On Mon, 20 Aug 2012, Daniel Sirim wrote: Your mailer is producing an illegible mess. I've reformatted some of your functions below. A basic point is that gretl functions "know about" only those variables that you pass as arguments, or that are defined inside the function itself. # Basic Functions # Nr. 1 GO function series GO (scalar a, scalar b1, series v1, scalar t0) return a1*(1-exp(-b1*v1-t0)) end function The symbol "a1" is not defined -- perhaps you meant "a"? # Nr. 2 GOS function series GOS (scalar a, scalar b2, series v1, scalar t0) return a1*(1-(1+b1*v1-t0)*exp(-b1*v1-t0)) end function This time both "a1" and "b1" are undefined. The computer is not going to guess what you mean. # Nr.3 HD function series HD (scalar a, scalar b3, scalar c3, series v1, scalar t0) return a3*(1-exp(-b3*v1-t0))/(1+c3*exp(-b3*v1-t0)) end function Here "a3" is undefined. function series model (scalar a, scalar b1, scalar b2, scalar b3, scalar c3, series v1) scalar n1 = GO (t1, a, b1, v1, t) scalar n2 = GOS(t2, a, b2, v1, t)+n1 scalar n3 = HD (t3, a, b3, c3 , v1, t)+n2 return (t<=v1)*GO (t1, a, b1, v1, t) \ + (t>v1)*(t<=t2)*GOS(t2, a, b2, v1, t)+n1 \ + (t>v1)*(t<=t3)*HD(t3, a, b3, c3 , v1, t)+n2 end function In this function the undefined variables are t1, t2, and t3. Also you are doing something funny with the inequalities relating to "t" (which, unless it is redefined, gives you an index of the observations). That is, you are comparing it against both scalars (I think t1, t2 and t3 are supposed to be scalars) and a series, v1. You can do that, but I'm not sure it's what you want. Further on, we have catch nls v4 = model "model" is a function with 6 parameters, as written, but you're not passing any arguments. You should start with a much simpler problem, to get the hang of programming in general and gretl scripting in particular. It should be fairly obvious that none of the functions above will work as written. Allin Cottrell