I have just joined the list, but searching the archive didn't give me a
clue so I thought I'd ask this list.

I have a finite element program that generates a scalar field on an
unstructured mesh. The solution is time dependent so I get results at a
number of positions in time. Following the User's Reference I have tried
to import these data as a group series into DX. After some trial and
error I finally got something that worked fine for 2D and 3D meshes.
Further DX hacking gave me something that I started to feel rather proud
of considering that I am a beginner.  Unfortunately I then started to
try more realistic cases. Not close to full scale, but at least a time
series with more than 4 points in time. This has not worked for me at
all. It seems that the number of positions can not even get close to
10 the way I have been generating my data.

This could very well be a problem with my netCDF format, but since it
worked so nicely for me for small cases I tend to think I have it right
now. 

If anyone has anything wise to way about this I'd be greatful. So as not
to pester everyone on the list I have left the details for the
interested reader at the URL http://parallel.hpc.unsw.edu.au/%7Esmeds/OpenDX/

My netCDF files are structured as outlined below. I have tried the
binary downloads from www.opendx.org. The versions I've tried are
Linux-4.1.1, Linux-4.1.3 and IRIX-4.1.1d. The IRIX and the Linux version
show the same behvaiour so it is not lack of memory that is the problem.
There is plenty of it on the SGI. The Linux 4.1.1 version seem to be
able to take more of the files I've generated than does the 4.1.3
version.

An error message from the IRIX version has gotten me to suspect that
there may be a problem when the import module reads my global
attributes. They get rather lengthy as the number of iterates (time
steps) increases. But this is a mere speculation. I have not yet
downloaded the source code to have a closer look.

Any insight that anyone can provide is appreciated

/Nils

netcdf Expand2D-longseries {
dimensions:
        pointnums = 221 ;
        sides = 4 ;
        quadnums = 192 ;
        axes = 2 ;
variables:
        float locations(pointnums, axes) ;
        int eldefs(quadnums, sides) ;
        float U_000(pointnums) ;
                U_000:positions = "locations" ;
                U_000:connections = "eldefs, quads" ;
        float U_001(pointnums) ;
                U_001:positions = "locations" ;
                U_001:connections = "eldefs, quads" ;
[...]
        float U_EXACT_030(pointnums) ;
                U_EXACT_030:positions = "locations" ;
                U_EXACT_030:connections = "eldefs, quads" ;

// global attributes:
                :series_u_soln = "U_SOLN; U_000, 2.5000E-01; U_001, 3.7500E-01; 
U_002, 5.0000E-01; U_003, 6.2500E-01; U_004, 7.5000E-01; U_005, 8.7500E-01; 
U_006, 1.0000E+00; U_007, 1.1250E+00; U_008, 1.2500E+00; U_009, 1.3750E+00; 
U_010, 1.5000E+00; U_011, 1.6250E+00; U_012, 1.7500E+00; U_013, 1.8750E+00; 
U_014, 2.0000E+00; U_015, 2.1250E+00; U_016, 2.2500E+00; U_017, 2.3750E+00; 
U_018, 2.5000E+00; U_019, 2.6250E+00; U_020, 2.7500E+00; U_021, 2.8750E+00; 
U_022, 3.0000E+00; U_023, 3.1250E+00; U_024, 3.2500E+00; U_025, 3.3750E+00; 
U_026, 3.5000E+00; U_027, 3.6250E+00; U_028, 3.7500E+00; U_029, 3.8750E+00; 
U_030, 4.0000E+00" ;
                :series_u_diff = "U_DIFF; U_DIFF_000, 2.5000E-01; U_DIFF_001, 
3.7500E-01; U_DIFF_002, 5.0000E-01; U_DIFF_003, 6.2500E-01; U_DIFF_004, 
7.5000E-01; U_DIFF_005, 8.7500E-01; U_DIFF_006, 1.0000E+00; U_DIFF_007, 
1.1250E+00; U_DIFF_008, 1.2500E+00; U_DIFF_009, 1.3750E+00; U_DIFF_010, 
1.5000E+00; U_DIFF_011, 1.6250E+00; U_DIFF_012, 1.7500E+00; U_DIFF_013, 
1.8750E+00; U_DIFF_014, 2.0000E+00; U_DIFF_015, 2.1250E+00; U_DIFF_016, 
2.2500E+00; U_DIFF_017, 2.3750E+00; U_DIFF_018, 2.5000E+00; U_DIFF_019, 
2.6250E+00; U_DIFF_020, 2.7500E+00; U_DIFF_021, 2.8750E+00; U_DIFF_022, 
3.0000E+00; U_DIFF_023, 3.1250E+00; U_DIFF_024, 3.2500E+00; U_DIFF_025, 
3.3750E+00; U_DIFF_026, 3.5000E+00; U_DIFF_027, 3.6250E+00; U_DIFF_028, 
3.7500E+00; U_DIFF_029, 3.8750E+00; U_DIFF_030, 4.0000E+00" ;
                :series_u_exact = "U_EXACT; U_EXACT_000, 2.5000E-01; 
U_EXACT_001, 3.7500E-01; U_EXACT_002, 5.0000E-01; U_EXACT_003, 6.2500E-01; 
U_EXACT_004, 7.5000E-01; U_EXACT_005, 8.7500E-01; U_EXACT_006, 1.0000E+00; 
U_EXACT_007, 1.1250E+00; U_EXACT_008, 1.2500E+00; U_EXACT_009, 1.3750E+00; 
U_EXACT_010, 1.5000E+00; U_EXACT_011, 1.6250E+00; U_EXACT_012, 1.7500E+00; 
U_EXACT_013, 1.8750E+00; U_EXACT_014, 2.0000E+00; U_EXACT_015, 2.1250E+00; 
U_EXACT_016, 2.2500E+00; U_EXACT_017, 2.3750E+00; U_EXACT_018, 2.5000E+00; 
U_EXACT_019, 2.6250E+00; U_EXACT_020, 2.7500E+00; U_EXACT_021, 2.8750E+00; 
U_EXACT_022, 3.0000E+00; U_EXACT_023, 3.1250E+00; U_EXACT_024, 3.2500E+00; 
U_EXACT_025, 3.3750E+00; U_EXACT_026, 3.5000E+00; U_EXACT_027, 3.6250E+00; 
U_EXACT_028, 3.7500E+00; U_EXACT_029, 3.8750E+00; U_EXACT_030, 4.0000E+00" ;
}

-- 
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
   Nils Smeds                        http://www.pdc.kth.se/
   Center for Parallel Computers     e-mail: [EMAIL PROTECTED]
   Royal Institute of Technology     Voice:  +46-8-7909115
   KTH                               Fax:    +46-8-247784 
   S-100 44 Stockholm, Sweden        Office: OB2, room 1546
-----------------------------------------------------------------------

-- 
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
   Nils Smeds                        
   HPCSU / Sci & Techn.              e-mail: [EMAIL PROTECTED]
   UNSW, NSW 2052                    Voice:  +61-2-9385 6915
   Sydney                            Fax:    +61-2-9385 ----
   AUSTRALIA                         Office: Red Centre, West, Rm 2075
-----------------------------------------------------------------------

Reply via email to