[CIG-MC] Re: geoid scaling in CitcomS-2.2.1

Eh Tan tan2 at geodynamics.org
Tue Oct 30 22:26:51 PDT 2007

Hi Shijie,

Actually, in CitcomS-2.2, dv/v and dp/p are 1.0 in the first iteration. 
But for some reason, which I cannot recall, I chose to print dv/v=dp/p=0 
for the first iteration. This is the code at line 242, 

fprintf(stderr,"AhatP (%03d) after %6.2f s v=%.3e div/v=%.3e "
                 "dv/v=%.3e and dp/p=%.3e for step %d\n",
                 count, CPU_time0()-time0, 
                 0.0, 0.0, E->monitor.solution_cycles);

The 2nd and 3rd last arguments are hard-coded as 0. So, Scott's output 
is normal.

I agree with you that outputting dv/v=dp/p=1.0 for the first iteration 
will make more sense. I will change the code soon.


PS: No need to feel sorry when cc'ing cig-mc. I think cig-mc mailing 
list is a very appropriate place to discuss this issue. There are many 
aspects of CitcomS that is not properly documented or covered in the 
manual. Our public discussion can clarify things and may result in 
additional documentation, which is beneficial to the user community.

Shijie Zhong wrote:

>Hi Scott and Eh,
>I think that dv/v=0 and dp/p=0 for the first iteration of both time steps 0 and 1 are strange to me. This basically means that the stokes flow solver did not do anything to solve the equation. 
>In fact, dp/p=1 not 0, for first iteration, if you do not restart with any previous velocity and pressure (at least in some versions of CitcomS). dv/v=0 and dp/p=0 occurs only when the C-G search step (i.e., alpha) is zero. 
>Did you run the calculation with some restart? If so, I suspect that there may be some problems with the restart. 
>Sorry to bother cig-mc list -- I thought that the problem was a bit odd.  
>Shijie Zhong
>Department of Physics
>University of Colorado at Boulder
>Boulder, CO 80309
>Tel: 303-735-5095; Fax: 303-492-7935
>Web: http://anquetil.colorado.edu/szhong

More information about the CIG-MC mailing list