Re: [Gretl-users] Initial values for simulating an AR(p) process

On Fri, 2 Dec 2016, Artur T. wrote: [...] filter() only takes a scalar for pre-sample values. If you want to simulate an AR(p) with p>1 and p initial values fixed, I guess your best bet is to re-cast the model as a VAR(1) in companion form and then use varsimul(), as in <hansl> clear set echo off set messages off open denmark.gdt -q set seed 1234 series Y = LRY list lD = const # time scalar nD = nelem(lD) scalar p = 4 ols Y lD Y(-1 to -p) matrix ARbeta = $coeff[(1+nD):] matrix mu = $coeff[1:nD] series m = lincomb(lD, mu) matrix A = ARbeta' | (I(p-1) ~ 0) matrix y0 = mreverse({Y}[1:p])' matrix U = {resample($uhat) + m} ~ zeros($T, p-1) matrix S = varsimul(A, U, y0) series ysim = S[,1] gnuplot Y ysim --with-lines --time-series --output=display </hansl> ------------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Scienze Economiche e Sociali (DiSES) Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti -------------------------------------------------------