kurtz le pirate <kur### [at] gmailcom> wrote:
> Le 30/06/2018 à 14:15, Bald Eagle a écrit :
>
> > This is very nice!
> > How long did it take you implement the code?
> > I'm glad to see that you're another person playing with the coding challenge
> > topics   :)    There's lots of fun and educational projects there.
> >
> > (I still have to fix the last problem I had with converting the Perlin noise
> > flow field
> > Coding Challenge #24: Perlin Noise Flow Field
> > https://www.youtube.com/watch?v=BjoM9oKOAKY
> > )
> >
> > I've always wanted to play with some of these systems and have the colors change
> > as a function of tension, rotational or translational velocity, etc.
> >
> > 1000 frames is "a lot" (Shut it, Balaska)
> > How long to render the whole animation?
> >
> >
> >
> >
>
>
> Yes, Coding Challenge is a good source of inspiration...
>
> The code took me about 1 hour. The tricky part was the persistence of
> the data between images. Then, 1000 frames, 800x600 with anti-aliasong
> take 15mn.
>
>
>
> --
> Kurtz le pirate
> Compagnie de la Banquise

I'm guessing what you were passing between frames were the two angles and
angular velocities in order to do a numerical solution of the ODEs. However, I
think your numerical solution scheme is adding energy. You start off with the
balls both stationary and horizontal. Later in the motion BOTH balls are above
the horizontal. A numerical scheme to conserve energy is surprisingly difficult.
What scheme did you use?

Even more surprisingly, there are complex double pendulum motions that are
periodic, many that are symmetric, but several that are asymmetric. Would you
like some data for one (or more) of those motions?

Jim T.

Post a reply to this message