Too many output parameters error in Fsolve when I turn Jacobian On

Hi all

why do I get Too many output parameters error in Fsolve when I turn Jacobian On ? I have 9 variables to solve , where can be the problem ? and does the Jacobian matrix calculation help me if this is the iteration report ? :

 Iteration  Func-count     f(x)          step         optimality    radius
     0         10    6.09647e+006                     3.27e+005               1
     1         20    5.00637e+006              1      2.98e+005               1
     2         30     2.7289e+006            2.5      2.24e+005             2.5
     3         40         37481.6           6.25      1.99e+004            6.25
     4         50         5.77916        1.28439            192            15.6
     5         60    2.82227e-007      0.0169644         0.0368            15.6
     6         70    1.34282e-021   4.00165e-006      3.17e-009            15.6
     7         80    8.78879e-025   2.98649e-013      1.58e-010            15.6
     
     

[Geoff Hayes]

farzad - please show the code that generates this error so that we can see what you are passing into the function and what you are expecting as outputs. The error messages is telling you that are expecting more output parameters than will actually be returned by the function.

[farzad]

Sorry can't share the code, I am solving for 9 variables, are they too many for Matlab 2007 ?

Answer 1

[Walter Roberson]

When you turn on Jacobian, then you need to return the Jacobian as a second output, but only when it is asked for. You need to test nargout for that.

function [obj, J] = fun_to_solve(x)
   obj = x.^2 - 10*x + 5;
   if nargout > 1
       J = 2.*x - 10;
   end
end

[farzad]

Thank you so much ! But I did not understand

so you are saying that I should change my Fsolve from

[x,fval,exitflag,output] = fsolve(fun,x0,options);

to yours ? what is the funciton x.^2 - 10*x + 5?

[Walter Roberson]

You should change your function so that it only assigns to the second variable if nargout > 1 . Your fsolve() call would not change.

x.^2 - 10*x + 5 is just some function to be solved, which I put in as a place holder. Your actual function is not named fun_to_solve, and your actual variables are not likely named obj and J, so the whole thing is intended to illustrate the pattern you should follow.

You cannot share the code, so I cannot possibly post the code with the exact change in terms of your variable names.

[farzad]

Thank you

Here you habe one function and the jacobian is the derivative of that one function, in my case I have muktuple functions, how Wil my Jacobian look like?

[Walter Roberson]

https://www.mathworks.com/help/optim/ug/fsolve.html#buta__s-fun

"If fun returns a vector (matrix) of m components and x has length n, where n is the length of x0, the Jacobian J is an m-by-n matrix where J(i,j) is the partial derivative of F(i) with respect to x(j). (The Jacobian J is the transpose of the gradient of F.)"