Error using eig Input matrix contains NaN or Inf.

I get the error shown below trying to run a DCC GARCH Model. How can I correct this? I tried to use abs(X) on each variable but it still does not work.

Thanks for all advice in advance.

Error using eig

Input matrix contains NaN or Inf.

Answer 1

[Jan]

The error message is clear: The input of eig() cannot contain NaN or Inf.

Without seeing your code, the readers cannot guess, where the NaN or Inf are coming from and how you could avoid this problem. Why do you assume that abs(X) helps and what does "on each variable" mean?

[Tilman Zottl]

R = mean(stdData,3);

r = abs(sqrt(diag(R)));

R = R ./ (r*r');

N = abs(mean(stdDataAsym,3));

%scale = max(eig(R^(-0.5)*N*R^(-0.5)));

scale = max(eig(R^(-0.5)*N*R^(-0.5)));

This is the specific code with scale being unable to execute. I meant R, r, and N by all variables.

[Tilman Zottl]

Also I am ussing the dcc.m function of the MFE Toolbox

[Jan]

Does r contain zeros? Then R must contain NaNs and eig() cannot work anymore.