机构地区: 中山大学工学院应用力学与工程系
出 处: 《中山大学学报(自然科学版)》 1997年第4期11-15,共5页
摘 要: 以子空间缩聚及正交分解为基础,根据实矩阵的奇异值分解定理,对广义特征值摄动问题,提出了一种能同时有效地处理孤立特征值、相重特征值及相近特征值三种不同情况的逐步逼近法.计算实例表明,该方法合理可靠。 For the generalized eigenvalue perturbation problem of real modes, an efficient method is put forward, which is applicable to all the three cases of eigenvalue problems: distinct, repeated and nearly equal eigenvalues. This method is a kind of successive approximation procedure, developed by performing an orthogonal decomposition and a sub space condensation and by using the singular value decomposition of real matrix. Some illustrative examples are presented to demonstrate the validity of this proposed technique.