[ODE] ODE spatial inconsistency
Marcin Pilat
pilat at cpsc.ucalgary.ca
Wed Jun 20 14:38:15 MST 2007
Actually, that's what I thought at first, but after doing some tests,
it doesn't seem like it is floating point problems (unless they are
horribly implemented inside ode). I posted some data in reply but
the message was too big and it is still awaiting approval. You can
see the graphs from a run of the code above with changing x from
-100.0 to 100.0: http://galapagos.cpsc.ucalgary.ca/~marcin/
odetestx.jpg (for x), http://galapagos.cpsc.ucalgary.ca/~marcin/
odetesty.jpg (for y), http://galapagos.cpsc.ucalgary.ca/~marcin/
odetestz.jpg (for z). If these changes are due to floating point
problems then why aren't they consistent - i.e. why are there
plateaus in the graphs with the same values? It doesn't sound right
to me. Any idea of what the problem and/or solution to this would be
(independent simulations or running everything at the same coordinate
works for some simple simulations, but in general is not a practical
option).
Thanks!
Marcin
More information about the ODE
mailing list