[vtkusers] LS-Dyna

[John Platt]

Hi Ian,

I am not familiar with AVS/Express so I cannot comment on how this code
would have handled the issues I listed (item 6 can be managed by VTK if
you use assemblies). If you approach VTK with the attitude that it is a
class library not to be changed, you will be challenged on some tasks.
On the other hand, if you are prepared to write some new filters or
specialise existing filters, life can become much easier and the effort
rewarding. 

Of course, the VTK List is almost exclusively problems. You don't hear
about the 80% which is a breeze. Also, my problems are particular to FE
analysis and I don't think VTK evolved from that background.

To answer your question, I develop code for solving coupled acoustic
problems in water and air using FE & BE and thin fluid films on flexible
structures (lubrication) - a rather esoteric mix!

John.


-----Original Message-----
From: Ian Curington [mailto:ianc at acm.org] 
Sent: 15 July 2004 14:27
To: jcplatt at lineone.net
Subject: re: [vtkusers] LS-Dyna

Dear John,

I can't really make non-vtk comments to the list, but
I find this response quite revealing. 
I'm working with both VTK and AVS/Express.
I am used to building apps dealing with complex data structs
like FEA, CFD, Oil Res, etc. with AVS/Express, which
has solved ALL 6 of your issues at least 5 or 6 years ago.
I still have a bit of hair left after the experience...

However, VTK/Paraview wins on flexible open source, XML, parallel
and large data methods, choice of GUIs.
I am sure VTK is catching up fast with AVS/Express, for which
development has slowed down to a crawl.

What kind of work are you involved in?

-Ian

In message John Platt <jcplatt at lineone.net> writes:
>Here are some areas in vtk 4.4 which may cause some hair loss
>
>1. Display of discontinuous node data.
>2. Quadratic elements (although I seem to recall that explicit codes
>tend to use lots of linear elements).
>3. Simultaneous display of faces and edges without duplication of point
>and scalar data.
>4. Filters which will load scalar/cell data on demand - particularly
for
>transient work.
>5. Coincident faces from 2-D & 3-D elements and merging of coincident
>points.
>6. Dealing with cells which don't have the selected result type.
>
>At this point I am hoping that Mathieu will respond with done, done,
>done, ... :-)
>
>I still have some hair left so it can be done but not in a week!

-- 
-- Ian Curington -- ianc at acm.org  ---