[LORENE] strange GRV2/3 behaviour (Newtonian)

Eric Gourgoulhon eric.gourgoulhon at obspm.fr
Wed Nov 19 23:12:12 CET 2003

Dear Reinhard,

Many thanks for having pointed this out. We haven't
noticed this yet (actually we haven't much practice
with Newtonian configurations...).

The strange behaviour that you have noticed was due
to the conjunction of two facts:

1/ the relaxation of logn and dzeta in Etoile_rot::equilibrium
   (parameter relax in the file parrot.d)
2/ the very fast convergence of static Newtonian configuration

Indeed, because of the relaxation, the gravitational potential
logn keeps some memory of its initial (very crude !) stage
and this spoils the final GRV2 and GRV3 if the number of steps
is very small, as in static Newtonian case.
You can recover a very good GRV2 and GRV3 by setting relax=1
(i.e. no relaxation), or by forcing the code to run for
a large number of steps (setting mer_max=100 and precis=1.e-20
in parrot.d).

In order to correct this behaviour, without setting manualy
relax=1 or running for a large number of steps,
I've modified the functions


so that the relaxation is started only at the 10th step.
Therefore, if you perform a cvs update, the problem you have
mentioned should have disappeared.

Best wishes,


Reinhard Prix wrote:

>I've just noticed a funny feature of the codes Rotstar and Sfluide in
>the Newtonian case:
>we had noticed a while ago that for very low rotation rates or static
>cases the virials in the Newtonian case become quite "bad", i.e. of
>the order of 1e-5 for a precision of 1e-10, while in the relativistic
>case one consistently gets GRV <= 1e-9. 
>However, if you set freq_ini_si=100 for example, even if freq_si=0,
>then you seem to get a solution that has GRV~1e-9...! 

Eric Gourgoulhon
Laboratoire de l'Univers et de ses THeories (LUTH)
UMR 8102 du CNRS / Observatoire de Paris, F-92195 Meudon Cedex, France
tel: +33 (secretariat : +33
e-mail: Eric.Gourgoulhon at obspm.fr    WWW: http://www.luth.obspm.fr/

More information about the Lorene.list mailing list