Abstract:Among recent insights into learning quantum states, online learning and shadow tomography procedures are notable for their ability to accurately predict expectation values even of adaptively chosen observables. In contrast to the state case, quantum process learning tasks with a similarly adaptive nature have received little attention. In this work, we investigate online learning tasks for quantum processes. Whereas online learning is infeasible for general quantum channels, we show that channels of bounded gate complexity as well as Pauli channels can be online learned in the regret and mistake-bounded models of online learning. In fact, we can online learn probabilistic mixtures of any exponentially large set of known channels. We also provide a provably sample-efficient shadow tomography procedure for Pauli channels. Our results extend beyond quantum channels to non-Markovian multi-time processes, with favorable regret and mistake bounds, as well as a shadow tomography procedure. We complement our online learning upper bounds with mistake as well as computational lower bounds. On the technical side, we make use of the multiplicative weights update algorithm, classical adaptive data analysis, and Bell sampling, as well as tools from the theory of quantum combs for multi-time quantum processes. Our work initiates a study of online learning for classes of quantum channels and, more generally, non-Markovian quantum processes. Given the importance of online learning for state shadow tomography, this may serve as a step towards quantum channel variants of adaptive shadow tomography.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to accurately predict the passing probability of a given test pair (i.e., a pair of input state and measurement operation) in the online learning of quantum processes. Specifically, the author explores how to effectively perform online learning of quantum processes in the face of adaptive and possibly adversarial test pairs, especially for quantum channels and Pauli channels with limited gate complexity. In addition, the paper extends these results to non - Markovian multi - time processes and provides corresponding results for online learning and shadow tomography. ### Problem Background Online learning of quantum processes refers to a serialized interaction process in which the learner needs to continuously update its understanding of the unknown quantum process based on past observation results and predict the passing probability of new test pairs accordingly. Different from traditional batch learning, online learning requires the learner to be able to process new data in real - time and quickly adjust its hypothesis, which is more flexible and practical in practical applications. ### Specific Objectives 1. **Online Learning of Quantum Channels**: - For quantum channels with limited gate complexity, the author shows that these channels can be learned online and gives upper bounds on the regret value and the number of errors. - For Pauli channels, the author also provides upper bounds on the regret value and the number of errors, and these upper bounds are well - scalable with respect to the system size. 2. **Non - Markovian Multi - Time Processes**: - The author extends the above results to non - Markovian multi - time processes, which describe the evolution of the system at multiple time points. - Provides results for online learning and shadow tomography of non - Markovian multi - time processes. 3. **Computational Complexity**: - The paper also discusses the computational complexity of online learning of these quantum processes and proves that in some cases, even if unbounded computational resources are allowed, efficient online learning cannot be achieved. ### Main Contributions - **Theoretical Results**: - The author provides upper bounds on the regret value and the number of errors in the online learning of quantum channels and Pauli channels with limited gate complexity. - These results are applicable not only to single - use quantum channels but also extended to multi - use and non - Markovian multi - time processes. - **Technical Methods**: - Uses techniques such as multiplicative weight update algorithm, classical adaptive data analysis and Bell sampling. - Introduces the quantum comb theory to deal with multi - time quantum processes. - **Practical Applications**: - These results provide a theoretical basis for online learning and shadow tomography of quantum processes and are helpful for developing more efficient quantum information processing algorithms. ### Summary This paper provides a series of theoretical results and technical methods by studying key problems in the online learning of quantum processes, especially for quantum channels and Pauli channels with limited gate complexity. These achievements not only enrich quantum information theory but also provide an important reference for learning quantum processes in practical applications.

Online learning of quantum processes

Adaptive Online Learning of Quantum States

Learning Quantum Processes and Hamiltonians via the Pauli Transfer Matrix

Learning quantum processes without input control

Learning Quantum Processes with Quantum Statistical Queries

Entanglement-enhanced learning of quantum processes at scale

Agnostic Process Tomography

Online learning of a panoply of quantum objects

Shadow process tomography of quantum channels

Exponential learning advantages with conjugate states and minimal quantum memory

Practical learning of multi-time statistics in open quantum systems

ShadowNet for Data-Centric Quantum System Learning

The power and limitations of learning quantum dynamics incoherently

A learning theory for quantum photonic processors and beyond

Classical Shadows for Quantum Process Tomography on Near-term Quantum Computers

Shadow Simulation of Quantum Processes

Implementation and Learning of Quantum Hidden Markov Models

Information-theoretic generalization bounds for learning from quantum data

Learning unitaries with quantum statistical queries

Learning the eigenstructure of quantum dynamics using classical shadows