adaptative time step

[Pascal-56]

Hello,

I want to accelerate a simulation and for that I want to know there
exists an adaptative time step method for ODE.

Sincerely,
Pascal
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[Jon Watte (ODE)-2]

Pascal wrote:
> I want to accelerate a simulation and for that I want to know there
> exists an adaptative time step method for ODE.
>  

ODE does not adapt the time step. You can, however, pass in whatever
time step you want, and change it between invocations. You can also
change how many iterations are used by the over-relaxation method in
dWorldStepFast().

That being said, I would counsel against doing so. Simulations that come
to rest through forces (like gravity) being counter-acted by responses
(such as contacts) will suddenly become unstable when you change the
time step size during a running simulation, because of time-step
dependent inaccuracies in the integrator.

Sincerely,

jw

--

  Revenge is the most pointless and damaging of human desires.


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[Oleh Derevenko-2]

dWorldStepFast has already been removed in the latest trunk revision. Daniel
told, nobody should be using it.

Oleh Derevenko
-- Skype with underscore


----- Original Message -----
From: "Jon Watte" <jwatte@...>
To: <ode-users@...>
Sent: Monday, October 19, 2009 8:46 PM
Subject: [ode-users] Re: adaptative time step



Pascal wrote:
> I want to accelerate a simulation and for that I want to know there
> exists an adaptative time step method for ODE.
>

ODE does not adapt the time step. You can, however, pass in whatever
time step you want, and change it between invocations. You can also
change how many iterations are used by the over-relaxation method in
dWorldStepFast().

That being said, I would counsel against doing so. Simulations that come
to rest through forces (like gravity) being counter-acted by responses
(such as contacts) will suddenly become unstable when you change the
time step size during a running simulation, because of time-step
dependent inaccuracies in the integrator.

Sincerely,

jw

--

  Revenge is the most pointless and damaging of human desires.






--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[Daniel K. O.-2]

Jon Watte escreveu:
> dWorldStepFast().

You mean dWorldQuickStep().

Another downside about changing step sizes is that non-deterministic
simulations are a nightmare to debug.

--
Daniel K. O.
"The only way to succeed is to build success yourself"

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[bazkie]

i agree with the other replies - keep the stepsize constant unless you
want to have nightmares at night.
there's nothing stopping you from having more iterations of these
constant steps per 'reallife timestep' though, i guess..
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[Pascal-56]

If I choose a time step dependant on the force amplitude you think
this can be problematic for the ODE solver?
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---

[Remi Ricard]

Pascal wrote:
> Hello,
>
> I want to accelerate a simulation and for that I want to know there
> exists an adaptative time step method for ODE.
>  

Some months ago someone sent on the list a way to adapt the ERP and CFM
when the timestep change.
You might want to do a quick search.

Remi

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "ode-users" group.
To post to this group, send email to ode-users@...
To unsubscribe from this group, send email to ode-users+unsubscribe@...
For more options, visit this group at http://groups.google.com/group/ode-users?hl=en
-~----------~----~----~----~------~----~------~--~---