| Among many high-risk application problems,the accuracy of the calculation results is necessary.However,due to the limitation of the real number accuracy,the accumulation of calculation errors often affects the accuracy of the calculation results.In many practical problems,even small errors may cause major accidents.At the same time,tensor features have important applications in image and signal processing,blind source separation,and statistics.Therefore,it is of great significance to study the credible verification of tensor eigenpairs.In this paper,we studies the problem of verified the symmetric tensor H-eigenpairs and the symmetric tensor B-eigenpairs.First,using the inverse Newton method,a credible verification algorithm for H-eigenpairs of symmetric tensor are given.This algorithm obtains the approximate solution of the H-eigenpairs and its corresponding error bound,so that an accurate H-eigenpairs must exist within the error range of the approximate solution.Second,using the adaptive trust region method,a trusted verification algorithm for the symmetric tensor B-eigenpairs are given.Finally,the above algorithm is implemented with the INTLAB software package in Matlab,and numerical experiments are conducted,and the results show the effectiveness of the algorithm. |
