convergence criterion, EV8 vs EV9

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

maxchen
Posts: 191
Joined: Fri Oct 10, 2008 4:03 pm

convergence criterion, EV8 vs EV9

Postby maxchen » Thu Jul 21, 2016 6:55 pm

It seems the behavior of convergence criterion in EV8 and EV9 are different.

using the data file download/file.php?id=3371

Code: Select all

%path = @linepath
%wf = %path + "dsdf.txt"
pageload %wf

system bg
bg.append df = c(1)
bg.append ds = c(2)

bg.arch(conv=0.01,showopts) @ccc c arch(1) garch(1)


EV8 shows
Estimation settings: tol=0.01000, derivs=analytic
Initial Values: C(1)=9.35833, C(2)=9.22667, C(3)=171.133,
C(4)=0.54664, C(5)=0.36116, C(6)=115.245, C(7)=0.75962,
C(8)=0.32894, C(9)=0.99109
Convergence achieved after 33 iterations

EV9 gives

Estimation settings: tol=1.0e-06, derivs=analytic
Initial Values: C(1)=9.35833, C(2)=9.22667, C(3)=171.118,
C(4)=0.54662, C(5)=0.36120, C(6)=115.196, C(7)=0.75926,
C(8)=0.32925, C(9)=0.99109
Convergence achieved after 54 iterations

the "tol" is scaled.

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13305
Joined: Tue Sep 16, 2008 5:38 pm

Re: convergence criterion, EV8 vs EV9

Postby EViews Gareth » Thu Jul 21, 2016 8:32 pm

Yep
Follow us on Twitter @IHSEViews

EViews Glenn
EViews Developer
Posts: 2671
Joined: Wed Oct 15, 2008 9:17 am

Re: convergence criterion, EV8 vs EV9

Postby EViews Glenn » Fri Jul 22, 2016 10:00 am

Tols are scaled for new estimation methods since the interpretation in the old engine and new engine differ. If you want the original behavior, you can select the legacy engine.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 35 guests