>Perhaps it should be sloppier still. But gretl is open source, no 
> need to guess what we're doing or not doing!

That's why I asked
With my "knowledge" of C it coulld have taken
hours to extract it from code
Dear Aliin,
I think, for starting values, something like 10^-5 is enough
NLS must not converge at the same point as MLE:
that's why attaining true convergence for NLS
for obtaining initial values is aimless exercise 

May I ask: why Hannan Rissanen method used only q+Q >=2
For q=1 even 1-st IGLS iteration I applied gives much better
results for arma(0,1), arma(1,1) for arma(2,1) the gain is moderate
It seems with growing p several IGLS iterations needed , or Hannan Rissanen 

Also if tрis is not convincing I even do not know what to do next

On Sat, 24 Nov 2018, oleg_komashko@ukr.net wrote:

> Dear Allin,
> It seems while tolerance for Hannan - Rissanen
> was resetted, this is not so for models that
> require non-linear autoregression:
>
> open AWM.gdt --quiet
> logs TFT
> arima 3 0 0;1 0 0; TFT --verbose
> arima 3 0 0;1 0 0; l_TFT --verbose
>
> produces
>
> ? open AWM.gdt --quiet
> Read datafile /usr/local/share/gretl/data/misc/AWM.gdt
> ? logs TFT
> Warning: generated missing values
> ? arima 3 0 0;1 0 0; TFT --verbose
> NLS: failed to converge after 1204 iterations

We are in fact setting a sloppy tolerance when we use NLS in 
arma_init, since quite recently -- see arma_init.c, line 1401:

libset_set_double(NLS_TOLER, 1.0e-5);

That can be compared with the default, namely pow(DBL_EPSILON, .75), 
which is about 1.82e-12.

Perhaps it should be sloppier still. But gretl is open source, no 
need to guess what we're doing or not doing!

Allin