[Algorithms] (no subject)

[Steven W. <St...@eu...>]

OK, this is my 2peneth on LCP's. 

An LCP is a known mathematical problem, it has a know mathematical solution
(the big matrix solution) and handily it is a logically equivalent problem
to that of solving collision/constraint forces for a group of bodies. As the
problems are equivalent there will be an equivalent to the big matrix
solution working within our original problem domain BUT we don't know what
it is and it may be very complex to execute. So we can translate our
original problem domain into that of an LCP, use the known mathematical
solution and translate it back. Jobs a goodun'. 

 

However if we start to say we are using the iterative solution on the LCP
rather than big matrix, we are suddenly not doing such complex maths on it
at all and I think we COULD work out what this means in our original problem
domain. If so would it not be better to do the work in the original problem
domain (remember the two domains are equivalent) rather than wasting time
translating into an equivalent (LCP) domain and back again.

 

It MIGHT be that impulse constraints are infact already equivalent to the
iterative solution to the LCP and the results of Dirks solution will be
identical to an iteratively solved LCP except with (logically) less
overhead.  

 

- Steven

 

- covers head and waits to be shot down ;)

 

 

>3) When you say LCP solver, you mean big matrix solver right? I've mostly

>been using iterative solver so far and the results look similar to those in

>your demo. Do you see any benefits of this system over iterative LCP
solver,

>for NextGen specifically?

>

>Thanks,

>Alen

 

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

This e-mail is confidential and may be read only by the intended recipient.
If you are not the intended recipient, please do not forward, copy or take
any action based on it and, in addition, please delete this email and
inform the sender.
We cannot be sure that this e-mail or its attachments are free from
viruses.  In keeping with good computing practice, please ensure that
you take adequate steps to check for any viruses.  Before replying
or sending any email to us, please consider that the internet is inherently
insecure and is an inappropriate medium for certain kinds of information.
We reserve the right to access and read all e-mails and attachments
entering or leaving our systems.