Dear All,

I tried Oleh's short script but did not find any crash under my win XP sp3. I am using gretl 1.10.2 (build date 2015-09-13, 32bit for windows). 
<hansl>
open okun.gdt
d_u = diff(u)
list lags_g =lags(4,g)

include armax.gfn
mmm = armax(2,2,d_u,lags_g,1,1,0,1,1)
</hansl>

<gretl output>
gretl version 1.10.2

Current session: 2015-09-24 09:52



? open okun.gdt



Read datafile C:\Program Files\gretl\data\PoE\okun.gdt

periodicity: 4, maxobs: 98

observations range: 1985:2 to 2009:3



Listing 3 variables:

  0) const    1) g        2) u      



? d_u = diff(u)

Generated series d_u (ID 3)

? list lags_g =lags(4,g)

Generated list lags_g

? include armax.gfn

 C:\Program Files\gretl\functions\armax.gfn

? mmm = armax(2,2,d_u,lags_g,1,1,0,1,1)

Generated matrix mmm

</gretl output>

I also tried Allin's script as follows. It also works fine.

<hansl>
nulldata 100
setobs 1 1 --special-time-series
set seed 77431
series y = normal()
loop p=1..2 -q
   loop q=1..2 -q
     catch arma p q ; y -q
   endloop
endloop
</hansl>

Those crashes seemed not appear under my XP.
Yi-Nnung

2015-09-23 14:22 GMT+08:00 <oleg_komashko@ukr.net>:
Dear Allin,
Doesh't crahes,
even with additional
modification below
doesn't

nulldata 100
setobs 1 1 --special-time-series
set seed 77431
series y = normal()
s = ""
loop p=0..2 -q
    loop q=0..2 -q
      catch arma p q ; y -q @s
    endloop
endloop

eval  $coeff

In my first example
on Ubuntu there was no
cases with err>0
so the dataset goes without catch
I removed catch: it crashes
In one-loop format -doesn't


Oleh








23 вересня 2015, 01:09:23, від "Allin Cottrell" <cottrell@wfu.edu>:

>   On Tue, 22 Sep 2015, oleg_komashko@ukr.net wrote:
>
> > Dear  Allin,
> > This script doesn't crashes
>
> Thanks, Oleh. So know we know, by the empirical method, that running
> a double loop over p and q (with a max of p=2, q=2) in arma
> estimation is not in itself a sufficient cause of a crash on Windows
> XP.
>
> The forensic method in software development often proceeds by
> seeking the minimal program that provokes a crash. So now we need to
> look a little further. Does it make a difference if the loops start
> at p=0, q=0, rather than p=1, q=1 as in my script (all else
> unchanged)?
>
> Thanks very much for helping to debug this issue.
>
> >> [C]ould I ask you just to confirm Yes or No, whether the script
> >> that I posted crashes on XP? Here it is again:
> >>
> >> nulldata 100
> >> setobs 1 1 --special-time-series
> >> set seed 77431
> >> series y = normal()
> >> loop p=1..2 -q
> >>     loop q=1..2 -q
> >>       catch arma p q ; y -q
> >>     endloop
> >> endloop
>
> Allin
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel@lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel


_______________________________________________
Gretl-devel mailing list
Gretl-devel@lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel