Undefined function '[function name] ' for input arguments of type 'double'.

I got this error message although my [function name] is no function but rather a input variable to an objective function used in fmincon. This is my code:

function Param_Optim = Optimizer
startSimulator;

Param_Start=[2;1.4;2;2;1.4;2];
lb=[0;0;0;0;0;0];%lower boundary
ub=[10;10;10;10;10;10];%upper boundary
options = optimoptions('fmincon','MaxIter',5,'Display','iter-detailed');
[Param_Optim,fval]= fmincon(@fun,Param_Start,[],[],[],[],lb,ub,[],options);
Param_Optim
fval
end
function J = fun(Param_Value)
sim_obj=sdo.SimulationTest('Hummingbird_Simulation_PositionControl_original');
Param_Value;
%Param=param.Continuous(Param_Value);
%sim_obj.Parameters=Param;
sim_obj.LoggingInfo.LoggingMode='LogAllAsSpecifiedInModel';
sim_out=sim(sim_obj);
Y_x=sim_out.LoggedData.get('logsout').getElement('X').Values;
Y_y=sim_out.LoggedData.get('logsout').getElement('Y').Values;
Y_z=sim_out.LoggedData.get('logsout').getElement('Z').Values;
W_x=sim_out.LoggedData.get('logsout').getElement('X_cmd').Values;
W_y=sim_out.LoggedData.get('logsout').getElement('Y_cmd').Values;
W_z=sim_out.LoggedData.get('logsout').getElement('Z_cmd').Values;
W=[W_x.get('Data') W_y.get('Data') W_z.get('Data')];
Y=[Y_x.get('Data') Y_y.get('Data') Y_z.get('Data')];
length_W=length(W);
%time=W_x.time;
E=[(W(1:length_W,1)-Y(1:length_W,1)).^2 (W(1:length_W,2)-Y(1:length_W,2)).^2 (W(1:length_W,3)-Y(1:length_W,3)).^2 ];
J=sum(sum(E));
% hold on
% plot(time,W(1:length_W,1),time,Y(1:length_W,1));
% figure
% plot(time,W(1:length_W,2),time,Y(1:length_W,2));
% figure
% plot(time,W(1:length_W,3),time,Y(1:length_W,3));
end

[Walter Roberson]

Which line is the problem occurring on? Which name is it giving?

[FloRida]

On the following line: Param_Value;

Matlab takes Param_Value as function, although it is in the definition of the function contained as input.

[Walter Roberson]

sdo.SimulationTest or Hummingbird_Simulation_PositionControl_original might contain a "clear all" somewhere in it.

[FloRida]

First of all, thank you for your help. My problem is, that i do not really get how fmincon works. I understood Param_Value as some kind of a control variable so it is not defined before. I thought that fmincon would call the function fun with its initial guess first and therefore Param_Value would take on the value Param_Start in the first simulation. Afterwards it would be changed by fmincon to minimize the value J. But neither of this happens.

[Walter Roberson]

That understanding is correct. However you are calling a simulation routine within your objective function, and if that simulation routine contains a "clear all" then you metaphorically shoot the leg you are standing on and Param_Value would be erased along with all of the other "all" that "clear all" clears.

[FloRida]

Okay. But how can a simulation contain a clear all? Where may I find it? However, when I run the optimizer file now, there is no error, but the result is the start vector with 0 iterations of fmincon and a f-count of 7. And if i change the starting vector it does the same, so the starting vector's not the optimum.

Answer 1

[Alan Weiss]

Sometimes you need to change the finite difference step size, as explained in Optimizing a Simulation or ODE.

Alan Weiss

MATLAB mathematical toolbox documentation

[FloRida]

That worked fine for me. Thanks