sgnull / particle-simulation Goto Github PK

Having an FPS Counter somewhere on screen would be really helpful as a basic performance metric.

Had some bugs with drawing, so I tried turning wrap around off, and the performance difference was visible. I don't have any hard numbers on this yet, I just know that it's pretty bad.

This is a particularly interesting issue for me, though, because IIRC wrap-around was a natural consequence of the physics calculations the last time I tried this project, not some conditional distance vector alteration. Perhaps this is related to issue #9

It seems like, from the simulations I've ran, that the normal force actually pulls particles in instead of pushing them away. This should be a simple fix.

Something is definitely wrong with the component force equations, because collisions do not work like they should. Collisions cause particles to jitter a lot, move really fast, then eventually separate.

Solving this would solve issue #3. The impact is extremely significant; drawing results in over a 100% increase in simulation times. There are two options:

• Replace graphics.py
• Reduce the pixels graphics.py has to update

First I'm going to try the latter, then possibly work on the former. One alternative to graphics.py is pygame, but I've heard of bad performance from that package too.

Switching to Numpy would allow for easier vector math, making force calculations much more elegant. It might also result in them being faster.

Separating the interact_with_particle function from the Particle object would hopefully allow for some optimization. Instead of calling this function twice for both particles in the pie slice, just call it once on both particles and handle both of their interactions at once. The only difference between their calculations is the direction of the distance_vector and the net force.

This is related to issue #3

As of right now, the simulation can run with about 20-30 particles before some serious lag sets in. This is not a huge issue at the moment, as this is sufficient for testing purposes, but it will be eventually. We're shooting for 500+ particles here.

I suspect the problem lies with the graphics.py package, so I will need to look for a better package to use for this project.

Basically title. The particle interactions don't take into account the wrap-around setting. This seems to be resulting in particles gathering at the corners of the screen.

With physics implemented, it is time to address the elephant in the room. The performance was known to be poor from the beginning, but with the addition of physics this has become far more pronounced. This more than likely comes down to optimizations, but it could also have something to do with the graphics.py package.

To be more specific: 200 particles now seriously lags the simulation.

I was hoping that this behavior was gone with issue #5, but it seems like some of the wrap around calculations might be incorrect.