Precision mistake after few time ?

What did you draw?
Post Reply
Arthur
Posts: 2
Joined: Tue Dec 27, 2016 2:42 pm

Precision mistake after few time ?

Post by Arthur »

Hello there !

So, I've just made a simple system where three massive stars are going down at a slow speed.
Two other planet are orbiting around these stars.
The system has been made so that it is perfectly symmetric for any given point of view.

However, after about 10 seconds at a time speed of 6, the orbiting planets start messing and taking a path completely non-symmetric from eachother.

I am running the version 0.32.04 on W7.

Has anybody experienced the same issue ? Is there a fix planned ?

Thanks ^^
User avatar
testtubegames
Site Admin
Posts: 1148
Joined: Mon Nov 19, 2012 7:54 pm

Re: Precision mistake after few time ?

Post by testtubegames »

Would you mind sharing the code for your setup (or some similar one that produces this problem?) I want to make sure we're seeing the same issue.

My sense is that this error you're running into has to do with floating point accuracy. Any calculation on a computer has a bit of error, and especially when you're running the system fast (at x6 speed, say), the minor difference might amplify pretty quickly. (Try running it at x10 or x100, and you'll see what I mean)

My guess is that this is what is happening here. If you run it slower, does the system remain symmetric better?

(Then again, I've heard of some problems on Win7 that are unlike any I've heard of... so I'm willing to believe you've discovered some novel bug)
Arthur
Posts: 2
Joined: Tue Dec 27, 2016 2:42 pm

Re: Precision mistake after few time ?

Post by Arthur »

Unfortunately, I couldn't give you the code for this system because after trying to copy & paste it on my own computer, all the planets and stars don't seem to be added properly (I've seen a similar issue in older topics). Running the time slower doesn't seem to prevent the bug from appearing.

I'll try to go a bit deeper in my tests to find a similar issue instead of relaying on floating-point mistake (which could be the problem anyways)
Post Reply