Solar System and All Moons - Moons fly off everywhere

[snotvomit]

I have been trying to run the simulation called "Our Solar System - all known moons".

No matter whether I set the simulation to accuracy mode or not, as soon as I increase the speed of the simulation, a bunch of moons for Jupiter, Saturn and Uranus fly off in all directions - they don't orbit at all.

I imagine they would still fly off at the speed that the simulation starts, but that's too slow to see anything happening.

Does anyone else have this problem with the simulation?

I am running the STEAM version of US.

[Naru523]

Increasing the timestep too high can cause objects to fly off.

[snotvomit]

I suspected so, but I am not increasing particularly high.  The problem is reapeatable just by setting 1 second = 5 days.

This isn't super-fast.

Please note that I am setting it to 1 second = 5 days, not 1 frame = 5 days.

Lots of moons fly off.

A few of moons fly off even when the setting is 1 second = 3 days.

The problem is, there's no indication in the simulation of what the "safe" rate is.  I'm concerned that the safe rate is so low that I can't really do the sandbox experiments I was hoping for because it will take hours to see the effects.

[atomic7732]

You see the green dot next to the timestep? If it goes red you're bound to get flying things.

Alot of the moons take <5 days to orbit, so if you're running 30 fps, and it's taking 1 day to orbit, you're getting 30 calculations for the circle running 1 second = 1 day, that's not even a circle, that's a polygon. And you're getting a decagon for 1 sec = 3 days. I'm afraid you can't do the simulations you want to do.

There is a program I know that will allow you to go faster and it will work (it does all the calculations but doesn't show the frames until it's gotten far enough), and it has a 15 day trial... But it's over $60.

[snotvomit]

The problem is that the green dot doesn't go red for the settings I gave.  The green dot remains green and the moons still fly off.

This means that I only find out what "too fast" a rate is when the moons lose orbit, and then I have to start again.  Simply increasing the time rate wrecks the simulation.

It's a shame.  The game is still great, but I wish there was some safety method to make sure things like this didn't happen.

I haven't found a way to rewind time within the simulator, so it's disappointing when the whole simulation can be wrecked by trying to "fast forward" - and not mega fast forward.

I'll just have to work around it.

[Dan Dixon]

I understand your frustration.

In simulating gravity there's no way to add a fast forward feature that doesn't decrease the accuracy of the simulation (which is why everything flies apart: inaccuracy)

In future versions I hope to better communicate this reality and provide better tools for controlling the time step.

[Eptesicus]

What about a kind of snapshot-rewind feature, where it marks every n frames to restore to in case you do mess things up? ...Obviously a full rewind would be out of the question due to memory, and running the physics backward would be just as inaccurate at high speed.

[yaang]

I'd rather get less frames per second than my refresh rate than getting false results. Can you make it that way with the accuracy mode enabled ?

Or another suggestion:

Can you put a Jump to : X years later option so when you press it the program calculates it properly and the whole systems jumps to that time without displaying the frames in between.

[de_Stan]

Can you put a Jump to : X years later option so when you press it the program calculates it properly and the whole systems jumps to that time without displaying the frames in between.

This.. i was about to post about that. That would be great. Go to X years later, the program wont have any output while calculating and then give us the result paused screen and a log to see details of what might happened eg. planet collision