Why do I get error, "Not enough input arguments"

Dear Group,

I am trying to estimate parameters from a non-linear model using lsqnonlin.

My function is

function F = myfun(x,y,X)
F = (y - (1 + x(1)*(x(2)/0.0025662))*X(:,1) - (1-x(1)^2)*(x(2)/x(3))*X(:,2) - x(4)*X(:,3))
end

y = is a column vector of 18026 observation
X is a matrix with 3 columns each with 18026 observations
x is my parameter matrix (I have four parameters - x(1), x(2), x(3), and x(4), to estimate.
x0 = [-0.06,0.6,0.006,0.85] are my initial values.

My parameters have upper and lower bounds:

lb = [-1,0,0,-1]
ub = [1,2,2,2]

Because I get different estimates when I use lsqnonlin as I change starting values, I wish to the multiple starting values feature of MATLAB to get a global minimum.

I am using this URL to write the code to feed multiple starting values: https://fr.mathworks.com/help/gads/multistart-using-lsqcurvefit-or-lsqnonlin.html

When I submit

problem = createOptimProblem('lsqnonlin','objective',myfun, 'xdata',X,'ydata',y, 'x0',x0,'lb',lb,'ub',ub);

I get the following error message:

 Not enough input arguments.
 Error in myfun (line 2)
 F = (y - (1 + x(1)*(x(2)/0.0025662))*X(:,1) - (1-x(1)^2)*(x(2)/x(3))*X(:,2) -  x(4)*X(:,3))

Can you kindly point out the error in my set up. Thanks!

best, Srinivasan Rangan

[Jan]

I've formatted your code today. Please read http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

Answer 1

[Jan]

                                                         | This is the problem
                                                         V
  problem = createOptimProblem('lsqnonlin', 'objective', @myfun, ...
                               'xdata',X,'ydata',y, 'x0',x0,'lb',lb,'ub',ub);

Using myfun only calls this function without input arguments, while @myfun provides the function handle.

Answer 2

[Walter Roberson]

Parameterize your function.

https://www.mathworks.com/help/optim/ug/passing-extra-parameters.html

Also, lsqnonlin cannot be used with problem based optimization. crrateOptimProblem is for linear optimization and mixed integer linear optimization only.

[Srinivasan Rangan]

Thank you Walter. I applied Jan's suggestion and it works. When I type problem, I get

problem =

struct with fields:

    objective: @myfun
           x0: [-0.0700 0.0600 0.0060 0.8500]
           lb: [-1 0 0 -1]
           ub: [1 2 2 2]
       solver: 'lsqnonlin'
      options: [1×1 optim.options.Lsqnonlin]

Then when I type [xmulti,errormulti] = run(ms,problem,50)

My error message is:

Not enough input arguments.

Error in myfun (line 2) F = (y - (1 + x(1)*(x(2)/0.0025662))*X(:,1) - (1-x(1)^2)*(x(2)/x(3))*X(:,2) - x(4)*X(:,3))

Error in lsqnonlin (line 206) initVals.F = feval(funfcn{3},xCurrent,varargin{:});

Error in fmultistart

Error in MultiStart/run (line 260) fmultistart(problem,startPointSets,msoptions);

Caused by: Failure in initial objective function evaluation. LSQNONLIN cannot continue. Failure in call to the local solver with supplied problem structure.

Do you know, how I can submit multiple starting values to lsqnonlin?

[Walter Roberson]

You do not appear to have followed my advice about parameterizing the function?

Answer 3

[Srinivasan Rangan]

Thanks Walter. I am a newbie to matlab. Unable to figure how to parametrize the function. Will post a new question, that might make it more clear.

best, Srini