NaN (Not a Number) problem in integration

2012 9 19
1 个回答
回答已采纳
9 次查看（30 天）

0 个投票

Hi all. I am trying to calculate this integration, it gives NaN ,instead of pi/2. How can I solve the problem?
mz=zeros(4,1); h=zeros(4); I=eye(4);
f=@(x)(arrayfun(@(x)(exp(-mz'*(1./(h+((sin(x).^2).*I)))*mz)),x));
out = quad(f,0,pi/2)
Thanks

请先登录，再回答此问题。

 采纳的回答

José-Luis
José-Luis 2012-9-19
编辑：José-Luis 2012-9-19

0 个投票

No, it should not give you pi/2. The way you have set up your matrices, the expression
h+((sin(x).^2).*I)))*mz
will always be equal to zero. Division by 0 = Inf.

4 个评论
显示 2更早的评论 隐藏 2更早的评论

Serhat
Serhat 2012-9-19
You misunderstood, ((h+((sin(x).^2).*I)).^-1)*mz is my function. So it equals 1/(I matrix) * mz and mz=0, then the function is 0=>exp(0)=1 isn't it?
José-Luis
José-Luis 2012-9-19
编辑：José-Luis 2012-9-19
I'm afraid it's not. Have you tried to visualize your function? Copy/pasting your code and running:
f(0) = NaN;
f(whatever) = NaN;
and that is because
exp(-mz'*(1./(h+((sin(x).^2).*I)))*mz)
will evaluate to NaN, with mz and I defined as you did. Maybe you have misplaced a parentheses?
Serhat
Serhat 2012-9-19
Now it works, you are right. I put parentheses at wrong places. Thank you very much.
f = @(theta)(arrayfun(@(x)(det(sigmaZ./sin(x).^2+eye(4,4)).^-1)*exp(-1*mZ'*1/(sigmaZ+sin(x).^2.*eye(4))*mZ),theta));

请先登录，再进行评论。

更多回答（0 个）

类别

在 帮助中心 和 File Exchange 中查找有关 NaNs 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by