- 积分
- 44
- 注册时间
- 2005-3-18
- 仿真币
-
- 最后登录
- 1970-1-1
|
发表于 2012-2-13 13:16:18
|
显示全部楼层
来自 日本
Lagrange multiplier法, 矩陣為不定
另外, 增廣Lagrange法部分, 從平衡方程(無論靜動態)的角度來理解的話(歸納如下圖),
增廣Lagrange法和Lagrange multiplier法一樣都比penalty法多了不等式之非嵌入條件(平衡方程中多一條判別式其接觸力算法也是Lagrange multiplier和penalty法中接觸力之和。 所以個人看來是有增加計算自由度的
Augmented: Maybe it is better to be translated to 增大
Pls refers to, e.g. Simo and Laursen: An augmented Lagrange treatmenet of contact problems involve friction, Computer & Structures 42(1992), pp97-116. In AL method, lambda is supposed to be constant during each step, and it should be augmeneted to lambda(last step)+penalty*penetration increment at the end of this step. In this case, penalty should be a not much value to avoid the bad number condition the stiffness matrix, and, because lambda is constant during the step, it doesn't introduce additional dof.
|
评分
-
1
查看全部评分
-
|