| CJ's profile老博士PhotosBlogLists |  Tools  Help |
老博士弱智儿童快乐多 June 29 发现我的一个弱点我一直把李开复当成一个娱乐圈人士. 直到最近, 我改变了这种看法.我发现他鼓吹的那些东西确实是我们这个社会已经缺失或者正在缺失的东西．即使他鼓吹的东西不对，最坏的情况，也就是相当于喝一杯白开水． 我发现我不是一个积极向上的人．我是一个消极向上的人．我八成是一个向上的人，向上的特点掩盖了消极的本质．我习惯兵来将挡水来土掩．一般来讲，我也确实可以做到兵来将挡水来土掩．但是，这使我成了一个消极的人．不善于主动创造机会，甚至机会来了也不会抓住．在新事物面前, 我习惯先端详一下它. 除非有足够的理由或者经过思考认为有必要, 否则我的习惯非常顽固． 我要努力克服这一弱点. May 22 How fast the Laplace equation was solved in 1995For uniform problems, which finite difference can be used,
FISHPACK, which use cyclic reduction or FFT is of course the winner!
even 10 times faster than the second best one.
SuperLU, which is based on complete decomposition of matrix, may be interesting if the matrix is small.
And another very excellent package is multi-level ILU preconditioning.
Yvan Notay may be one of the most excellent person in solving this kind of equations. The general package NGILU, is only outerperformed by FISHPACK which only works under a very strict condition.
Also, he is a very interesting person. Last week, I wrote an e-mail to consult him about his
agmg. His reply was very funny. It seems that he was just sitting at the same desk with me, and telling a story: you were happy here..... May 12 结果与成果的区别周一老板对我说, 这次去三堡(石门山庄)讲得比上次在组里做报告时清楚, 但是总得来看, 就是讲了一个结果而不是成果. 今天王博拿了一个老外的一篇文章. 感慨啊.
什么叫结果, 什么叫成果. 这就是差别啊.
一个简单的实验结果, 分析一下背后的东西, 可以变成成果.
这让我想起了洛伦兹变换. 第一个导出来的, 并不是爱因斯坦, 而是洛伦兹.
可惜他被埋在一堆数学里并没给出物理解释.
这就是结果与成果的区别.
April 29 时间差分与积分的关系时间上的差分换一个角度看，就是积分．
欧拉向前/向后差分，其实就是矩形积分公式．
Crank-Nilcon就是梯形积分公式．
从这一个角度上讲，要想time marching做的好，就是要找到好的积分计算方式．
|
||||
|
|
