⚛️ quantum physics

Error Mitigation by Zero-Noise Extrapolation (ZNE) in Indirect-Control System

This paper proposes a Zero-Noise Extrapolation (ZNE) error mitigation strategy for Variational Quantum Eigensolvers (VQE) within the indirect-control paradigm by adapting unitary folding techniques to time-evolution operators and leveraging the symmetry of one-dimensional XY spin chains to overcome the challenges of noise amplification in continuous-time quantum systems.

Original authors: Arijit Das, Masaki Owari

Published 2026-10-12

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Arijit Das, Masaki Owari

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

In the quest to build machines that can solve problems beyond the reach of today's computers, scientists are racing to harness the strange rules of quantum mechanics. These future devices, known as quantum computers, rely on tiny units of information called qubits. Unlike the bits in a standard laptop, which are either zero or one, qubits can exist in a delicate superposition of both states simultaneously. This property allows them to explore vast numbers of possibilities at once. However, this power comes with a severe weakness: qubits are incredibly fragile. The slightest interaction with their environment, such as a fluctuating temperature or a stray electromagnetic wave, causes them to lose their quantum state and introduce errors. For decades, the prevailing belief was that to build a useful machine, we would need to correct these errors in real-time using complex codes that require thousands of physical qubits just to create one stable, error-free "logical" qubit. This path, known as fault-tolerant quantum computing, remains a distant goal.

In the meantime, researchers are exploring ways to make the noisy, imperfect machines we have today useful. One promising strategy is to change how we control the machine itself. Instead of reaching out to touch every single qubit individually, which invites more noise, a newer approach called "indirect control" suggests we should only manipulate a small handful of qubits at the edge of the system. The rest of the machine is left to evolve naturally according to its own internal laws. This method reduces the number of entry points for noise and simplifies the hardware. Yet, even with this cleaner architecture, errors still creep in. The challenge for scientists is to find a way to clean up the results after the calculation is done, without needing the impossible task of fixing the machine while it runs. This is where a technique called zero-noise extrapolation comes in, a method that tries to guess what the answer would have been if the machine were perfect, by comparing results from machines that are intentionally made noisier.

A team of researchers at Shizuoka University in Japan has now taken a significant step in testing whether this error-cleaning technique can work within the indirect-control framework. Their work focuses on a specific type of quantum system known as a one-dimensional spin chain, which can be visualized as a line of tiny magnets connected to their neighbors. In this setup, the researchers can only apply direct commands to the first two magnets in the line. The rest of the chain is "inaccessible," meaning no one can touch them directly; they only move because the influence of the first two magnets ripples through the line over time. The researchers wanted to use this setup to solve a difficult math problem related to finding the lowest energy state of a system, a task known as the Variational Quantum Eigensolver.

The central obstacle they faced was how to apply the error-cleaning technique. To guess the perfect answer, the method requires running the same calculation multiple times with different levels of noise. In standard quantum computers, scientists create this extra noise by inserting extra, redundant steps into the circuit that cancel each other out in theory but amplify errors in practice. However, in the indirect-control system, the calculation relies on the natural flow of time through the chain. Reversing this flow to create the redundant steps is usually impossible because the internal laws of the chain do not allow for a simple backward motion. The researchers realized that if they could find a way to reverse the flow, they could create the necessary noise amplification. They discovered that by adding a very specific, limited ability to flip the state of every other magnet in the inaccessible part of the chain, they could effectively reverse the time evolution. This small relaxation of the strict "no-touch" rule allowed them to construct the redundant steps needed for the error-cleaning process.

To test their idea, the team ran detailed computer simulations of a seven-qubit and an eight-qubit system. They programmed the machine to solve for the lowest energy state of a specific magnetic model, first running it with a natural level of noise, and then running it again with the noise artificially boosted using their new method. By comparing the results from the noisy runs and mathematically projecting them back to a zero-noise state, they were able to recover an answer that was remarkably close to the true, error-free value. Their simulations showed that even when the noise was significant, this technique could pull the result back to within a few percent of the correct answer. The success was consistent across different levels of noise and different sizes of the system, suggesting that the method is robust.

However, the researchers are careful to note that their success relies heavily on the specific symmetries of the magnetic chain they studied. The trick of reversing time by flipping every other magnet works because of the unique mathematical structure of that particular system. They argue that this approach cannot be easily copied for other types of quantum systems that do not share these specific symmetries, such as the more complex Heisenberg model. For those more general systems, the method of reversing time to create noise amplification may not be possible, and scientists will need to find entirely different ways to clean up errors in indirect-control machines.

Despite this limitation, the study provides a crucial proof of concept. It demonstrates that the indirect-control architecture, which offers a simpler and potentially more scalable path to building quantum computers, is not incompatible with advanced error correction. By showing that it is possible to manipulate the noise levels in a controlled way within this restricted environment, the researchers have opened a door for future experiments. Their work suggests that if we can build these indirect-control systems in the lab, we will not be helpless against the noise that plagues them; we will have a mathematical tool to see through the distortion and find the truth hidden beneath. The path forward involves testing these ideas on real hardware and exploring whether similar tricks can be found for other, more complex quantum systems.

Technical Summary: Error Mitigation by Zero-Noise Extrapolation (ZNE) in Indirect-Control Systems

Problem Statement
Indirect-control quantum computing offers a scalable architecture where external operations are restricted to a small subset of qubits (the control unit), while the rest of the system evolves under its intrinsic many-body Hamiltonian. This paradigm reduces hardware complexity and limits noise injection from control channels. While recent work (Anan et al.) demonstrated the feasibility of implementing the Variational Quantum Eigensolver (VQE) within this framework, the integration of Quantum Error Mitigation (QEM) techniques remains unexplored. Specifically, Zero-Noise Extrapolation (ZNE), a leading QEM strategy, relies on the controlled amplification of noise, typically achieved via "unitary folding" (inserting $G^\dagger G$ sequences). In indirect-control systems, the primary computational resource is continuous time-evolution under the system Hamiltonian. Reversing this evolution to construct the necessary inverse operations ( $U^\dagger$ ) is non-trivial, as indirect-control systems generally lack time-reversibility through local operations on the control unit alone. Consequently, standard ZNE techniques cannot be directly applied.

Methodology
The authors investigate the feasibility of implementing ZNE within the indirect-control VQE framework using a one-dimensional XY spin chain. The study proceeds through the following methodological steps:

System Definition: The system is modeled as a 1D XY spin chain where the first two qubits serve as the control unit. The ansatz circuit consists of instantaneous unitary operations on the control unit interleaved with free time-evolution under the system Hamiltonian $H_{XY}$ .
Addressing Time-Reversibility: To implement unitary folding for the time-evolution gates ( $U(t_n, t_{n+1})$ ), the inverse gate $U^\dagger$ is required. The authors exploit a specific symmetry of the 1D XY Hamiltonian (in the absence of local magnetic fields): applying Pauli-Y gates to all odd-numbered qubits transforms the Hamiltonian such that $H_{XY} \to -H_{XY}$ . By relaxing the strict indirect-control constraint to allow the application of Y-gates on odd-numbered qubits (both accessible and inaccessible) before and after the time-evolution, the authors demonstrate that $U^\dagger$ can be effectively synthesized.
Noise Modeling: Unlike previous ZNE studies assuming global depolarizing noise, this work models noise as a local depolarizing channel acting independently on each qubit. This is deemed more physically realistic for many-body systems. The authors analyze the commutation properties of local noise with unitary folding, noting that while the simple linear scaling of noise parameters (valid for global noise) does not strictly hold for local noise, the technique remains effective.
Simulation Framework: Numerical simulations were performed using the Qulacs library. The study utilized a 7-qubit and 8-qubit system with a 30-layer indirect-control VQE ansatz. The target Hamiltonian was a transverse-field Ising model. Simulations employed the Lindblad master equation approximated via Trotter decomposition to model decoherence.
ZNE Implementation: The authors applied Richardson extrapolation to the expectation values obtained from circuits with varying noise levels. Noise levels were scaled by increasing the folding number ( $K$ ) of the time-evolution gates (and other gates in multivariate scenarios). Both single-variate and multivariate ZNE approaches were tested.

Key Contributions

Feasibility of ZNE in Indirect-Control: The paper establishes that ZNE can be implemented in indirect-control VQE by leveraging the symmetry of the 1D XY model to construct inverse time-evolution operators.
Relaxed Control Constraint: The study proposes a minimal relaxation of the indirect-control paradigm, allowing Y-gate operations on odd-numbered qubits. The authors argue this is physically realizable using spin chains with two types of qubits having distinct resonance frequencies, allowing selective addressing via global resonant pulses.
Local Noise Analysis: The work extends ZNE analysis to local depolarizing noise models, demonstrating that unitary folding remains effective even when the noise does not commute with unitary transformations in the first-order approximation.
Multivariate ZNE: The authors successfully applied multivariate Richardson extrapolation to handle heterogeneous noise sources across different gate types (single-qubit, two-qubit, and time-evolution gates).

Results

Error Mitigation Efficacy: For a 7-qubit system with a decoherence rate of $\gamma = 0.001$ , ZNE successfully recovered the noise-free ground-state energy estimates. Specifically, 3rd-order Richardson extrapolation achieved results within the statistical uncertainty of the noise-free simulation (98% relative accuracy).
Scaling and Variance: Consistent with ZNE theory, the variance of the estimator increased with the order of extrapolation. The study confirmed that while higher-order extrapolation improves bias reduction, it incurs a significant sampling overhead.
Parameter Sensitivity: The accuracy of ZNE was found to be sensitive to the range of initial time parameters ( $T_{max}$ ). When $T_{max}$ was large, the accumulated decoherence in the noisy circuits became too severe for ZNE to fully correct, leading to a divergence between noisy and noise-free estimates.
Multivariate Performance: In an 8-qubit simulation with heterogeneous noise probabilities (0.001 for single-qubit gates, 0.005 for CZ gates, and $\gamma=0.001$ for time-evolution), multivariate ZNE (orders 3–5) successfully mitigated errors, bringing estimates close to the noise-free baseline.

Significance and Claims
The paper claims that indirect-control systems provide a promising alternative architecture for NISQ devices by limiting noise injection pathways. However, the realization of this potential is contingent on effective error mitigation. The primary significance of this work is demonstrating that ZNE, a standard error mitigation technique, can be adapted to the indirect-control paradigm.

The authors explicitly note the limitations of their approach:

Symmetry Dependence: The method relies heavily on the specific symmetries of the 1D XY model. The authors state that extending this unitary folding technique to general many-body systems (e.g., the Heisenberg model) with different symmetries is difficult.
Scope of Application: The study is restricted to numerical simulations. The authors acknowledge that experimental implementation on current mainstream hardware (e.g., IBM, Google) is not feasible as these platforms do not utilize indirect-control.
Future Directions: The paper suggests that for systems without specific symmetries, alternative ZNE methods (non-unitary folding) or Probabilistic Error Cancellation (PEC) adapted to indirect-control are necessary areas for future research. It also highlights the need to identify which many-body quantum systems are most suitable for the indirect-control approach in experimental settings.

In summary, the work provides a theoretical and numerical proof-of-concept that error mitigation is viable in indirect-control VQE, provided the system Hamiltonian possesses specific symmetries that allow for the construction of inverse time-evolution operators.

More like this