• Advertisement
Sign in to follow this  

Kutta-Runge on the Trial

This topic is 4682 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Can any body present his impression about Kutta-Runge method compared mainly to the verlet integration method

Share this post


Link to post
Share on other sites
Advertisement
haven't used verlet, but used RK4 . In any case, theoretically and practically, it is reasonable to implement simulator so you'll be able to switch from one simulation method to other, or ideally, use different methods for different things. Runge-Kutta is avaliable in different forms, so if it gets unstable you can just switch to more stable one (from RK4 to RK6).

Share this post


Link to post
Share on other sites
Quote:
Original post by arithma
Can any body present his impression about Kutta-Runge method compared mainly to the verlet integration method


Verlet requires only one derivative computation per step, as opposed to RK4 (for example), which takes four (this may or may not be more expensive, depending on the timestep(s) that can be used for the simulation).

For cloth or molecular dynamics, Verlet is hard to beat in terms of stability and performance.

For stiff springs (as opposed to Verlet-style "rigid constraints"), RK is more stable, but loses energy over time (may or may not be a problem, depending on application).

RK2, or Midpoint, takes two derivatives per step, and can work well in many applications. I have found that RK has improved accuracy for high angular velocity rigid body simulation (proper simulation of precession, stability).

Symplectic Euler takes one derivative per step and is more stable than Forward Euler. Verlet is also a Symplectic method.

See:
Spring Simulation Comparison (Java)
Modern Methods (2005)

Share this post


Link to post
Share on other sites
Be aware though that RK4 is harder to implement for systems which aren't isolated or entirely known before hand. For example, if you use RK4 to integrate rigidbody motion where the forces and torques which act on the body can have dependancies on the body position/orientation and which aren't known prior, you'll have to link your rigidbodies to the force/torque equations via function pointers or some similar method so they can be called with each RK step.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement