量子路径积分算法及其在大地电磁反演中的应用

常规非线性反演方法虽然对初始模型的依赖大为减弱，但局部收敛现象和计算速度慢仍然是瓶颈. 本文提出了一种新的反演方法——量子路径积分算法(Quantum Path Integral Algorithm,简称QPIA). 该方法引入量子力学的横向场、传播子等概念，并充分利用量子隧穿效应，大大提高反演的效率，具体是通过对反演目标函数的构建，并以Feynman的传播子来构成模型的接收概率来实现. 在对一维大地电磁模型和实际数据进行试验后，表明该方法比常规反演方法更能够精确、稳定和快速地逼近真实模型.

Quantum path integral algorithm and its application in magnetotelluric inversion

A quantum path integral algorithm (QPIA) is proposed in this paper by introducing quantum conceptions, such as the transverse field and propagator in quantum mechanics. Different from general methods, QPIA, which fully applies quantum-tunneling effect, can improve inversion efficiency greatly in the convergent rate and avoiding to be trapped in local minimum. By composing the objective function and regarding Feynman propagator as the acceptable probability, we accomplish the algorithm. Inversion results of one-dimensional synthetic data set and real data set show that QPIA is more precise, stable and fast to approach the true model compared with traditional nonlinear inversion methods.