[Source code and Dataset] FrozenQubits: Boosting Fidelity of QAOA by Skipping Hotspot Nodes

FrozenQubits is a novel quantum divide-and-conquer approach that leverages the insight about most real-world applications having power-law graphs to boost the fidelity of QAOA applications. In FrozenQubits, we freeze some of the nodes with the highest degree of connectivity. This drops the edges connected to these nodes resulting in a circuit with fewer CNOTs and lower depth.FrozenQubits是一种新颖的量子分治法（quantum divide-and-conquer），其依托现实场景中多数应用均具备幂律图（power-law graphs）的特性，以此提升量子近似优化算法（Quantum Approximate Optimization Algorithm，QAOA）应用的保真度。在FrozenQubits框架内，我们将冻结部分连接度最高的节点，移除与这些节点相连的边，最终得到包含更少受控非门（Controlled-NOT，CNOT）且深度更低的量子电路。