[Ifeffit] A very quick question

Hello, I would like to ask a very quick question and I would appreciate if someone can help me. While running XANES fitting for mixed valence state metals such as cerium (3/4), should I use two arctangent functions with selected group of Lorentzian functions or I have to use only one arctangent function with a group of Gaussian functions? I would appreciate if someone give some advice. Yours, Ahmed