[ODE] New to ODE

Lewis Foster doof205 at gmail.com
Tue Nov 7 10:37:13 MST 2006


Hi David,

Thanks for pointing out a couple of ideas.

To be honest I don't think I could manage either of those having little to
no ODE experience at the minute. The one thing I do have is time to learn
though.

I've spoken to my lecturer and he suggested working with ODE to reduce
effects such as objects bouncing around long after the point at which they
would realistically have settled down. Like when you have a big pile of
objects. The problem I see is that the disable function and auto-disable
seem to have problems like this already under control, not a problem for
ODE, just for me :-)

Other than that I have been thinking about other modifications:

ODE calculates everything based on a time period since the last update
right? Perhaps I could update a single object many times, then move to
another and update several times. In esscence just updating more times per
object, rather than just once. Could improve speed?

I'm not sure if this is already in place but how about a value of
significance for objects. Objects with higher significance could be modelled
more accurately whereas insignificant objects are not so important. Disable
could be used to reduce calculations.

All of these ideas feel to me to be too shallow. I'm not even sure they
would require modifications of ODE code, rather just using the code and for
someone who is familiar with ODE it's probably somewhat trivial?

Thanks again for the reply,

Lewis

On 10/21/06, david moloney <moloned at gmail.com> wrote:
>
> Hi Lewis,
>
> some interesting work that could be done one ODE would be the following:
>
>    1. Porting of the solver to a GPU similar to whats being done in
>    Bullet
>    2. Modification of the solver to use lower precision (32 or even
>    16-bit openEXR arithmetic) by using an inner low precision and outer high
>    precision loop in a manner similar to that used by Goeddeke et al. http://numod.ins.uni-bonn.de/research/papers/public/GoStTu05double.pdf
>
>
> I think either of these topics would make for very novel and interesting
> work and suitable for a thesis.
>
> Regards,
>
> - David
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://q12.org/pipermail/ode/attachments/20061107/8cb11d35/attachment.htm


More information about the ODE mailing list