I run a matlab code. I get the plots but do not get the variables in workspace. Somebody please help me to fix this.

I am running the following matlab code.

function whisker_curvature
clc
clear all
%global t  x y
x=[1 1 1 1 1 1 1];
% options=optimset('Display''iter');
x1=fsolve(@solver,x);

end

function  F=solver(x)
%options = odeset('RelTol',1e-8,'AbsTol',[1e-5, 1e-5, 1e-5, 1e-5, 1e-5, 1e-5]);
[t,y] = ode45(@equation,[0, 1], [0 0 0 x(1) x(2) x(3) 0]);%,options); % theta-zero_0 is also OK/
s = length(t);
F = [y(s,1)-14.96, y(s,2)-4.48, y(s,3)- 30];%, u(s,4), u(s,5), u(s,6)]; %at theta(1)=11, we have intersting behavior.
figure(1)
 plot(t,y(:,1),'--ro',t,y(:,2),'--bo',t,y(:,3), '--g',t,y(:,4),t,y(:,5), '--b',t,y(:,6), 'go',t,y(:,7), '--r')  % ,t,y(:,2),'--bo',t,y(:,3),t,y(:,4),t,y(:,5),t,y(:,6), 'go'
%  plot(t,y(:,1),'--ro')
 %plot(t,y(:,2),'--bo')
xlabel('sigma')
ylabel('unknowns')
%grid on

end

function dy = equation(t,y)  % t is my \sigma and unknowns (x,y,\theta,F,\theta_{obj},s_{obj},x_u)=(y(1),y(2),y(3),y(4),y(5),y(6),y(7))
L = 20;
dy = zeros(7,1);
dy(1)=y(6)*cos(y(3));
dy(2)=y(6)*sin(y(3));
dy(3)=2*0.02/(1+(2*0.02*y(7))^2)^(3/2)+(y(6)*y(4))/3*(pi/4*((L-y(6)*t)/L)^4)*((15.13-0.25*sin(y(5))-y(1))*cos(y(5))+(4.29+0.25*cos(y(5))-y(2))*sin(y(5)));
dy(4)=0;
dy(5)=0;
dy(6)=0;
dy(7)=y(6)*1/(sqrt(1+(2*0.02*y(7))^2));

end

I get the plots but do not get the variables value in the workspace window.

I suspect this happens due to the following error message received:

Warning: Trust-region-dogleg algorithm of FSOLVE cannot handle non-square systems; using Levenberg-Marquardt algorithm instead. > In fsolve (line 298) In whisker_curvature (line 7)

Solver stopped prematurely.

fsolve stopped because it exceeded the function evaluation limit, options.MaxFunctionEvaluations = 1400 (the default value).

Someone please help me to fix this problem.

[Jan]

Your code is not readable, if you do not use the "{} Code" button to format it. It is your interest to provide readable code.

[vikas mishra]

Thanks and sorry for that. By the way, how to use the "{} Code" button to format it?

[Stephen23]

@vikas Mishra: first select the code text, then click the {} Code button. Ensure that there are blank leading and trailing lines.

[Jan]

I've edited the question and applied the formatting now.

The "clear all" on top of a function is completely meaningless. It deteles all variables of the current workspace, but on top of a function the (local!) workspace is empty at all. In addition clear all removes all functions from the memory, and reloading them from the disk is a waste of time. Simply omit the darn clearing. It does not help.

[Stephen23]

Basically this:

function whisker_curvature
clc
clear all

is a good example of this:

https://en.wikipedia.org/wiki/Cargo_cult_programming

Answer 1

[Niklas Nylén]

Variables created in a function do not get assigned to the base workspace. To debug the variable values you should add a debug stop in your function. When the debugger is stopped the function you will have access to the function's local variables.

[vikas mishra]

thanks. please tell me how to do that?

[Stephen23]

How to use the debugging tools is explained in the MATLAB documentation:

https://www.mathworks.com/help/matlab/debugging-code.html

https://www.mathworks.com/help/matlab/matlab_prog/debugging-process-and-features.html

Answer 2

[Jan]

The warning tells you, that the default "Trust-region-dogleg" cannot be applied to your problem. You can use optimset to enable a suiting method, e.g. the chosen "Levenberg-Marquardt".

It sounds strange, that fsolve stops with an error, but you still get plots. The message tells you, that the maximum number of function evaluations is exceeded. Either it is a bug in the formula, or the problem does not converge, or you have to increase the number of accepted function evaluations.

[vikas mishra]

@ Jan , Thank you. I could not use optimset properly. As I have now provided the readable code, would you like to suggest some suitable optimset.

[Jan]

What about:

opt = optimoptions('Algorithm', 'levenberg-marquardt', ...
                   'MaxFunctionEvaluations', 5000);
fsolve(@solver, x, opt)

See https://www.mathworks.com/help/optim/ug/fsolve.html#buta__s-options. The "Tolerances and Stopping Criteria" influence the number of required function evaluations also.

I cannot try this, because I do not have the Optimization Toolbox.

[vikas mishra]

@ Jan , Thank you. I tried but happened to get the following error:

??? Undefined function or method 'optimoptions' for input arguments of type 'char'.

Error in ==> whisker_curvature at 7 opt = optimoptions('Algorithm', 'levenberg-marquardt', 'MaxFunctionEvaluations', 5000);

>>

[Jan]

Then let me encourage you to read the documentation of your Matlab versions. In the current version (and the online documentation) the command is called "optimoptions", but in your version it might be "optimset".

[vikas mishra]

@ Jan Thanks, it worked.