Non-Markovian Noise Suppression Simplified through Channel Representation
Zhenhuan Liu, Yunlong Xiao, Zhenyu Cai · 2024
Read Non-Markovian Noise Suppression Simplified through Channel Representation by Zhenhuan Liu, Yunlong Xiao, Zhenyu Cai (2024) 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-2412-11220.txt.
Introduction
Zhenhuan Liu, Yunlong Xiao, and Zhenyu Cai tame a sticky kind of quantum noise in a 2024 paper. Some noise keeps a memory of past states, which breaks the plain tools built for memory-free noise. Such memory noise haunts real quantum chips and dulls both error fixes and error softeners.
Their trick recasts the whole memory-bound dance as one plain noise channel, a view they call the Choi channel. In that view, old channel tools can be used as-is, then mapped back to the chip. The gains and costs of the old tools often carry over whole. Design of memory-noise care turns far simpler.
Catalog record
| CREATOR | Zhenhuan Liu, Yunlong Xiao, Zhenyu Cai |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-11220
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Non-Markovian Noise Suppression Simplified through Channel Representation Zhenhuan Liu, Yunlong Xiao, Zhenyu Cai Abstract Non-Markovian noise, arising from memory effects in the environment, poses substantial challenges to conventional quantum noise suppression protocols, including quantum error correction and mitigation. We introduce a channel representation for arbitrary non-Markovian quantum dynamics, termed the Choi channel, which translates the complex dynamics of non-Markovian noise into the familiar picture of noise channels. It therefore provides a systematic way to design non-Markovian noise suppression protocols: one can apply existing channel-level error-suppression techniques in the Choi-channel picture and then translate them back to the circuit picture. The performance of the resulting non-Markovian protocols, including their noise-suppression effects and complexity, can often be inherited directly from the corresponding Choi-channel protocols without requiring a separate analysis. With this framework, we devise new protocols using Pauli twirling, probabilistic error cancellation, and virtual channel purification. Pauli twirling reduces non-Markovian noise to noise with only classical temporal correlations; probabilistic error cancellation can fully cancel non-Markovian noise; and virtual channel purification can suppress non-Markovian noise without detailed knowledge of its specific form. Through these examples, the Choi channel serves as a foundational bridge for systematically converting existing channel-level techniques into non-Markovian noise suppression protocols. I Introduction Noise presents the most significant obstacle to the practical implementation of quantum computation. A range of techniques has been proposed for noise suppression at different levels, including dynamical decoupling [ 1 ] , quantum error correction (QEC) [ 2 , 3 ] and mitigation (QEM) [ 4 ] . Most error suppression techniques are developed based on the assumption that the noise processes occurring at different time slices are independent and can be described as independent quantum channels. Under this assumption, one can develop QEC and QEM techniques to address the noise channels at a given time slice and simply repeat the QEC and QEM procedure. However, since in practice noise is caused by continuous interaction between the environment and the quantum computer, it is no surprise that the environment can exhibit memory effects that lead to classically or even quantumly correlated noise at different time slices [ 5 , 6 , 7 ] , which are often referred to as non-Markovian noise. Other than certain specific circumstances [ 8 , 9 ] , in general non-Markovian noise can significantly affect the performance of QEC and QEM protocols [ 10 , 11 ] . Figure 1: Constructing non-Markovian noise suppression protocols from channel noise suppression protocols. A general non-Markovian noise process is first mapped, through a purely mathematical transformation, to its Choi-channel representation. In this channel picture, one can apply existing channel-level noise suppression protocols. The resulting construction is then translated back, again mathematically, to the original circuit picture, yielding an explicit non-Markovian noise suppression protocol. The solid arrows denote physical noise suppression steps, while the dotted arrows denote mathematical transformations between the non-Markovian process and the Choi-channel picture. Attempts have been made to deal such noise in several noise-suppression settings, including hardware-level techniques [ 1 , 12 , 13 , 14 ] , QEC [ 15 , 16 ] , and QEM [ 17 ] . However, these studies typically either rely on specific assumptions about the system-environment interaction or treat non-Markovian effects at an abstract theoretical level without yielding specific protocols. The primary challenge in analysing general non-Markovian noise lies in its complex nature, which requires higher-order quantum operations for a complete description. These include frameworks such as quantum strategies [ 18 ] , quantum combs [ 19 , 20 ] , process tensors [ 21 , 22 ] , or quantum circuit fragments [ 23 , 24 ] . In contrast, most existing error suppression techniques are designed for noise modelled as quantum channels [ 25 , 26 , 27 ] , making them unsuitable for direct application to non-Markovian scenarios [ 28 , 29 , 30 , 31 ] . In this work, we present a simple yet surprisingly general and effective solution: a mapping between non-Markovian quantum dynamics and quantum channels, which we call the Choi channel representation. While the Choi channel can in principle be further converted to the standard Choi state representation of a non-Markovian process [ 19 ] , there is no reason to do that since our channel representation is much more natural for studying noise-suppression protocols, given that these protocols are usually constructed to act directly on noise channels, not the Choi state representation of them. By applying existing channel-level error-suppression protocols directly in the Choi-channel picture, we can then translate the resulting protocols back to the non-Markovian circuit picture to obtain new protocols for tackling non-Markovian noise, which is shown in Fig. 1 and demonstrated through several examples later in this paper. Moreover, when a channel-level protocol can be translated back to the circuit picture, the error suppression performance and complexity of the resulting non-Markovian noise-suppression protocol is inherited directly from that of the corresponding channel-level protocol, without requiring a separate analysis. We begin by developing the \chi -matrix representation and Pauli twirling protocol for non-Markovian noise, showing that Pauli twirling can eliminate the quantum temporal correlation in arbitrary non-Markovian quantum dynamics. Then, using Choi channels, we manage to construct explicit protocols for performing probabilistic error cancellation and purification-based QEM on general non-Markovian noise [ 32 , 33 , 34 , 35 , 36 ] . All of these examples have demonstrated the simplicity and effectiveness of using Choi channels to directly import channel noise suppression techniques for tackling non-Markovian noise, thereby opening up a broad spectrum of new possibilities for addressing non-Markovian noises. Figure 2: (a) Non-Markovian noise caused by interaction with the same environment system at different time steps, where wires in different colours represent systems from different time steps. (b) Shifting the input and output wires of the same time step to the same register outputs the Choi channel. A simplified graph representation is shown in the lower panel. The figure here only shows two circuit layers, which can easily be generalised into an arbitrary number of circuit layers. II Choi channel No quantum system operates in complete isolation. In a quantum computer, repeated interactions with the environment generate temporal quantum correlations, giving rise to non-Markovian noise that poses significant challenges to reliable information processing. The dynamics of such noise are inherently complex. In general, a single time slice of such noise cannot be adequately described by a quantum channel or even by more general linear maps [ 37 , 38 , 39 ] . We can model a noisy circuit using Fig. 2 (a) where layers of gates, denoted as \mathcal{U}_{i} , are interlaced by layers of noisy interactions, denoted as \mathcal{E}_{i} . In general, these interactions \mathcal{E}_{i} can be connected through the environment and thus can be quantumly correlated over time. We will denote the entire non-Markovian quantum dynamics as \mathscr{E} , which can be mapped to a quantum channel \mathcal{C}_{\mathscr{E}} by shifting the input and output wires of the i -th time step to the i -th auxiliary register as shown in Fig. 2 (b). We will call this the Choi channel of the non-Markovian quantum dynamics. For simplicity, let us illustrate this using the example of a non-Markovian noise