[Carpet] time_refinement_factors and prolongation order

[Erik Schnetter]

On Aug 31, 2006, at 10:04:11, Michael Koppitz wrote:

> On the carpet mailing list archive
> I found a setup ment to be used
> for runs without time refinement.
> http://lists.carpetcode.org/archives/developers/2006-July/001732.html
>
> I tried this (and other confiogurations and end up with an error:
> "Assertion `prolongation_order_time == 2' failed"
>
> Using prolongation_order_time =2 I get an error:
> Internal error: extrapolation in time.  time=2  times=[1,0,-1]
>
> maybe the two are not connected, but they sure look like they are.
>
> Could somebody suggest a remedy?

I think I answered this question already in private.  Please do not  
send private emails to me if you ask the question later on the list  
-- instead, send questions directly to the list, so that they are  
publicly archived for all to see.


These settings were suggested by Yosef.  As Yosef already said in  
this email, these settings do not work.  As I found out and (so I  
hope) described in a later posting on the list, you should use the  
following settings:

prolongation_order_time = 1   (because you will likely be using 2  
time levels, which will use linear time interpolation when regridding)

init_3_timelevels = no   (since you do not need to initialise 3 time  
levels if you do not interpolate in time, and since the three  
timelevel initialisation scheme does not work with a time refinement  
factor of unity)

Alternatively you can use three time levels and  
prolongation_order_time = 2.



With mesh refinement, all time levels need to be initialised, unless  
you do not interpolate in time.  Keep in mind that regridding needs  
to interpolate in time, even if time evolution does not.  You should  
therefore use e.g. init_all_timelevels or MoL::initial_data_is_crap,  
unless you can guarantee that you do not change the grid hierarchy  
during the first few coarse grid time steps.

-erik

-- 
Erik Schnetter <schnetter at cct.lsu.edu>

My email is as private as my paper mail.  I therefore support encrypting
and signing email messages.  Get my PGP key from www.keyserver.net.



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
Url : /archives/developers/attachments/20060831/535f8f7c/attachment.pgp