Re: st: No convergance?

["Christer Thrane" <ch-thran@online.no>]

I use Intercooled Stata, and I don't think matsize is the problem

The ...

qreg earnings gender age agesq ... and so on

... just stops after about 60 iterations and states:

convergence not achieved
r(430);

The same happens after bsqreg and clad.

However, after

xtgls earnings gender age agesq ... and so on

I get the "matsizie to small message."


I will continue to investigate the matter. Perhaps I should install the SE 
version.


Thanks!

Chris
----- Original Message ----- 
From: "Andrei Malinovschi" <a_malinovschi@yahoo.com>
To: <statalist@hsphsun2.harvard.edu>
Sent: Friday, December 16, 2005 4:28 PM
Subject: Re: st: No convergance?



> Hi Christer,
>
> If you are looking at matsize help you can see that
> the 800 limitation is only for Intercooled STATA. For
> STATA/SE you can go up to 11000 (I have STATA 8.2, but
> I suppose that this is true also for STATA 9). It's
> not clear from your mail which version of STATA 9 do
> you have - Intercooled or SE.
>
> Hope this helps,
> Andrei
>
>
> --- Christer Thrane <ch-thran@online.no> wrote:
>
> > To my earlier posting, Josep replied:
> >
> > "I don't know about the robust regressions not
> > converging, but you can -set
> > matsize- to overcome the problem with the
> > matsize-too-small error."
> >
> > I have set the matsize to 800. As far as i know,
> > that's maximum.
> >
> > Chris
> >
> > ----- Original Message ----- 
> > From: "Joseph Coveney" <jcoveney@bigplanet.com>
> > To: "Statalist" <statalist@hsphsun2.harvard.edu>
> > Sent: Friday, December 16, 2005 1:52 PM
> > Subject: Re: st: No convergance?
> >
> >
> > > Christer Thrane wrote:
> > >
> > > I have a panel data set with approximately 5 000
> > persons surveyed in 12
> > > consecutive years totalling about 60 000
> > observation/years.
> > >
> > > If a regress earnings on gender, age, agesq, exp,
> > gpa, year (the panel
> > > identifier; i.e. the time dimension) plus a few
> > dummies for region with
> > > the
> > > cluster option [cluster(id)], Stata produces the
> > expected results results
> > > in
> > > about two seconds.
> > >
> > > If i ty to estimate:
> > >
> > > qreg
> > > bsqreg
> > > clad
> > >
> > > on the same variables (ignoring the cluster
> > option) I get:
> > >
> > > convergence not achieved
> > > r(430);
> > >
> > > I have tried to set # of iterations to 100, 500,
> > 800--nothing helps...
> > >
> > > If I try:
> > >
> > > xtgls
> > >
> > > on the same variables I get:
> > >
> > > matsize too small - should be at least 5185
> > > r(908);
> > >
> > > Any suggestions?
> > >
> > >
> --------------------------------------------------------------------------------
> > >
> > > I don't know about the robust regressions not
> > converging, but you can -set
> > > matsize- to overcome the problem with the
> > matsize-too-small error.
> > >
> > > It might be helpful if you posted your exact code.
> > >
> > > Joseph Coveney
> > >
> > > *
> > > *   For searches and help try:
> > > *
> > http://www.stata.com/support/faqs/res/findit.html
> > > *   http://www.stata.com/support/statalist/faq
> > > *   http://www.ats.ucla.edu/stat/stata/
> > >
> > >
> > >
> >
> >
> > *
> > *   For searches and help try:
> > *
> > http://www.stata.com/support/faqs/res/findit.html
> > *   http://www.stata.com/support/statalist/faq
> > *   http://www.ats.ucla.edu/stat/stata/
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/