Hybrid Quantum-Classical Computation

混合量子-经典计算结合了量子和经典处理器的能力，以更快地计算。这在当前的噪声中间尺度量子（NISQ）计算时代尤其重要，因为量子处理器仍然非常有限。此外，混合计算具有实现实际量子优势的最大潜力，因为容错量子计算机完全取代经典计算机还需要很长时间。由于这些原因，混合量子-经典架构在最近对量子计算机潜力的启发式探索中占据了中心地位。然而，令人痛苦的是，迄今为止缺乏一个严格的框架。HQC项目的目标是将这一重要领域置于坚实的理论基础之上。我们的目标是解决有关混合计算的力量的基本问题：-什么问题更适合混合量子经典方法？- 我们用什么算法来解决这些问题呢？在这种方法下，什么数据结构和编码性能更好？- 我们能否在这种情况下对这些问题进行分类，即我们能否定义混合量子-经典计算复杂性类？ 为了解决这些问题，HQCC项目被构造成三个互补的部分：一个处理复杂性理论方面（例如量子优势可能的制度），一个涉及混合计算的算法（例如，变分量子算法和将数据编码到它们中的最有效方法），一个处理学习和训练过程。这三个组成部分将在相互对话中发展，以全面理解混合量子-经典计算。

Hybrid quantum-classical computing combines the power of quantum and classical processors to compute faster. This is of particular relevance in the current Noisy Intermediate-Scale Quantum (NISQ) computing era, when quantum processors are still very limited. Moreover, hybrid computing bears the strongest potential for reaching a practical quantum advantage, as it will take a long time before fault-tolerant quantum computers completely replace classical machines, if ever. For these reasons, hybrid quantum-classical architectures have taken center stage in the recent heuristic exploration of the potential of quantum computers. A rigorous framework is, however, painfully lacking to date. The goal of project HQCC is to put this important field onto solid theoretical footing. We aim to address the fundamental questions about the power of hybrid computation:- What problems are more amenable to the hybrid quantum-classical approach? - With what algorithms can we solve them?- What data structures and encodings perform better under this approach? - Can we classify how hard these problems are in this context, namely can we define hybrid quantum-classical computational complexity classes? To address these questions, project HQCC is structured into three complementary parts: one dealing with complexity-theoretic aspects (such as the regimes in which quantum advantage is possible), one concerned with algorithms for hybrid computation (for example, variational quantum algorithms and the most efficient ways to encode data into them), and one dealing with learning and training processes. These three components will be developed in dialogue with each other to extract a comprehensive understanding of hybrid quantum-classical computation.