标题: 谈谈计算数学(转自校内) [打印本页] 作者: mzszrj 时间: 2010-1-23 09:06 标题: 谈谈计算数学(转自校内) 虽然不是我写的,但我觉得很好,希望与大家分享。以下的内容转自校内:/ Z' `& H5 j* A# |6 x" Q1 z: x6 l
5 X9 {5 Y: ]' Y4 ?
从计算数学的字面来看,应该与计算机有密切的联系,也强调 ; p" ` M1 H4 `0 K3 C: w了实践对于计算数学的重要性。也许Parlett教授的一段话能 + \. Q; O, w$ x- n最好地说明这个问题: : `/ Z+ V2 M; ~8 B% G) o( Z' z. e% ^, z$ h5 s$ _
How could someone as brilliant as von Neumann think / D( ?- [( v O' dhard about a subject as mundane as triangular factoriz - O U7 {) D" Y) C) @( X# I-ation of an invertible matrix and not perceive that,+ a4 ~# W5 D3 i6 }* e
with suitable pivoting, the results are impressively 1 v* }& r7 i) l; t+ g" S/ ?1 mgood? Partial answers can be suggested-lack of hands-on) Y/ _5 V( f8 Q2 K, @. V) {
experience, concentration on the inverse rather than on; ~7 ^4 \, @4 ^
the solution of Ax = b -but I do not find them adequate. # j- ^% x$ l0 Q! Q0 {Why did Wilkinson keep the QR algorithm as a backup to a ( [# r5 T* C# Z. Y' J: Y' NLaguerre-based method for the unsymmetric eigenproblem3 n/ o, `; l) F& d2 U& \3 F! H
for at least two years after the appearance of QR? Why 1 w! S3 H9 ^- I( U+ E0 X2 E. n- S$ jdid more than 20 years pass before the properties of1 I; r0 i% b5 T# B- ^; l3 T
the Lanczos algorithm were understood? I believe that0 y, j* |( [# R
the explanation must involve the impediments to ; d4 g8 w9 F$ @3 Gcomprehension of the effects of finite-precision & Y) d5 N. C5 Y4 B* t [arithmetic.(引自www.siam.org/siamnews/11-03/matrix.pdf)0 p( l# s- ~& o, \ d
- f# ~- U: x. r7 R% X- `Problems, Inverse Problems in Sciences and Engineering(以前 % T7 x+ s. _0 |7 Y: l6 q) G- J2 y & X& N0 p Z, a5 W2 q4 u6 f: j叫Inverse Problems in Engineering).第一本杂志最好,第二本杂' \5 |+ x5 r+ x5 h" L, N* B- c! ~
( X4 S+ \5 e8 F, ?3 n
志上面有很多苏联人的工作,第三本偏向于应用。在很多高档次的 5 S" U0 S( s% `( ?& e6 f6 L9 ^% T& H6 j* j2 N7 i% }
杂志中都有反问题方面的文章,比如SIAM Journal on Numerical * m; f q I7 ]7 D: F8 F5 d' N2 b/ t- |4 r( e) F
Analysis,SIAM Journal on Mathematical Analysis, SIAM * g4 T4 G5 \6 o( X6 }( M, v. W- o' M6 B, S. j( x; X" g
Journal on Matrix Analysis and Applications,SIAM Journal on4 P+ d- `2 _' |2 x5 G0 J
- _# l- {; f8 X \- A4 NScientific Computing上也有不少反问题方面的文章。/ D' E m" \. b+ O( R
* l- H0 k; T* a5 _- f* M8 K F
在国内做反问题做的最好的应该是复旦大学的程晋老师,他在反问4 ~4 P1 y( H& z
6 o x4 ?+ n1 M( ]9 N题的理论估计方面有不少工作,南京大学的金其年老师也有不少好 $ b1 N( z `, F' a 2 B3 p" p6 E7 e" |$ \( f的结果(很年轻!),哈工大有几个人是做应用方面的工作的(他 ( E$ \9 M; p0 B! l- n# U! }+ c8 [
们的前校长就是做地球物理中的反问题的)。国际上知名的有HW 5 [& z3 i: O z L. ]: S9 [1 \ * [ c7 R( \! b! X/ ~6 p8 d2 j3 Q) {Engl(澳大利亚),Yamamoto(日本), Kress(德国), Martin1 @. K. \0 q( ?* N% q6 ?5 E
3 ?. Q' c( p+ `$ F, e
Hanke(德国), Isakov(美国)等。$ R- Z' g" b; ?' c) t( {' H4 G