Kamaelia.Support.Particles.ParticleSystem

• Discrete time particle physics simulation
  • Example Usage
  • How does it work?

Discrete time particle physics simulation

A discrete time simulator of a system of bonded and unbonded particles, of multiple types.

The actual physics calculations are deferred to the particles themselves. You can have as many, or few, spatial dimensions as you like.

Example Usage

Create 3 particles, two of which are bonded and move noticeably closer after 5 cycles of simulation:

>>> laws = SimpleLaws(bondLength=5)
>>> sim = ParticleSystem(laws)
>>> sim.add( Particle(position=(10,10)) )
>>> sim.add( Particle(position=(10,20)) )
>>> sim.add( Particle(position=(30,40)) )
>>> sim.particles[0].makeBond(sim.particles, 1)   # bond 1st and 2nd particles
>>> for p in sim.particles: print p.getLoc()
...
(10, 10)
(10, 20)
(30, 40)
>>> sim.run(cycles=5)
>>> for p in sim.particles: print p.getLoc()
...
[10.0, 13.940067328]
[10.0, 16.059932671999999]
[30, 40]
>>>

How does it work?

Set up ParticleSystem by instantiating, specifying the laws to act between particles and an (optional) set of initial particles.

Particles should be derived from the Particle base class (or have equivalent functionality).

Particles can be added or removed from the system by reference, or removed by their ID.

ParticleSystem will work for particles in space with any number of dimensions - so long as all particles use the same!

Bonds between particles are up to the particles to manage for themselves.

The simulation runs in cycles when the run(...) method is called. Each cycle advances the 'tick' count by 1. The tick count starts at zero, unless otherwise specified during initialization.

The following attributes store the particles registered in ParticleSystem: - particles -- simple list - particleDict -- dictionary, indexed by particle.ID

ParticleSystem uses a SpatialIndexer object to speed up calculations. SpatialIndexer reduce the search space when determining what particles lie within a given region (radius of a point).

If your code changes the position of a particle, the simulator must be informed, so it can update its spatial indexing data, by calling updateLoc(...)

The actual interactions between particles are calculated by the particles themselves, not by ParticleSystem.

ParticleSystem calls the doInteractions(...) methods of all particles so they can influence each other. It then calls the update(...) methods of all particles so they can all update their positions and velocities ready for the next cycle.

This is a two stage process so that, in a given cycle, all particles see each other at the same positions, irrespective of which particle's doInteractions(...) method is called first. Particles should not apply their velocities to update their position until their update(...) method is called.

---