实验1.11.21.3课件.pptVIP

实验1.1（病态问题） （1）选择充分小的扰动项系数，反复进行实验.如果扰动项系数很小,标准解和扰动后解的差很小.计算中你有什么出乎意料的发现?表明有些解关于如此的扰动敏感性如何? 当扰动项的系数为0.001时，在MATLAB中运行程序: format long ve=zeros(1,21); ve(2)=0.001; x=roots(poly(1:20)+ve); y=20:-1:1; y’; z=x-y’ ;我们得到如下的数据：这是扰动后得到的解和标准解的差 7.08168416422633 + 5.03812474527973i 8.08168416422633 - 5.03812474527973i 1.53367775966902 + 9.16640451159825i 2.53367775966902 - 9.16640451159825i -2.17645443590958 + 7.77166672445647i -1.17645443590958 - 7.77166672445647i -3.27885935670453 + 5.46090122700207i -2.27885935670453 - 5.46090122700207i -3.08717836418682 + 3.47317252626141i -2.08717836418682 - 3.47317252626141i -2.30732178826853 + 1.89883745787881i -1.30732178826853 - 1.89883745787881i -1.24238968519640 + 0.65471389945693i -0.24238968519640 - 0.65471389945693i -0.04792488665565 0.00061053326870 -0.00000223655074 0.00000000267371 0.00000000000512 -0.00000000000031 ;当扰动象系数是10^10时,两解的差为: -9.99999981000000 -0.00000002383234 -0.00000002195180 + 0.00000000234581i -0.00000002095180 - 0.00000000234581i -0.00000001814125 + 0.00000000345381i -0.00000001714125 - 0.00000000345381i -0.00000001449707 + 0.00000000346550i -0.00000001349707 - 0.00000000346550i -0.00000001135223 + 0.00000000295378i -0.00000001035223 - 0.00000000295378i -0.00000000864749 + 0.00000000227750i -0.00000000764749 - 0.00000000227750i -0.00000000625923 + 0.00000000159428i -0.00000000525923 - 0.00000000159428i -0.00000000408886 + 0.00000000096337i -0.00000000308886 - 0.00000000096337i -0.00000000207427 + 0.00000000040647i -0.00000000107427 - 0.00000000040647i -0.00000000014326 -0.00000000000000 ;观察分析数据可知，有些解（根值较小的解）关于扰动的敏感性差，几乎不受扰动的影响。 （2）将方程(E.1.2)中的扰动项系数加到次数