MoM是否是网格划分越精确计算越准确
但是我听有人说网格细了以后，所要求解的矩阵会更大，解矩阵时引入的误差也会增加，
比如求解矩阵的迭代次数增加，历次截断误差造成的误差积累也会增加，这样会造成网格刨细以后计算结果反而会不准确
我只了解MoM或FEM的基本原理，没有自己写过算法，从我自己使用电磁计算软件（IE3D、Ansoft Ensemble、HFSS等）的经验，没有出现过这种问题，所以很怀疑这种说法，但是也想不出确切的理由来反驳，所以在这里请教精通MoM、FEM方法的朋友，望不吝赐教
1) mom has no accumulated error because each element has the relationship with all the others by green function.
fem element only has relationship with neighbour elements,so if the tareget object is electric-big/mesh-small enough, it has the above problem.
2) the mesh can be 1/10lembda as usual, but it's ok to be 1/20lembda to compute.
too small meshes make big matrix as you mentioned, which makes it difficult to get good result.If the object is big enough, the problem will be obvious.
3) How to define the "accuracy" to a EM problem is a problem itself when use db as the unit,sometime 3db is acceptable. maybe one can have a look at the inductive current, which is more sensitive to the mesh shape and size.
4) it is no sense to go after the accuracy too much.just take it easy.
相关文章：
- FDTD的网格划分有什么比较好的方法吗？(05-08)
- FDTD:半径为正方网格边长的圆柱探针附近的场迭代(05-08)
- XFDTD6.0能用亚网格吗？(05-08)
- 大家编fdtd程序的怎么对cad文件作网格划分啊？(05-08)
- 高手指点－－HFSS初始网格的问题(05-08)
- fdtd网格划分的问题(05-08)
