Compiling a high-level quantum circuit down to a low-level description that can be executed on state-of-the-art quantum computers is a crucial part of the software stack for quantum computing. One step in compiling a quantum circuit to some device is quantum circuit mapping, where the circuit is transformed such that it complies with the architecture's limited qubit connectivity. Because the search space in quantum circuit mapping grows exponentially in the number of qubits, it is desirable to consider as few of the device's physical qubits as possible in the process. Previous work conjectured that it suffices to consider only subarchitectures of a quantum computer composed of as many qubits as used in the circuit. In this work, we refute this conjecture and establish criteria for judging whether considering larger parts of the architecture might yield better solutions to the mapping problem. We show that determining subarchitectures that are of minimal size, i.e., of which no physical qubit can be removed without losing the optimal mapping solution for some quantum circuit, is a very hard problem. Based on a relaxation of the criteria for optimality, we introduce a relaxed consideration that still maintains optimality for practically relevant quantum circuits. Eventually, this results in two methods for computing near-optimal sets of subarchitectures$\unicode{x2014}$providing the basis for efficient quantum circuit mapping solutions. We demonstrate the benefits of this novel method for state-of-the-art quantum computers by IBM, Google and Rigetti.
翻译:编译高层量子电路成为低层描述使其可以在现代量子计算机上运行是量子计算软件栈的关键部分。量子电路映射是将电路转换为符合体系结构的有限量子比特连接的过程。由于量子电路映射的搜索空间随比特数的增长呈指数级增长,因此希望在此过程中考虑尽可能少的设备物理量子比特。以前的研究假设只考虑与电路中使用的电路一样多的量子比特构成的量子计算机的子架构就足够了。在本研究中,我们驳斥了这个假设,并建立了判断是否考虑体系结构的更大部分可以产生更好的映射解决方案的标准。我们表明确定最小大小的子架构是一个非常困难的问题,即不能删除任何物理量子比特而不失去某些量子电路的最佳映射解决方案。基于标准最优性的松弛,我们引入了一种松弛考虑,它仍然保持了实际相关量子电路的最优性。最终,这导致两种计算近似最优的子架构集合的方法,为有效的量子电路映射解决方案提供了基础。我们通过IBM、Google和Rigetti的最先进的量子计算机展示了这种新方法的优点。