Playing around with the demo, I did enjoy making binary systems, then exploding one of the stars, was fun watching the planets new and totally bizarre orbits
this enjoyment quickly plateaus when you realise youre not really getting much of an indication of why the simulation runs the way it does. Adding a report function to explain, for example, what would ultimately happen to the moons of Saturn if you flung them all out into deep space, would instantly improve the entire experience.
so you want a computer program to be able to read someones mind when they ask "why", correctly understand said question, and then output an easy to understand answer?
this enjoyment quickly plateaus when you realise you’re not really getting much of an indication of why the simulation runs the way it does. Adding a report function to explain, for example, what would ultimately happen to the moons of Saturn if you flung them all out into deep space, would instantly improve the entire experience.
so you want a computer program to be able to read someones mind when they ask "why", correctly understand said question, and then output an easy to understand answer?
No, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
I agree that this might be an excellent teaching aid for schools, and depending on the accuracy of the simulation, also useful for helping college students understand orbital mechanics in a more visual way (biggest complaint I get from my engineering physics friends is that its hard to visualize the stuff)
Originally Posted by CardJoe No, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
Well due to the n body problem, I still think thats asking a lot.
It would take a while to calculate it for each simulation you do, and if it was preprogrammed it would have to be very specific like
"Did you know? Titan will eventually collide with Earth if given a speed of 27Km/s with initial direction vector (0.7238, 0.4791, 0.0102) at time 0"
Not that i think that particular statement is a useful one... because given any speed, vector and time you can make anything collide. "Did you know? The Sun will eventually collide with Earth if given a speed of 0.9c (where c = speed of light) with initial direction vector (direction of eath in 9ish minutes) at time 0"
Originally Posted by CardJoe No, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
Well due to the n body problem, I still think thats asking a lot.
It would take a while to calculate it for each simulation you do, and if it was preprogrammed it would have to be very specific like
"Did you know? Titan will eventually collide with Earth if given a speed of 27Km/s with initial direction vector (0.7238, 0.4791, 0.0102) at time 0"
Not that i think that particular statement is a useful one... because given any speed, vector and time you can make anything collide. "Did you know? The Sun will eventually collide with Earth if given a speed of 0.9c (where c = speed of light) with initial direction vector (direction of eath in 9ish minutes) at time 0"
In which case a more useful report would just be a detailed rundown of what's happened thus far in the simulation with any body you aren't focused on - "Titan was propelled from point X, Y, Z at a speed of A. Io was propelled at..."
Often it's clear that the information is there, as the simulations obviously track this data, but you can't easily access it without wading through complex menus.
A good (though suspect) reference point would be something like JFK Reloaded, which ended each (admittedly short and contained) simulation with a detailed ballistics report - "Shot 1 entered Guard A at X and exited at Y. Shot 2 went shot nothing."
Originally Posted by CardJoe No, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
Well due to the n body problem, I still think thats asking a lot.
It would take a while to calculate it for each simulation you do, and if it was preprogrammed it would have to be very specific like
"Did you know? Titan will eventually collide with Earth if given a speed of 27Km/s with initial direction vector (0.7238, 0.4791, 0.0102) at time 0"
Not that i think that particular statement is a useful one... because given any speed, vector and time you can make anything collide. "Did you know? The Sun will eventually collide with Earth if given a speed of 0.9c (where c = speed of light) with initial direction vector (direction of eath in 9ish minutes) at time 0"
In which case a more useful report would just be a detailed rundown of what's happened thus far in the simulation with any body you aren't focused on - "Titan was propelled from point X, Y, Z at a speed of A. Io was propelled at..."
Often it's clear that the information is there, as the simulations obviously track this data, but you can't easily access it without wading through complex menus.
A good (though suspect) reference point would be something like JFK Reloaded, which ended each (admittedly short and contained) simulation with a detailed ballistics report - "Shot 1 entered Guard A at X and exited at Y. Shot 2 went shot nothing."
My brain just melted with all the numbers and letters....
Looks like a fun game, something different which is nice..
I tried out the demo and was bored within about 40 mins, If you increase the time per frame past about 1-2 days per frame the planets and objects seem to somehow gain veloxity and spin out of control into deep space, it wasn't even slightly realistic at all. I basically got almost every planet except jupiter to fly off into deep space by 2010-05-17...... But maybe the simulation was true and we are all already dead -.-
Originally Posted by ObeyTheCreed I tried out the demo and was bored within about 40 mins, If you increase the time per frame past about 1-2 days per frame the planets and objects seem to somehow gain veloxity and spin out of control into deep space, it wasn't even slightly realistic at all. I basically got almost every planet except jupiter to fly off into deep space by 2010-05-17...... But maybe the simulation was true and we are all already dead -.-
Each time step the planets travel in straight line. After each step the body velocity is changed based on current force acting on it. If you have a 1 year step the earth will shoot of 1 year worth of distance travelled but in a straight line according to it's vector at the step start with no regard to the suns gravity.
It's a good sim but only if your prepared to run it at very small time steps. But with only 20 or so step per second possible it's rather slow like that.
Originally Posted by CardJoe No, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
Whilst a good feature it would be difficult to implement - I guess what you really want is a button that you click and it will tell you if there are any significant events in the objects future (specifying x years or whatever) - which is possible but would require alot of time (given that significant events are defined by close body encounters, and for close body encounters you need the slow algorithm on max efficiency, which for the time span you are talking about would actually take a very long time on a conventional computer).
I bought this to teach with as its excellent though - really really good. The algorithms it employ arent the greatest for celestial body calculations but they do provide a good enough estimate for what this sim does, so what you see a few thousand years in the future, unless you have been running it at max algorithm accuracy, is pretty much going to be wrong - but really, thats not what this is for :)
Quote:
Originally Posted by ObeyTheCreed I tried out the demo and was bored within about 40 mins, If you increase the time per frame past about 1-2 days per frame the planets and objects seem to somehow gain veloxity and spin out of control into deep space, it wasn't even slightly realistic at all. I basically got almost every planet except jupiter to fly off into deep space by 2010-05-17...... But maybe the simulation was true and we are all already dead -.-
The faster you make it go the less accurate it is - sadly, this is not a limitation of the program but rather your computer and the mathematics the calculations are based upon (not how its implemented).
edit: if you want a fast MORE ACCURATE implementation check out my sim at http://webdev.sweeto.co.uk/orbits/orbits.php - right click TWICE to use the RK4 Algorithm (which is the one universe sandbox uses) and speed it up. You will be watching the solar system unfold, and iirc after 20 000 years I had an error of something silly like 20 meters (double click on it to switch between speed mode and graphical mode)
Comments 1 to 17 of 17
ReplyNo, I'd just like it if you ran a simulation, for example, where you exploded the moons of Saturn out of the immediate orbit and, when it ended, it offered something like "Did you know? Titan's trajectory would eventually cause it to collide with Earth at a speed of XXX"
Well due to the n body problem, I still think thats asking a lot.
It would take a while to calculate it for each simulation you do, and if it was preprogrammed it would have to be very specific like
"Did you know? Titan will eventually collide with Earth if given a speed of 27Km/s with initial direction vector (0.7238, 0.4791, 0.0102) at time 0"
Not that i think that particular statement is a useful one... because given any speed, vector and time you can make anything collide. "Did you know? The Sun will eventually collide with Earth if given a speed of 0.9c (where c = speed of light) with initial direction vector (direction of eath in 9ish minutes) at time 0"
In which case a more useful report would just be a detailed rundown of what's happened thus far in the simulation with any body you aren't focused on - "Titan was propelled from point X, Y, Z at a speed of A. Io was propelled at..."
Often it's clear that the information is there, as the simulations obviously track this data, but you can't easily access it without wading through complex menus.
A good (though suspect) reference point would be something like JFK Reloaded, which ended each (admittedly short and contained) simulation with a detailed ballistics report - "Shot 1 entered Guard A at X and exited at Y. Shot 2 went shot nothing."
That is all ( skip to 2:30 for the fun bit ).
My brain just melted with all the numbers and letters....
Looks like a fun game, something different which is nice..
Each time step the planets travel in straight line. After each step the body velocity is changed based on current force acting on it. If you have a 1 year step the earth will shoot of 1 year worth of distance travelled but in a straight line according to it's vector at the step start with no regard to the suns gravity.
It's a good sim but only if your prepared to run it at very small time steps. But with only 20 or so step per second possible it's rather slow like that.
I bought this to teach with as its excellent though - really really good. The algorithms it employ arent the greatest for celestial body calculations but they do provide a good enough estimate for what this sim does, so what you see a few thousand years in the future, unless you have been running it at max algorithm accuracy, is pretty much going to be wrong - but really, thats not what this is for :)
edit: if you want a fast MORE ACCURATE implementation check out my sim at http://webdev.sweeto.co.uk/orbits/orbits.php - right click TWICE to use the RK4 Algorithm (which is the one universe sandbox uses) and speed it up. You will be watching the solar system unfold, and iirc after 20 000 years I had an error of something silly like 20 meters (double click on it to switch between speed mode and graphical mode)
-
« Previous
-
1
-
Next »
Discuss in the forums