How do define a lognormal pdf function handle inline?

Lognormal is just an example, I know that lognormpdf built-in function is existed in MATLAB. But suppose we want to define it. We can write: lognormal = @(p,x) exp(-(log(x)-p(1)).^2./p(2).^2/2)./x/p(2)/sqrt(2*pi); The problem is, that lognormal([1,1],0) is NaN. Is it possible to modify the definition easily to restrict lognormal([anything],0) = 0?

Answer 1

[Star Strider]

For the lognormal distribution functions, I usually replace the zero values with sqrt(eps) or realmin, whatever seems most appropriate at the time. You can use the isnan function to detect the NaN values and replace them with the value you choose as appropriate, depending of course on the function you are using.