⚛️ quantum physics

Protocol for Purifying Noisy Preparation and Measurements of Qubits

This paper presents a protocol that effectively suppresses noisy state preparation and measurement (SPAM) errors in qubits to arbitrarily low levels by repeating noisy operations and utilizing ancillas, thereby enabling high-fidelity quantum information tasks in computing and communication.

Original authors: Jaemin Kim, Seungchan Seo, Jiyoung Yun, Benjamin Lienhard, Joonwoo Bae

Published 2026-06-30

📖 4 min read🧠 Deep dive

CC0 1.0

Original authors: Jaemin Kim, Seungchan Seo, Jiyoung Yun, Benjamin Lienhard, Joonwoo Bae

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 send a secret message using a walkie-talkie, but the device is old and glitchy. Sometimes, when you try to say "Start," the static makes it sound like "Stop." Sometimes, when you try to listen for "Stop," the static makes it sound like "Start." In the world of quantum computers, these glitches are called SPAM errors (State Preparation and Measurement errors). They happen when the computer tries to set up a qubit (the basic unit of information) or when it tries to read the result.

This paper introduces a clever "noise-cancelling" protocol that fixes these glitches without needing a brand-new, perfect machine. Here is how it works, using simple analogies:

The Problem: The "Noisy Room"

Think of a quantum computer as a room where people are trying to whisper a single word: "Zero."

Preparation Noise: When the person tries to whisper "Zero," their voice cracks, and sometimes it sounds like "One."
Measurement Noise: When the listener tries to hear the word, their ears are clogged with static, so they might hear "One" even when the speaker said "Zero."

In current technology, these errors happen about 5% to 10% of the time. This is too messy for complex tasks like quantum error correction or building a quantum internet.

The Solution: The "Voting System"

The authors propose a protocol that acts like a voting system to filter out the noise. Instead of relying on one person to say "Zero," you bring in a few friends (called ancilla qubits) to help.

The Setup: You have your main person (the system qubit) and a few friends (the ancilla qubits). Everyone is in a noisy room, so everyone is prone to making mistakes.
The Connection: You link them all together with a special "handshake" (a quantum gate called CNOT). This handshake ensures that if the main person is "Zero," the friends should also be "Zero."
The Vote: You ask all the friends to shout out what they hear.
- If everyone shouts "Zero," you accept the answer. You are now very confident that the main person was indeed "Zero," because it is highly unlikely that everyone made the exact same mistake at the same time.
- If even one friend shouts "One," you ignore the whole group and try again.

The Magic: Turning "Bad" into "Good"

The paper shows that even if your friends are just as noisy as the main person, this voting system works wonders.

With 2 friends: You can reduce the error rate from 5% down to 0.1% (1 in 1,000).
With 4 friends: You can reduce the error rate to 0.0001% (1 in 1,000,000).

It's like having a noisy microphone, but if you record the same speech with four other noisy microphones and only keep the recording where all five agree, the final result sounds crystal clear.

Real-World Application: Using What You Have

The authors didn't just do this on paper; they showed how to do it on superconducting quantum processors (the kind used by companies like IBM and Google).

The Trick: These processors already have little switches called "tunable couplers" that connect qubits. Usually, these are just used to link qubits together.
The Innovation: The authors suggest temporarily turning these switches into the "friends" (ancilla qubits) needed for the voting system. This means you don't need to build new hardware; you just use the parts you already have in a smarter way.

Why This Matters

The paper claims this method can make quantum computers much more reliable for specific tasks:

Quantum Networks: It helps in "entanglement distillation" (cleaning up shared quantum connections) and "entanglement swapping" (linking distant quantum computers), even if the equipment is noisy.
Efficiency: It achieves these high levels of cleanliness using very few extra resources (just a couple of extra qubits).

The Catch

There are two main limitations mentioned:

The "Try Again" Factor: Because the system only accepts the result if everyone agrees, you might have to run the experiment several times to get a successful "vote." However, the paper notes that the success rate is still high (over 75% in realistic scenarios).
The Gate Limit: The "handshake" (the CNOT gate) used to link the qubits must be reasonably good. If the handshake itself is too broken (too noisy), the whole voting system fails. The paper calculates a specific threshold: as long as the handshake isn't too bad, the system works.

In summary: This paper presents a practical "software fix" for noisy quantum hardware. By using a few extra qubits as a voting committee, it can filter out static and make the computer's inputs and outputs almost perfect, using resources that are already available in today's machines.

Technical Summary: Protocol for Purifying Noisy Preparation and Measurements of Qubits

Problem Statement
Noise inherent in qubit state preparation and measurements (SPAM) constitutes a significant source of error in quantum information processing. This is particularly critical for near-term hybrid quantum-classical algorithms (e.g., variational quantum algorithms), quantum error correction, and entanglement distribution in quantum networks. While previous efforts have addressed measurement noise through information-theoretic analysis or active error mitigation, many existing proposals rely on the assumption of noiseless ancillary qubits. It remains unclear whether noisy preparations of ancillary qubits can be utilized to suppress readout errors, and a cost-effective protocol using a minimal number of ancillas is desired.

Methodology
The authors propose a practical protocol to purify noisy SPAM using only noisy SPAM resources and a set of auxiliary qubits. The protocol operates under the assumption that the initial noise parameters satisfy specific conditions: the preparation fidelity $f > 1/2$ and the measurement noise fraction $q < 1/2$ (performing better than random guessing).

The protocol consists of two main components:

Purification of State Preparation: To improve the fidelity of a noisy initial state $\rho$ , the protocol employs $n$ ancillary qubits initialized in the same noisy state. A collective CNOT operation (implemented as a sequence of standard two-qubit CNOT gates) is applied between the system qubit (control) and the $n$ ancillas (targets). Subsequent noisy measurements are performed on the $n$ target qubits. The system state is accepted (post-selected) only if all measurement outcomes are zero ( $0^n$ ). The authors demonstrate that the fidelity of the post-selected state, $f^{(n)}$ , converges to 1 as $n$ increases.
Purification of Measurements: To purify a noisy measurement described by POVM elements $\tilde{M}_k$ , the protocol initializes $m$ ancillary qubits in the noisy state $\rho^{\otimes m}$ . A collective CNOT is applied to the system and the ancillas. Measurements are performed on all $m+1$ qubits. The outcome is accepted only if all results are identical (either $0^{m+1}$ or $1^{m+1}$ ). The resulting effective POVM element, $\tilde{N}^{(m)}_k$ , exhibits a noise fraction $q^{(m)}$ that converges to 0 as $m$ increases.

The authors also analyze the protocol's robustness when the CNOT gates themselves are noisy (with error rate $\epsilon$ ). They derive a "purification condition" showing that purification is only possible if the initial fidelity improvement in the first iteration is positive, establishing a threshold for the CNOT error rate $\epsilon_c$ .

Key Contributions and Results

Arbitrary Suppression of Errors: The paper proves that under realistic assumptions ( $f > 1/2, q < 1/2$ ), the protocol can suppress SPAM errors to arbitrarily low levels using a finite number of ancillas.
Quantitative Performance: In a realistic scenario where preparation and measurement error rates are approximately 5% ( $1-f = q = 0.05$ $1 - f = q = 0.05$ ):
- Using two ancillas, the error rate can be suppressed to approximately $10^{-3}$ .
- Using four ancillas, the error rate can be suppressed to approximately $10^{-6}$ .
- Success probabilities for these configurations remain high (e.g., $>0.77$ for two ancillas).
Robustness to Noisy Gates: Even with imperfect CNOT gates, the protocol can improve fidelity and reduce measurement noise, provided the CNOT error rate $\epsilon$ is below a critical threshold $\epsilon_c$ . For balanced SPAM errors of 5%, the threshold is approximately $\epsilon_c \approx 0.146$ . If $\epsilon$ exceeds this, purification fails.
Experimental Feasibility: The authors propose an implementation for superconducting quantum processors. They suggest repurposing existing tunable couplers (typically used to minimize crosstalk) as the required ancillary qubits. This approach leverages existing hardware resources without requiring new components, utilizing the couplers' ability to act as qubits for conditional phase accumulation and readout.
Application to Quantum Networks: The protocol is applied to two key tasks:
- Entanglement Distillation: The protocol enables the distillation of entangled bits (ebits) from weakly entangled states even when local measurements are noisy, a task previously shown to be impossible with noisy measurements alone.
- Entanglement Swapping: The protocol improves the fidelity of entanglement swapping in quantum repeaters, allowing for the creation of high-fidelity ebits between distant parties despite noisy local measurements.

Significance and Claims
The authors claim that this work provides a general, resource-efficient approach to mitigating SPAM errors, which are a primary obstacle to reliable quantum algorithms and error correction. By demonstrating that noisy ancillas can be used to purify noisy SPAM, the protocol removes the requirement for perfect auxiliary resources. The authors emphasize that the protocol is immediately cost-effective, requiring only a few ancillary qubits to reach error rates ( $10^{-3}$ ) suitable for fault-tolerant quantum computing and quantum communication tasks. The proposed use of tunable couplers in superconducting processors highlights the protocol's compatibility with state-of-the-art hardware, enabling effectively noiseless quantum networks despite inherently noisy SPAM.