Error with lsqnonlin : Error in lsqncommon (line 14) if any(~isfin​ite(initVa​ls.F))

Hello everybody,

I have a question : when I run my programm, I have an error in lsqncommon. So I went in this function to understand my error and the programm stop at this line of the function lsqncommon :

if any(~isfinite(initVals.F))

error(message('optimlib:commonMsgs:UsrObjUndefAtX0', caller));

end

But I don't understand the meaning of this line, is there someone who can explain me this line please ?

Thanks !

Franck

Answer 1

[Alan Weiss]

Apparently, some initial values are not finite. You gave a value of x0 so that your objective function evaluated at x0 gives some NaN or Inf value. The solver cannot proceed from such a point.

Alan Weiss

MATLAB mathematical toolbox documentation

[Franck Farail]

Okay Thank you !

Answer 2

[Franck Farail]

Hello everybody,

So I ckecked my programm and I don't understand : I checked my initials values to see if they are finite and they are finite and then wenn I put these values in the lsqnonlin function, I have the same error.

TF =

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1 Finite value

Check for missing argument or incorrect argument data type in call to function 'isfinite'.

Error in lsqncommon (line 14)

if any(~isfinite(initVals.F))

[Alan Weiss]

Please show the call to lsqnonlin.

Please show the entire error, every single red line.

Please show the values of your arguments, including at least the first line of your nonlinear function.

Alan Weiss

MATLAB mathematical toolbox documentation

[Franck Farail]

opts = optimoptions('lsqnonlin','Algorithm','levenberg-marquardt');

x0 = 3.92699081698724 1.20000000000000 0.300000000000000 3.37745703122409 0

2.35619449019235 2.30000000000000 1.40000000000000 3.21163801526966 0

1.57079632679490 0.100000000000000 3.26000000000000 0.143811735939238 0.523598775598299

0.785398163397448 0.0300000000000000 2.14000000000000 1.27939824687061 0

1.78539816339745 3.15000000000000 0.0200000000000000 0.177638833777887 0

3.92699081698724 1.26000000000000 0.0300000000000000 0.00682008492641159 0;

[param_identifies] = lsqnonlin(@fonction_erreur,x0,[],[],opts);

Error :

Check for missing argument or incorrect argument data type in call to function 'isfinite'.

Error in lsqncommon (line 14)

if any(~isfinite(initVals.F))

Error in lsqnonlin (line 260)

lsqncommon(funfcn,xCurrent,lb,ub,options,defaultopt,optimgetFlag,caller,...

Error in Etalonnage_robot_6R (line 224)

[param_identifies] = lsqnonlin(@fonction_erreur,x0,[],[],opts);

[Alan Weiss]

If you evaluate

fonction_erreur(x0)

you will most likely get an error. This error is the one to fix.

If you don't get an error, then you will probably obtain some strange values of the function, such as characters or some other data type. The function must return double values only.

Alan Weiss

MATLAB mathematical toolbox documentation

[Franck Farail]

Okay so I evaluate the function

fonction_erreur(x0)

I changed some things and now I have this error message

Unable to perform assignment because brace indexing is not supported for variables of this type.

Error in fonction_erreur (line 20)
        erreur_pose{j,1} = Pos_M{j,1} - Pos_F{j,1};

Error in Etalonnage_robot_6R (line 201)
        fonction_erreur(DHM_e);
        

Here is my function :

         function  erreur = fonction_erreur(dhm_e)
    
    global Routil Poutil Pos_F Pos_M j;
    erreur_pose = zeros(100,100);
   

        T06_e = calc_mgd(dhm_e);
        disp('T06_e =');
        disp(T06_e);
        %Tb = matrice_passage_instrument(DHMolt1);
        
        To = matrice_passage_outil_cas_2(Routil,Poutil);

        Pos_M{j,1} = T06_e*To; %Tb
        disp('Pos_M =');
        disp(Pos_M);
        
        erreur_pose{j,1} = Pos_M{j,1} - Pos_F{j,1};
        
        save erreur_de_positions erreur_pose 
        save positions_finales_mesurées Pos_M
        
        erreur = erreur_pose;

    end

[Alan Weiss]

It looks like you are mixing up cell arrays and matrices. This line creates a matrix:

erreur_pose = zeros(100,100);

Then you try to refer to it using cell array indexing:

erreur_pose{j,1} = Pos_M{j,1} - Pos_F{j,1};

I don't know what type of variable Pos_M is (or Pos_F either), whether it is a matrix or a cell array. Assuming that it is a matrix, you should write

erreur_pose(j,1) = Pos_M(j,1) - Pos_F(j,1);

Then again, I don't see a loop here defining the indices j and i. Nor do I see a need for any such indices. Maybe you shoud write

erreur_pose = Pos_M - Pos_F;

I really cannot understand your function, either what it is or what it is trying to accomplish. But this indexing suggestion should help.

Alan Weiss

MATLAB mathematical toolbox documentation