Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Need some help with nonlinear integration

  1. Nov 21, 2004 #1

    I study animation, and I wanted to make a script where I can automate pendulums in my scene with some kind of realistic physics, rather than animate them manually.

    So 'right' I think, 'I need to go study some physics and maths' - but it wasnt as easy as I hoped, Ive hit a wall, and it is the nonlinear integration of the second order D.E:

    y" = -(g/L)sin(y)

    So from my limited understanding, this equation is non-linear due to the absence of y', and because y is a function of sin.

    Its seems the only hints Ive been able to get is that integrating it is not a particually easy thing to do, with references only giving me parameters to enter into mathematics software for computation.The problem is, I essentially need to write my own software(script), and without understanding what is going on, I cant.

    If I could get some better idea of what is involved, then I can decide if its going to be worth my while spending more time trying to do this now, or if I need a diploma in advanced mathematics first :)

  2. jcsd
  3. Nov 21, 2004 #2


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Non linear problems are best dealt with numerically. If you must solve the nonlinear equation look into a Runga Kutta method. If you keep the oscillations of your pendulum small (< .2 rad or ~10deg) you can use the usual approach and let sin [itex] \Theta [/itex] = [itex] \Theta [/itex] then the problem is linear and easily solved.
  4. Nov 21, 2004 #3
    Thankyou Integral, I will look into the Runga-Kutta method.

    I would like to be able to calculate large angle pendulums, upto 180 degrees, so unfortunatly(for me) I cannot linearise the equation.
  5. Nov 21, 2004 #4


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member
    Dearly Missed

    You should, however, integrate once:
    Multiply your equation with y':
    Integrated from t=0 to some arbitrary t-value, you get:
    Last edited: Nov 21, 2004
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook