fmincon-Getting error Unable to perform assignment because dot indexing is not supported for variables of this type

k=garchFunction(u)
u=[0 0 0]
A=[0 1 1]
b=1
Aeq=[]
beq=[]
lb=[0 0 0]
up=[1 1 1]
[x,fval]=fminunc(@garchFunction,u,A,b,Aeq,beq,lb,up)


function y = garchFunction(x) %function for likelihood
Values1814=xlsread('C:\Users\pc\OneDrive - stud.unifi.it\Desktop\Econometrix\CAC40(2014-2018).csv','E2:E1024')
Values=Values1814/(10^6)
returns=tick2ret(Values)
returns=returns-mean(returns)
T=length(returns)

omega=x(1);
alpha=x(2);
beta=x(3);

sigma=zeros(T,1)
sigma(1)=0.011207749644594;
likelihood=0
    for i=2:T;
        sigma(i) = sqrt( omega + alpha*returns(i-1)^2 + beta*sigma(i-1)^2 );
        likelihood = likelihood + 0.5*(-log(2*pi)-log(sigma(i)^2) - (returns(i)/sigma(i))^2);
    end
y = -likelihood;
end

I have this function for the estimation of the maximum likelihood for garch. However, although it estimats perfectly the likelihood, once I want to minimize it trough the function fmincon i get the error shown in the Title. Can someone help me understand what is wrong? Thanks.

[Matt J]

Possibly, but we need to see copy/pastes of the error message. We also need you to attach the .csv file so that we can run the example.

[Diego Tintori]

yes, sorry...

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

Error in fminunc (line 230)
options.GradObj = optimget(options,'GradObj',defaultopt,'fast',allDefaultOpts);

[Matt J]

I don't get any error messages from fminunc, but your .csv file read doesn't return anything

>> Values1814=xlsread('CAC40(2014-2018).csv','E2:E1024')

Values1814 =

     []

[Walter Roberson]

I seem to recall that there was a bug in one release that was triggered only if you did not pass in your own options. Which release are you using?

[Diego Tintori]

Sorry, I don't understand what a realease is...However, it's not possible the .csv file doesn't work. I get the full array and furthermore the function estimates the likelihood as well as the one estimated by the garch function by Matlab(I tried setting the optimal parameters estimated by Matlab's function). The problem lies only when I try to minimize it through fmincon and I don't understand why...

[Walter Roberson]

Release is matlab version, such as r2017a

[Diego Tintori]

R2019b

[Walter Roberson]

Use readtable instead of xlsread. The file is NOT csv, it is semicolon separated. It has a text header line and the first column is a date. If you are not using Windows with Excel installed then xlsread is going to turn the reading over to csvread which cannot handle the file.

[Diego Tintori]

Actually I restructured the file such that i could separate the commas in columns in order to get the column E which I was interested in. I don't think that's the problem since the function works perfectly and the problem arises only with fmincon function. I have windows with excel installed.

[Walter Roberson]

The code calls fminunc but you write fmincon repeatedly. The difference could be important to reproducing the problem.

[Diego Tintori]

I'm really sorry, the function was indeed fmincon and I coded fminunc, that's why it didn't run.

Thank you!

[Walter Roberson]

To confirm, with fmincon in place the problem disappears?