The ‘icdf’ function computes the inverse of a cdf function, which means that it receives a probability value or values and obtains the corresponding random value(s) in the indicated cumulative distribution. In your case you used values from ‘normrnd’ as the second argument instead of probability values. That is a mistake. Whenever your P values fall outside the range [0,1] as they are bound to do, you will get a NaN as a consequence. You need to rethink what you are trying to do. Look up the site:
and see what it really says.
Your statement "calculate the inverse of the cumulative normal distribution function of a set of random numbers" indicates a misunderstanding. To get the inverse of a cumulative distribution function you don't start with random number values, you start with cumulative probability values and obtain corresponding random variable ranges, and of course those probability values must always lie between zero and one.
