is phun/algodoo programed with a touch of random?

[izacque]

If not, why does a very thin box fall over when it touches the ground? or why do a bunch of circles in the exact same place explode?

[Mystery]

It is probably as a result of low simulation frequency, as, at a perfect precision simulation accuracy the simulation will probably run the exact same, as without perfect precision, things are going to hit other things at different speeds (things will go inside each other at faster speed).

Sim.timeDelta = 0.01 is the default, at 0.0000001 (10,000,000 hz) it would probably run perfectly, and the same each time, as the variance in each movement will be much smaller.

[izacque]

aha... so the "random" behavior is due to rounding...

[savask]

But why sometimes one complicated scene acts differently on a big number of tries?

[Mr. crash]

savask: same as zoomed in acts differently than zoomed out?

[savask]

The Bell Curve

conundrumer

Rating:
Filesize: 46.18 kB
Comments: 16
Ratings: 8

Ok, I'll show you example with the Bell curve.
If you will play this scene more times (start, undo, start, undo...), you will get different results.

[KarateBrot]

But why sometimes one complicated scene acts differently on a big number of tries?

it's because of the butterfly effect which is caused by the roundings gradyfitz already mentioned and maybe because of very small differences in initial conditions. algodoo doesn't calculate everything with perfect precision.

[savask]

But I don't think that there are other variants of rounding (or calculating). If 0.541~0.5, so it can't round to 0.6 on next try, for example.

[KarateBrot]

yeah and because of that there can also be small differences in initial conditions or some calculations are not syncronal to the frequency of 100Hz. for example a calculation starts at a very very small different point in time each time you start the simulation.

[savask]

Is frequency calculating from computer clock?
If yes, this mean that your explanation is right and when computer works slower, Algodoo may start calculation later than timer will count up.

[Mr. crash]

whats the real life frequency

[KarateBrot]

It's approximately 1.85*10^43 Hz. Every physics formula doesn't work with higher real life frequencies.

[KarateBrot]

Is frequency calculating from computer clock?
If yes, this mean that your explanation is right and when computer works slower, Algodoo may start calculation later than timer will count up.

I don't know but although the sim isn't running there are calculations and maybe thats why it's off-sync if you start the simulation not exactly at specific points in time.

whats the real life frequency

Maybe you didn't expect a serious answer but it's approximately 1.85*10^43 Hz Every physics formula doesn't work with higher real life frequencies.

[savask]

I don't know but although the sim isn't running there are calculations and maybe thats why it's off-sync if you start the simulation not exactly at specific points in time.

Your explanation is rather plausible, so I'll think it is right

[Mystery]

Isn't that the Planck time frequency?

[KarateBrot]

yeah that's it. i just made the reciprocal value of planck time to get the frequency.

[izacque]

what about circles in the same place?

the sim starts at exactly the same time, and their attributes would all have been rounded exactly the same...

[KarateBrot]

what about circles in the same place?

the sim starts at exactly the same time, and their attributes would all have been rounded exactly the same...

they behave the same every time i start the simulation. they fly apart and hit the ground at the same time and distance, they have got the exact same velocity (of course mirrored) but they do it the same every time i start the simulation. but maybe i can't see a difference because the scene is just too simple.

[izacque]

but why would they fly in different directions? shouldn't they all have the same velocity? (speed and direction)

[KarateBrot]

If they don't fly in different directions they should stay on each other. i don't know how it is programmed but they get pushed apart even if they have got the same position. I checked it. both have got the position [0,0]. they behave completely the same. the only difference is the velocity in x-direction after starting the simulation. it's predefined in which direction they get pushed. it's dependent from geomID or entityID i guess. it's necessary otherwise they would stay in each other.

[izacque]

... which strictly speaking would be more realistic...

anyway, I get it now. thanks

[KarateBrot]

hmm yeah maybe but you can't spawn two balls into each other in reality too so who knows