There are a few combinations of values for which your function does not define the output argument. If you call the function with those values, it will throw an error becase the output is undefined. For example, consider y2==0 and x2==0. It might be that you expect the code evaluation to "fall through" like it does in some languages, but in MATLAB only one block of an if-elseif-...-else block will be executed, so once y2==0 is matched then that block will be executed, and if x2==0 then you do not define theta. Thus the error.
Tip1: note that the MATLAB approach would use simpler vectorized code, e.g. based on modulo and other vectorized operations. While your approach works for individual values, you should learn how to write code that handles entire vectors/matrices/arrays of data:
Tip2: numbering variables like that is a sign that you are doing something wrong. The MATLAB approach is to use indexing, which is automatically expandable, generalizable, does not require copy-and-paste of code, and makes both looping and code vectorization easy.
