[Nek5000-users] about fix_geom

[nek5000-users at lists.mcs.anl.gov]

Hello, Paul,

Thanks for your quick answer. So the original usage of this subroutine
is to correct the interface between two non-matching neighboring
elements. The way I understand is wrong. It functions in two steps, first
to average and second to propagate info from the rectified boundary.

However, from this post
https://lists.mcs.anl.gov/mailman/htdig/nek5000-users/2011-May/001319.html
I realize that it seems that fix_geom should do the job I want: after I 
change the
gll coordinates on the boundaries based on the Poisson solver, a call to 
fix_geom
will fix the gll points inside the elements, reflecting the 
corresponding boundary
displacements. Is there misunderstanding here?

thanks very much.

lailai


On 1/21/2014 3:22 PM, nek5000-users at lists.mcs.anl.gov wrote:
>
>
> Hi Lailai,
>
> fix_geom was intended to simply repair small rips in the
> mesh fabric, i.e., to ensure that there are no gaps between
> curved edges or corners that are slightly separated because of
> round-off errors in the .rea file, etc.  So, it's not surprising
> that there would be no change for a single element.
>
> The way fix_geom is intended to work is to simply average the
> values of x,y, and z that are shared between elements, and then
> to use Gordon-Hall to propagate the resultant changes into the
> interior of each element.   Periodic BCs are excluded from the
> averaging process via a mask.   If an element is unchanged by
> the averaging process then there is no change propagated into
> its interior.
>
> Hope this helps...
>
> Paul
>
>
>
> On Tue, 21 Jan 2014, nek5000-users at lists.mcs.anl.gov wrote:
>
>> Hello, guys,
>>
>> Have you any experience with the routine fix_geom, which has not been 
>> tested
>> for 2d cases as reported. I am trying to use it for a 2d case with 
>> curved
>> boundaries not necessarily describable by expressions. When the gll 
>> points on
>> bc are rectified to capture the right shape, a call to fix_geom is 
>> supposed
>> to reposition the internal gll points following the so-called 
>> Gordon-Hall
>> algorithm?
>>
>> I have tested for a simple mesh with only one square element, whose 
>> corners
>> are located at (-1,-1), (-1,1), (1,1) and (1,-1). I find that simply 
>> moving
>> the internal gll points of one edge (like (-1,-1) to (1,-1)) followed by
>> calling fix_geom does nothing to the internal gll coordinates. 
>> Probably some
>> key ingredients are missing here?
>>
>> thanks in advance,
>> lailai
>>
>>
>> _______________________________________________
>> Nek5000-users mailing list
>> Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users