Passage of time in US

[LordAnubis85]

Hey everyone, I had this interesting idea regarding time flow and just time in general in US. So currently the only way to pass time is to let it flow, correct? This can be done slowly for accurate results or very quickly with bodies flying out into oblivion. What if we had the ability to specify a specific time to jump to?

Basically my idea would be to make the current Day/year field editable, then be able to type in something like XX Days or XX years (replacing XX with any numer). Then hopefully, using the data from each body such as velocity and mass, etc, be able to calculate what our current simullation would look like at that time. Is this feasable or would it require too much processing power/time to make these calculations?

My goal with this idea is to be able to type in something like 10,000 years and have the simulator accuratly show me what everything would look like at that time (assuming every simulation starts at a time of 0). If this is feasable, I will add it to the community idea spreadsheet. If this is just completely not possible with current tech, then no worries. Thanks guys

[Dan Dixon]

To get the results at 10000 years of a non stable system you're going to have to do the gravity simulation at a small enough time step for those 10000 years. Setting up the UI to simulation for X time and then stop might make sense.

There would be a way to do a quick but inaccurate simulation of 10000 if the simulation was mostly stable buy just advancing the values of the orbital elements for each body. I hadn't really thought about that before.

Go ahead and add the idea to the list.

[LordAnubis85]

There would be a way to do a quick but inaccurate simulation of 10000 if the simulation was mostly stable buy just advancing the values of the orbital elements for each body. I hadn't really thought about that before.

That's along the lines I was thinking. It would be a quick way of showing where bodies would be after X time. The bigger problem I foresee with this would be if one was to add, say for example, a rouge star traveling toward the main part of the simulation or some other element that would affect the default orbits of any bodies in the simulation.

For example, lets say I load up our solar system and add a rogue star or a black hole traveling toward the Sun. I put that object so far out and have it travel so slow that it takes about 500,000 years for it to be close enough to have any gravitational effect on objects in the system. If I simply advanced the values of that rogue object to say 600,000 years, so as to have it pass through the solar system and be well on its way to the next, will the simulation be able to calculate the gravitational effects that object had and be able to reposition everything?

My concern is that my idea might be too much for the current game engine to handle. Like you said Dan, it may only work for simulations that are stable. At any rate, I've posted this to the Community Ideas spreadsheet for users to vote on.

[Dan Dixon]

If I simply advanced the values of that rogue object to say 600,000 years, so as to have it pass through the solar system and be well on its way to the next, will the simulation be able to calculate the gravitational effects that object had and be able to reposition everything?

No. The only way to get an accurate result is to simulation the system at a small time step over the entire length of time. This is a fundamental issue with n-body/gravity simulators.

[bterrill]

I know this is a bit of an older thread... but... on the whole "stuff flings out to oblivion"

Even if you "jump 500 years" or "10000 years" I understand the math must be done. Is the problem with "flying out of oblivion" or "inaccuracies" the math processing or the fact you're drawing everything out as the math happens?.

If you did not need to show the 1,000 objects flying around in real time could it not accuractly calculation such leaps in time? I mean, you could even limit it to not be editable , but a drop down box. Even if we could only go 50,100,500 years in the future you'd be able to re-choose them enough to go 10,000 years.

If it's the latter, couldn't you just pause the visual part of the sim and throw up something like "Calculating ...." and "Estimated calc time: <X>" and not draw anything until the final calculation is reached?


I only ask because I notice when I "turn trials off" my auto time goes up about 25-30% just from that (not drawing them).

[Dan Dixon]

Trails are currently being drawn quite inefficiently.

As you increase the time step you decrease the accuracy of the simulation. Here's more information:
http://universesandbox.com/faq/#sim