基于随机并行梯度下降算法的多模光纤快速精确模态分解

Fast and accurate modal decomposition of multimode fiber based on stochastic parallel gradient descent algorithm.

作者信息

Lü Haibin, Zhou Pu, Wang Xiaolin, Jiang Zongfu

机构信息

College of Optoelectric Science and Engineering, National University of Defense Technology, Changsha, China.

出版信息

Appl Opt. 2013 Apr 20;52(12):2905-8. doi: 10.1364/AO.52.002905.

DOI:10.1364/AO.52.002905

PMID:23669702

Abstract

A new approach for the complete modal decomposition of the optical fields emerging from the multimode fiber is presented in this paper. Based on the stochastic parallel gradient descent algorithm, mode coefficients for all the bound modes in the multimode fiber can be exactly calculated by utilizing one intensity profile of the beam. Numerical simulation validates the feasibility, and the reconstructed error is below 0.1%. In the case of six modes within the fiber, the running time is about 2 s.

摘要

本文提出了一种对多模光纤出射光场进行完整模态分解的新方法。基于随机并行梯度下降算法，利用光束的一个强度分布就能精确计算出多模光纤中所有束缚模的模式系数。数值模拟验证了该方法的可行性，重构误差低于0.1%。对于光纤中有六个模式的情况，运行时间约为2秒。