[petsc-users] DMPlexCreateGmsh

[Gorman, Gerard J]

Hi Matt

First - really sorry for having inflicted this on you. I agree that there are many problems with the format and it’s not like the world needs a new file format. I only happened to use it for a case where there was an existing toolchain…which I’m now rewriting anyhow.

What is your current least detested file format for FEM that plays nicely with DMPlex? Exodus II?

Dharmendar - you might just want to write a simple translation script if you are locked into gmsh for some reason. It is also a pain that the format does not work with paraview and other common tools out of the box.

Cheers
Gerard


On 11 Jan 2014, at 01:14, Matthew Knepley <knepley at gmail.com<mailto:knepley at gmail.com>> wrote:

On Fri, Jan 10, 2014 at 3:44 PM, Dharmendar Reddy <dharmareddy84 at gmail.com<mailto:dharmareddy84 at gmail.com>> wrote:
Hello,
        I have a question and feature request regarding DMPlexCreateGmsh.

Gmsh, mesh format outputs, physical regions in the mesh as
For example in the a particular 2 d mesh with eight physical regions
it may look like the one below
$PhysicalNames
8
1 1 "RegionName1"
1 2 "RegionName2"
1 3 "RegionName3"
2 4 "RegionName4"
2 5 "RegionName5"
2 6 "RegionName6"
2 7 "RegionName7"
2 8 "RegionName8"
$EndPhysicalNames

It would be nice to have the DMPLexCreateGmsh process this information.

The first number in each row of name is the topological dimension (tdim),

If the code can group all the regions with dimension < mesh dimension
(meshDim) as "Boundary"  and all regions with dimension =
meshDimension as "Region"

Gmsh outputs (if requested) the lower dimension cells in regions with
tdim  < meshDim. If interpolate mesh is enabled, i would like to have
cells of boundary region, which in the above example will be lines and
points, included in the boundary strata with global numbering for the
facets.

I really hate GMsh right now. The format is a complete cluster of idiocy. They allow
outputting facets as separate lower-dimensional cells, unconnected to the original
cell, so we would have to initiate a search for every insertion. Not to mention that sizes
are not declared up front so that we have to run through the file in multiple passes. I
really really hate these guys. Is there an overwhelming reason to waste my time messing
around with a format designed by the mentally infirm?

I am willing to listen to reason, but reading Gmsh files would make anyone unreasonably angry.

    Matt

This ways the total number of Cells in the mesh will be : NumNodes +
NumLines+ NumTriangles or NumNodes+ NumTriangles (if interpolate is
off)


In the Current implementation, correct me if i am wrong, I see the the
total number of cells = numNodes + (some Lines which are elements of
the lower dimensional regions) + numTriangles


Thanks
Reddy


--



--
What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140113/43cdff48/attachment.html>