Re: [Gretl-devel] autoregressive "genr" bug

Am 11.06.2018 um 18:10 schrieb Riccardo (Jack) Lucchetti: > On Mon, 11 Jun 2018, Allin Cottrell wrote: > >> I've just discovered something that I think is a bug, though it can also >> be a convenience in some cases. I'd be interested to hear what people >> think about it. >> >> Simplest case: >> >> series u = normal() >> u = 0.5 * u(-1) >> >> The thing is that after the second line is executed, the first value of u >> is not NA, it's just the original value. > > I've always considered this as a very nice feature. Suppose you want to > generate a random walk as a discrete realisation of a Brownian motion. > > series x = 0 > x = x(-1) + normal() > > I _want_ the first value to be 0, because that's what Wiener processes do. > I'd be rather annoyed if my initialisation was bypassed silently. I was going to write the same thing. The initial value is absolutely essential in time series, can't just overwrite it with NA. This would also break a lot of code, and I absolutely don't think it's a bug.