problems with norm()

Question

Barbara Schläpfer 2020-12-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/687373-problems-with-norm

回答： Manvi Goel 2020-12-14

Hi

I am trying to calculate the cost for my optimization of two graphs.

For have implied a cost funciton, but even if my values are pretty small, the value of my cost function is very big. Any tips or suggestions why that happens?

cost = norm(experimental_force_interp - abaqus_force) / lenght(abaqus_force);

2 个评论
显示无隐藏无

dpb 2020-12-8

Either the difference is large or the force is small, one.

Mayhaps there's one observation very near zero that is the culprit overall?

Standardizing data can often help; without knowing a whole lot more about the problem it's not really feasible to have much else to suggest that might be pertinent.

David Goodmanson 2020-12-8

在 MATLAB Online 中打开

Barbara,

there is no substitute for actually seeing the data:

n = length(abaqus_force);
plot(1:n, experimental_force_interp, 'o-', 1:n, abaqus_force, 'o-')
grid on

Here the 'o-' options are so that no point in an unobvious place gets missed. If the o markers are too dense the options can be dropped.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Manvi Goel 2020-12-14

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/687373-problems-with-norm#answer_574975

The reason for a bigger value of cost function could be wither a higher difference in experimental_force and abaqus_force or very small denominator, i.e., length of abaqus_force.