On January 6, 2004 06:19 pm, Allen H. Nugent wrote:
> Dear Fred,
Hi
> I have been struggling with this kind of problem for 2 months. DX does not
> seem to care much for closed surfaces (eg. {Connect} will fail because it
> expects an open surface with a single, all-purpose normal).
Yes, I agree connect will fail, but I suspect there is a way
to do this.
> I have been experimenting on a somewhat more complex geometry than your
> sounds, and I have just verified a new approach. I write a file that
> contains 3 sets of coords: 1 representing the surface, 1 representing an
> inner layer, and 1 representing an outer layer. The data component is then
> a flag: "0" for the actual surface, "1" for the inner layer, and "-1" for
> the outer layer. I then use {AutoGrid} to form a cubic grid and
> {Isosurface} (with <value> set to "0") to render the surface.
Right, I have a similar kludge that i have used. I have simply
created a function that depends on r linearly (r is the distance from the
origin) but has value 1 at the surface (i.e. it's gradient is theta and
phi dependent but is a constant with respect to r). I then pass my data
through it, and compute my kludge-function on a 3-d grid. I then use
this in opendx and display the isosurface whose value is 1.
Unfortunately, this is clearly inefficient and leads to much
larger datafiles.
I can get close to what I need. See
http://www.physics.utoronto.ca/~nastos/spinplots/spins3.gif
for example. I just want to be able to produce a "smoother"
surface.
> It sort of works, but is a bit patchy and glitchy, and wastes time and
> memory, so I'm trying to figure out where to go next.
>
> I think you can forget about an elegant, generic solution to this problem
> until some one writes such a module.
>
> I would be interested in the technique you used to form the connections,
> because I think native DX format will turn out to be the only way I can get
> satisfaction.
Let me get back to you on that off the list. I sort-off-lied in my
original post since I never got it to completely work. I then came
up with the kludge that I mention above, and used that. I won't be
able to get back to that problem until the weekend though.
> P.S. I'm a relative newby, so don't put too much weight on my advice.
Maybe someone else on the list can comment? I suspect this can done
through 'warp' or something similar, but I haven't been able to figure
that out.
Thanks for your input.
Cheers,
Fred
> Regards,
>
> Allen H. Nugent
> Graduate School of Biomedical Engineering
> University of New South Wales
> Sydney NSW 2052 Australia
> Tel: +61 2 9385 3916 Fax: +61 2 9663 2108
