modeled as controlled-rotation errors on CNOT gates), with infidelities ranging from 0.01 to 0.04.

Accuracy: The discrepancy between the actual process infidelity and the estimated infidelity (using the derived lower bound) was found to be negligible, not exceeding $1 \times 10^{-4}$ across all simulated noise channels.
SPAM Robustness: Simulations confirmed that the protocol remains robust against SPAM errors, with the estimation converging to the actual value regardless of the SPAM error rate.
Sampling Efficiency: The study analyzed the required sample sizes for "outer sampling" (selecting $(P,Q)$ pairs) and "inner sampling" (selecting Pauli sequences). It was determined that a moderate number of outer samples ( $M=30$ ) and inner samples ( $M'$ ) suffices for convergence, and the number of circuit repetitions has minimal impact on the estimation accuracy.

Significance and Claims
The paper claims to provide a theoretical pathway for benchmarking non-Clifford gates using exclusively local operations. This is presented as a significant step toward fault-tolerant quantum computing, as non-Clifford gates are indispensable for universal quantum computation yet difficult to characterize with current local-twirling methods. The authors emphasize that their method solves the challenge of characterizing multi-qubit non-Clifford gates without the overhead of multi-qubit twirling groups.

The authors remain modest regarding the immediate scalability and variance of the method. They acknowledge that for multi-qubit gates with exponentially decaying PTM terms (like the controlled-X gate), the current framework may require large sampling numbers to achieve convergence. Additionally, they note that the current protocol exhibits non-negligible variance in fidelity estimation for Toffoli gates, suggesting room for optimization in near-term applications. Future work is suggested to explore alternative twirling groups (like the local Clifford group) or alternative fidelity metrics to further reduce sampling overhead.