基于GPU集群的大规模量子线路仿真理论与方法研究

量子计算是信息科学领域研究前沿之一。在量子计算机实用化之前，量子计算仿真技术是开展量子计算理论研究的重要手段。本课题采用量子线路计算模型，运用大规模GPU集群技术，搭建量子计算仿真平台，开展量子线路仿真理论与方法的研究。为建设具有国际领先水平、可模拟多比特量子计算的超级计算机提供理论研究基础。研究内容如下：1.针对量子计算仿真单指令多数据流的特点设计低网络延迟的GPU集群架构；2.针对量子计算仿真中的研究对象和操作步骤，从存储和计算两方面进行仿真计算方法研究；3.针对CPU+GPU异构集群中多级存储结构和多层次通信的特点，研究通用量子计算仿真的GPU集群计算方法；4.对量子计算仿真算法的复杂性和集群软硬件环境进行抽象，开展量子计算仿真系统的性能分析研究。通过上述研究，从计算仿真模型、量子线路仿真、GPU算法优化和系统性能评估等方面为大规模量子线路仿真提供综合化理论支撑和系统可行性验证。

Quantum computing is one of the frontiers in science research. Before the utilization of quantum computer, quantum simulation is an important method to study quantum computing theory. The project is related to theory and method of quantum circuit simulation and constructing GPU quantum computing with the quantum circuit computing model based on large-scale GPU cluster, which can provide a theoretical basis for construction of an international leading level quantum simulation supercomputers. The research contents are as follows: 1. For the features of quantum computing simulation, single instruction multiple data streams, GPU cluster architecture with low latency network will be designed; 2. For the study object and operation steps of quantum computing simulation, we will research on the method of quantum computing simulation from storage and computation aspects; 3. For the characteristics of heterogeneous CPU + GPU cluster, multi-level storage architecture and multi-level communications, oriented general quantum computing simulation, the approach of GPU cluster computing will be studied; 4. Performance of quantum computing simulation system will be analyzed through abstracting the complexity of the quantum computation algorithm and the cluster environment of hardware and software. From aspects of computation simulation model, quantum circuit simulation, GPU algorithm optimization and system performance evaluation, above studies will provide comprehensive theoretical support and verify the feasibility of large-scale quantum circuit simulation.