AQER: a scalable and efficient data loader for digital quantum computers
Kaining Zhang, Xinbiao Wang, Yuxuan Du, Min-Hsiu Hsieh, Dacheng Tao
TL;DR
This paper introduces AQER, a scalable and efficient data loader for digital quantum computers that reduces entanglement to improve loading accuracy and efficiency, supported by theoretical bounds and extensive experiments.
Contribution
The paper presents a unified theoretical framework for approximate quantum loaders and introduces AQER, a novel method that systematically reduces entanglement to enhance data loading performance.
Findings
AQER outperforms existing methods in accuracy and gate efficiency.
Theoretical bounds relate infidelity to entanglement entropy.
Experimental validation on datasets with up to 50 qubits.
Abstract
Digital quantum computing promises to offer computational capabilities beyond the reach of classical systems, yet its capabilities are often challenged by scarce quantum resources. A critical bottleneck in this context is how to load classical or quantum data into quantum circuits efficiently. Approximate quantum loaders (AQLs) provide a viable solution to this problem by balancing fidelity and circuit complexity. However, most existing AQL methods are either heuristic or provide guarantees only for specific input types, and a general theoretical framework is still lacking. To address this gap, here we reformulate most AQL methods into a unified framework and establish information-theoretic bounds on their approximation error. Our analysis reveals that the achievable infidelity between the prepared state and target state scales linearly with the total entanglement entropy across…
Peer Reviews
Decision·ICLR 2026 Poster
- The approach is grounded in a clear and well-motivated theoretical insight, which lends credibility to the method and provides a room for future extensions. Given that quantum state preparation is, in general, an infeasible problem in terms of computational complexity, it is reasonable that the paper does not pursue purely theoretical guarantees. - Numerical benchmarks across multiple datasets demonstrate consistent performance improvements over prior methods.
- Since the target hardware setting is the NISQ regime, the lack of experiments on real quantum devices, or even simulations under realistic noise models (e.g., depolarizing noise), makes it difficult to assess how the proposed method would perform in practice. This limitation is particularly relevant given that several prior works in this area include evaluations on real hardware.
The paper's primary strength is Theorem 3.1, which establishes a formal information-theoretic bound between the approximation infidelity and the proposed measure. This provides a solid theoretical justification for the algorithm's design, moving beyond purely heuristic approaches. The entanglement-reduction-guided strategy for building the circuit is a novel and intelligent heuristic. It offers a structured method for ansatz construction that aims to position the optimization in a favorable reg
1. The paper compellingly argues that AQER mitigates barren plateaus in the final optimization. However, a critical discussion or numerical experiment is missing on the optimization landscape. The algorithm's success hinges on this greedy step being efficient. The paper would be significantly strengthened by a discussion of why the local structure of this cost function makes the optimization in Step I tractable. 2. The experimental comparison relies heavily on the two-qubit gate count, which is
- The paper presents a clear, useful bound for all AQLs, formulated in terms of the entanglement measure S. - The AQER algorithm is intentionally designed to minimize the S, which the theory connects directly to infidelity. - Once target states are available, training AQER relies only on local measurements. - Experiments provide encouraging evidence that supports the paper’s central claims
Although the above strength, I do not think the current version of the paper matches the ICLR criteria for the following reason: - My main concern is the computational cost required to reach low infidelity in the general case. The theoretical bound in the paper is meaningful only in the limit S goes to 0. For target states with moderate or high entanglement, achieving a small S requires deeper circuits U, and it is unclear whether this can still be done with circuits of polynomial size. - Als
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsQuantum Computing Algorithms and Architecture · Quantum many-body systems · Quantum-Dot Cellular Automata