object function iteration gets stuck

Help!!!I'm comfused!!! why it always get stuck when I tried a iteration of the object function ?both in the optimization tool and command window?

As in the title,I'm calculating the minimum values of a 6 parameters-function with the

fminsearch function, both tried in the tool box and the commmand window,and it always get seized up and be no responsible at all, and the 'stop'and 'pause' buttons will not work at all. Did anyone has met the same problem?

Thank you!

And part of the function codes:

options=optimset('OutputFcn',@outf,'Display','iter','MaxIter',20);
hold on
objfun=@fitmeson;
[x,fval,exitflag,output]=fminsearch(objfun,[0.0021,0.062,0.329,0.875,0.394,10.009],options)
hold off

and the results screenshots:

    >> OutputFcn

r = 0.0021    0.0698836     0.289626       0.8604       0.4159           10

   Iteration   Func-count     min f(x)         Procedure
       0            1         0.102388         

r = 0.0021    0.0698836     0.289626       0.8604       0.4159           10
     1            7        0.0980085         initial simplex

r = 0.0021    0.0698836     0.304107       0.8604       0.4159           10
     2            9        0.0832537         reflect

r = 0.002135    0.0710483     0.294453      0.87474     0.395105      10.1667
     3           10        0.0832537         reflect

r = 0.002135    0.0710483     0.294453      0.87474     0.395105      10.1667
     4           11        0.0832537         reflect

r = 0.002135    0.0710483     0.294453      0.87474     0.395105      10.1667
     5           13        0.0753749         expand

r = 0.00196194    0.0685895     0.306789     0.911387     0.397416      10.0093
     6           15        0.0753749         contract inside

r = 0.00196194    0.0685895     0.306789     0.911387     0.397416      10.0093
     7           16        0.0753749         reflect

r = 0.00196194    0.0685895     0.306789     0.911387     0.397416      10.0093
     8           17        0.0753749         reflect

r = 0.00196194    0.0685895     0.306789     0.911387     0.397416      10.0093
     9           19        0.0753621         reflect

r = 0.0021031    0.0678226     0.305696     0.896988     0.386462      10.0147
    10           20        0.0753621         reflect

r = 0.0021031    0.0678226     0.305696     0.896988     0.386462      10.0147
    11           28        0.0743308         shrink

r = 0.00203252     0.068206     0.306243     0.904187     0.391939       10.012
    12           30        0.0743308         contract outside

r = 0.00203252     0.068206     0.306243     0.904187     0.391939       10.012
    13           31        0.0743308         reflect

r = 0.00203252     0.068206     0.306243     0.904187     0.391939       10.012
    14           33        0.0736525         reflect

r = 0.00207703    0.0665389     0.305426     0.895153     0.391695      10.1499
    15           35        0.0729338         reflect

r = 0.0020862    0.066819     0.30623     0.89782     0.39315      9.9468
    16           37        0.0729338         contract inside

r = 0.0020862    0.066819     0.30623     0.89782     0.39315      9.9468
    17           38        0.0729338         reflect

r = 0.0020862    0.066819     0.30623     0.89782     0.39315      9.9468
    18           39        0.0729338         reflect

r = 0.0020862    0.066819     0.30623     0.89782     0.39315      9.9468

And whatever I set the iteration as 100 or 40 or some other values,it will stuck here at 18th iteration~!!!!

Answer 1

[Titus Edelhofer]

Hi,

what happens If you hit Ctrl-C? Do you see an error message that tells you, where you interupted the calculation? If yes, set a breakpoint there and start again. When you come to the 18th iteration, take a look at the sizes of variables ...

Titus

[Sheng Liu]

thank you so much!I will try!

[Sheng Liu]

hello，Thank you for your good advice,I've already tried for several times,and I get the same result.

After Ctrl-C,it appears in the command window like this:

Operation terminated by user during prufer

In propagate_solution (line 193)

[theta0,theta1] = prufer(y(1),y(2),y1(1),y1(2),e,vbar,h,S);

In locate_eigenvalue (line 39)

[yr,yer,theta] =

propagate_solution(eold,yf,yef,part.imatch+1,truncb,part,Vmatch,false);

In calculate_eigenvalues>init_ev (line 454)

[prufc,enew] = locate_eigenvalue(ec,part,vmin,trunca,truncb);

In calculate_eigenvalues>calculate_eigenvalue (line 303)

[e,varargout{2},varargout{3},varargout{4},varargout{5},small]...

In calculate_eigenvalues (line 171)

[ev,e1,pruf1,e2,pruf2,pruf,fail] =

calculate_eigenvalue(e1,pruf1,e2,pruf2,n,bp,Vmin,Vmax,trunca,truncb);

In cpm/get_ppm (line 12)

varargout{1} = calculate_eigenvalues(varargin{3:end});

In ppm/get_eigenvalues (line 130)

E=get_ppm(ppm_child,'eigenvalues',emin,emax,nmin,nmax);

In massp (line 43)

E=get_eigenvalues(cp,0,4,true);

In fitmeson>@(r)sum(abs(massp(r)-mp)./mp) (line 23)

Ep=@(r) sum(abs(massp(r)-mp)./mp);%sums of the average error

In fitmeson (line 29)

E=(Ef(r)+Ep(r))/N;%calculating average error

In fminsearch (line 362)

x(:) = v(:,j); fv(:,j) = funfcn(x,varargin{:});

In OutputFcn (line 2)

[x,fval,exitflag,output]=fminsearch(@fitmeson,[0.0021,0.062,0.329,0.875,0.394,10.009],options)

K>>

and when it get stuck all the other applications have response except MATLAB,and the occupancy rate of CPU were not very high,just about 30~40 percent.

I have no idea about what the error message mean,so please help me,thank you!!!

Answer 2

[Titus Edelhofer]

Hi,

what you see is the stack trace, i.e., fminsearch calls fitmeson, fitmeson calls massp etc. Take a look at the top most lines:

In propagate_solution (line 193)
[theta0,theta1] = prufer(y(1),y(2),y1(1),y1(2),e,vbar,h,S);

In locate_eigenvalue (line 39)
[yr,yer,theta] =
propagate_solution(eold,yf,yef,part.imatch+1,truncb,part,Vmatch,false);

In calculate_eigenvalues>init_ev (line 454)
[prufc,enew] = locate_eigenvalue(ec,part,vmin,trunca,truncb);

What happens here in the functions? Is there perhaps some loop in the locate_eigenvalue function that does not terminate? Or in propage_solution? When you stopped there, use the debugger to identify loop counters (if there) or other variables that might diagnose what happens ...

Titus