Function lsqnonlin() has no optimization effect

2020 5 17
1 个回答
更新时间：2020 5 18
2 次查看（30 天）

0 个投票

It is a teaching example, but I can't get the right answer in my software. How can I solve this problem?

4 个评论
显示 2更早的评论 隐藏 2更早的评论

Ang Feng
Ang Feng 2020-5-17
Well, I tried the code, and it produce
x =
0.2578 0.2578
resnorm =
124.3622
also with the warning information as well
Local minimum possible.
lsqnonlin stopped because the size of the current step is less than
the value of the step size tolerance.
<stopping criteria details>
Perhaps, we can ignore this warning.
Matt J
Matt J 2020-5-17
编辑：Matt J 2020-5-17
@Yuwen,
Please do not post your code as an embedded image. We cannot conveniently copy-paste it in that form.
@Ang Feng,
It is not a warning. It is an announcement of success! (Or at least, lsqnonlin believes that it was successful).
Yuwen Zhu
Yuwen Zhu 2020-5-18
@Matt J@Ang Feng Thanks for your answers. I try this code again, and it works right.
Yuwen Zhu
Yuwen Zhu 2020-5-18
And I have another question. My code has no optimization effection.
function f=objfunc(x)
PNO=[60.8,60.8,60.8,60.8,60.8,60.8,15.2,30.4,60.8,121.59,151.99,202.65,60.8,60.8,60.8,60.8,60.8];
PO2=[253.3,506.6,1013.3,2026.5,5066.3,10132.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5,2026.5];
rNO=[7.12e-05,9.66e-05,1.31e-04,1.77e-04,2.66e-04,3.60e-04,5.32e-05,9.72e-06,1.77e-04,3.25e-04,3.948e-04,5.07e-04,1.79e-04,1.77e-04,1.79e-04,1.76e-04,1.79e-04];
a=log(PNO);
b=log(PO2);
y=log(rNO);
f=x(1)+a*x(2)+b*x(3)-y;
x0=[-16.800207,1.2008,0.4151];
[x,resnorm]=lsqnonlin(@objfunc,x0)
Local minimum found.
Optimization completed because the size of the gradient is less than
the value of the optimality tolerance.
<stopping criteria details>
x =
-16.8802 1.2008 0.4151
resnorm =
4.4249

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Matt J
Matt J 2020-5-18

0 个投票

And I have another question. My code has no optimization effect
In this case, your initial guess already is optimal. You can see this by solving for the optimum x analytically, which is possible in this case because your model is linear, and because you have no bound constraints:
a=log(PNO(:));
b=log(PO2(:));
y=log(rNO(:));
x=[ones(size(a)),a,b]\y(:)
x =
-16.8802
1.2008
0.4151

类别

在 帮助中心 和 File Exchange 中查找有关 Choose a Solver 的更多信息

标签

提问:

Yuwen Zhu
Yuwen Zhu
2020-5-17

回答：

Matt J
Matt J
2020-5-18

Community Treasure Hunt

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

Start Hunting!

Translated by