Relativistic collisions

[stormdot5]

In previous KSP you could only reach ludicrous speeds modded engines, cheats, or an encounter with the kraken. In KSP2 by default there will be engines that will be able to take you to an appreciable fraction of c. Do you think they will be adding in anything to simulate the effects of time dilation on the travelling craft or changing the way collisions occur at relativistic speeds?

Mostly we expected the game to bug out when we pushed the limits before. But people can set up a collision specifically at relativistic speeds now. In old KSP I would just expect the crafts to just pass through each other between physics ticks. In KSP2 I'd expect something more consistent.

 

With time dilation I reckon it's going to get real complicated for life support mods. 

Edited September 12, 2019 by stormdot5

[magnemoe]

Don't think so or getting faster than 20% of lightspeed will be hard even if going full kerbal. assuming they are using the 1/10 scale stars will be 0.4-2 LY away to other stars making travel time practical. 

Collisions, even coliding at orbital speeds in KSP is a bit hard simply as, game only updates 20 times second. At 2 km/s you are moving 100 meter between each update. 
At 30.000 km/s its 150 km, you would have problems hitting an small moon. 

[GoldForest]

2 hours ago, stormdot5 said:

 In KSP2 by default there will be engines that will be able to take you to an appreciable fraction of c.

 

1 hour ago, magnemoe said:

Don't think so or getting faster than 20% of lightspeed will be hard even if going full kerbal.

Project Daedalus is in game. Irl version would achieve 12% of C after 4 years. 

Kerbal version might hit 12% in 1 year. 

But the way KSP physics work it doesn't allow for faster speed collisions. You can actually pass through a planet, using time warp of course and before the time warp changes, and be fine. I assume the same thing could happen in KSP 2.

[magnemoe]

47 minutes ago, GoldForest said:

 

Project Daedalus is in game. Irl version would achieve 12% of C after 4 years. 

Kerbal version might hit 12% in 1 year. 

But the way KSP physics work it doesn't allow for faster speed collisions. You can actually pass through a planet, using time warp of course and before the time warp changes, and be fine. I assume the same thing could happen in KSP 2.

I once managed to clip trough Eve, I was on an collision trajectory and wanted to adjust to an nice aerobrake a week before entering SOI, however I managed to miss keys and rater went full warp and found myself leaving Eve behind I was already outside it SOI. 

Also in warp parts don't collide, has had the scary experience of having an lander passing trough the transfer stage under warp. More scary then using mechjeb landing autopilot as it will disengage warp then it is ready to burn and aborting the landing also take you out of warp 

[Guest]

Yeah what @magnemoe said. This is going to have near-zero gameplay impact, you'd have to explicitly and carefully set up your relativistic collision and the outcome would be fairly obvious. To make them at all interesting they'd have to add a lot of stuff like terrain deformation and simulation of meteorite strikes, and I don't see that happening. It would be a significant amount of work that wouldn't support much meaningful gameplay. I don't think they want to make KSP an asteroid bombardment simulator.

Edited September 12, 2019 by Guest

[GoldForest]

44 minutes ago, magnemoe said:

Also in warp parts don't collide, has had the scary experience of having an lander passing trough the transfer stage under warp. More scary then using mechjeb landing autopilot as it will disengage warp then it is ready to burn and aborting the landing also take you out of warp 

Oh gosh, I remember accidently doing this sometimes and accidently ending time warp while they were both merged. 

[DStaal]

7 hours ago, GoldForest said:

But the way KSP physics work it doesn't allow for faster speed collisions. You can actually pass through a planet, using time warp of course and before the time warp changes, and be fine. I assume the same thing could happen in KSP 2.

Counterpoint: I don't see a reason to assume they couldn’t change that either.  It would take a bit more processing, but it's not that difficult to put in checks which would allow even fairly small parts to reliably impact each other at even high timewarps.  (A quick idea would be to plot a straight line between the center of mass of objects between time steps, and see if those lines intersect.  You might only do it for objects inside the physics bubble.)

[Snark]

10 hours ago, stormdot5 said:

Do you think they will be adding in anything to simulate the effects of time dilation on the travelling craft

I'd guess almost certainly not.  There's just no point.

1. Time dilation wouldn't make much difference to most aspects of the game, since generally speaking "elapsed time on board" isn't something players need to care about.  If it makes no significant difference to gameplay, there wouldn't be much reason to model it.
2. Even if you did care about time dilation... there likely wouldn't be enough of it to make much of a big difference to gameplay.  The Lorentz contraction requires that you get really close to c to have a big dilation effect.  Even if you're going 90% of lightspeed, the contraction factor is barely more than 2.
3. Even with the interstellar ships in KSP2, it seems unlikely to me that they're going to shave things really close to lightspeed.  They'll have a heckuva lot more dV than we're used to in KSP1, certainly, but c is fast.  About the only way to get close to it would be to have an antimatter-powered photon drive, and even then you'd need a really high wet/dry mass ratio.  Tsiolkovsky still applies, and the rocket equation gets even more tyrannical when you toss relativity into the mix.

 

8 hours ago, GoldForest said:

Project Daedalus is in game. Irl version would achieve 12% of C after 4 years. 

Kerbal version might hit 12% in 1 year.

And at 12% lightspeed, the time dilation factor is...

...wait for it...

...drum roll please...

1.0073

That's right, ladies and gents, less than 1% dilation.  In other words, practically imperceptible to the player-- and that's even if they cared about time spent in transit, which for the most part they don't.

It's also worth noting that Daedalus assumes (per Wikipedia) a 54-kiloton ship for a 500-ton scientific payload, i.e. less than 1% payload mass, which seems much lower than what kerbal players are used to or would prefer.  And that doesn't allow for any deceleration-- the Daedalus design is to just do a flyby and go screaming past the target at relativistic speeds.  If you want to allow enough dV to stop at your destination, that means you'll need to cut your peak velocity in half.  And at 6% lightspeed, the time dilation factor goes down to 1.0018, i.e. just 0.18%.  So on a 10-year voyage, you'd save less than a week of on-board time.  (And even that's being overly generous; it assumes that you're going the full max speed for the entire journey, which of course you aren't-- there's a long period of acceleration and deceleration when you're going a lot slower.)

TL;DR:  relativity is pretty much irrelevant to Daedalus.  It doesn't go anywhere near fast enough.

 

7 hours ago, Brikoleur said:

Yeah what @magnemoe said. This is going to have near-zero gameplay impact, you'd have to explicitly and carefully set up your relativistic collision and the outcome would be fairly obvious. To make them at all interesting they'd have to add a lot of stuff like terrain deformation and simulation of meteorite strikes, and I don't see that happening. It would be a significant amount of work that wouldn't support much meaningful gameplay. I don't think they want to make KSP an asteroid bombardment simulator.

^ This.

10 hours ago, stormdot5 said:

or changing the way collisions occur at relativistic speeds?

I would say almost certainly not, because,

1. expensive to implement, lots more new attack surface for potential bugs
2. basically no effect on gameplay that players would care about

9 hours ago, magnemoe said:

Collisions, even coliding at orbital speeds in KSP is a bit hard simply as, game only updates 20 times second. At 2 km/s you are moving 100 meter between each update. 
At 30.000 km/s its 150 km, you would have problems hitting an small moon.

^ This.

8 hours ago, GoldForest said:

But the way KSP physics work it doesn't allow for faster speed collisions. You can actually pass through a planet, using time warp of course and before the time warp changes, and be fine. I assume the same thing could happen in KSP 2.

^ This.

1 hour ago, DStaal said:

Counterpoint: I don't see a reason to assume they couldn’t change that either.  It would take a bit more processing, but it's not that difficult to put in checks which would allow even fairly small parts to reliably impact each other at even high timewarps.

The problem isn't just timewarp.  At relativistic speeds, even without any timewarp at all, objects move really far even just between physics ticks.

Here's a little challenge for you.

1. Put two craft in LKO, in basically the same orbit but in opposite directions, so that  their relative speed (if they were to collide) is on the order of 4500 m/s.  No cheats allowed, i.e. you're not allowed to use "set orbit" to hack them into identical orbits.  You have to fly them to that, and eyeball it.
2. Now get them to collide.

It's really, really hard, verging on impossible.  Not just because it's hard to get their paths to intersect precisely enough that a collision is even geometrically possible, but also because craft moving at 4500 m/s relative will travel over 200 meters between ticks, so unless they're kilometer-sized ships, they're likely to just pass through each other.

Now take that to the relativistic scenario, where they're moving five thousand times faster, and also you don't have the luxury of putting them in a stable orbit where you can fine-tune the orbits to make them identical.

It gets solidly into "ain't gonna happen" territory.

And even if you could somehow, against all odds, with many many hours of trial and error somehow cajole them into colliding (why the dickens they'd go to the bother is beyond me, but never mind that)... what then?

What happens when two things collide at high speed in KSP?  They get destroyed.  That's what happens.  They get destroyed at 100 m/s, let alone 4500.  If they were going 5000 times faster than 4500 m/s, they're not going to be 5000 times deader; they'll just be destroyed, that's all.  So what was the point?

It just wouldn't be worth it for the devs to plow a whole bunch of time into trying to make something work that, 1. will basically never happen, even if players are trying to make it happen, and 2. even if it does happen, doesn't do anything interesting in gameplay terms.

 

 

[pizzaoverhead]

1 hour ago, DStaal said:

Counterpoint: I don't see a reason to assume they couldn’t change that either.  It would take a bit more processing, but it's not that difficult to put in checks which would allow even fairly small parts to reliably impact each other at even high timewarps.  (A quick idea would be to plot a straight line between the center of mass of objects between time steps, and see if those lines intersect.  You might only do it for objects inside the physics bubble.)

From what I remember, something similar was done to fix issues with vessels hitting the surface of planets hard enough to go through them: There's a collision enhancer that does a raytrace forwards from the vessel and moves it above ground if it were to go below for explosions if it would go through it otherwise. There was some restriction that prevented it from having any effect at higher time warps, probably that the body hadn't loaded its collider yet if you were too far away, but moving fast enough. Potentially it could be enhanced to at least destroy your vessel when warping through bodies.

[DStaal]

4 minutes ago, Snark said:

but also because craft moving at 4500 m/s relative will travel over 200 meters between ticks, so unless they're kilometer-sized ships, they're likely to just pass through each other.

This is what I think could be solvable, if they wanted to.  You don't need to be super-precise - my intersecting-lines based on tick locations would be good enough.  And yes, all that would happen is that the ships would be destroyed - but that's valuable as well: You can actually hit something at speed, and not just 'phase' through it.  It's a minor hole in the reality of the game, but it's a hole that could be addressed.

Whether they want to, or if it's important enough to spend the effort on - I'm not sure.  But if they're beefing up timewarp, it's a related system that would become more noticeable and prominent, so I can see them spending some time on it.  Computing intersections of lines based on two points each is a relatively well-known problem, so it wouldn't take that much.

[magnemoe]

1 hour ago, DStaal said:

Counterpoint: I don't see a reason to assume they couldn’t change that either.  It would take a bit more processing, but it's not that difficult to put in checks which would allow even fairly small parts to reliably impact each other at even high timewarps.  (A quick idea would be to plot a straight line between the center of mass of objects between time steps, and see if those lines intersect.  You might only do it for objects inside the physics bubble.)

This is kind of plausible, at least for planets, moons and targets you intercept. Warp has changed after all.
This is also effective at orbital speeds outside of warp.
People will do combat in KSP multiplayer, we all know that and this would be an nice hook to let missiles hit. 

[Snark]

29 minutes ago, DStaal said:

This is what I think could be solvable, if they wanted to.

Virtually anything is solvable, if one wants to.  Whether it's possible isn't really relevant (it almost certainly is).

What's relevant is, 1. how expensive would it be to implement, and 2. how risky would it be, and 3. what benefit would doing so provide, and 4. how often would that benefit accrue.

And in this case, the answers are, 1. very, and 2. very, and 3. virtually none, and 4. essentially never, respectively.  So putting on my commercial-software-developer's hat for a moment, that's what puts this solidly into the "never gonna happen" category, as far as I'm concerned.

29 minutes ago, DStaal said:

And yes, all that would happen is that the ships would be destroyed - but that's valuable as well: You can actually hit something at speed, and not just 'phase' through it.  It's a minor hole in the reality of the game, but it's a hole that could be addressed.

A hole that could be addressed, at very great development cost, to address a situation that is vanishingly improbable and has essentially no gameplay effect in the 99.99999% use case.

Space is big.  Collisions simply don't happen (especially when everything not in the physics bubble is on rails), unless you've somehow hyper-crowded the space, which generally doesn't happen in KSP.  Suppose you've got literally 1 million spacecraft, and every single one of them is big (say, a 1000 m² cross-sectional area), all crammed into a relatively small area of space (say, 1 million square kilometers in cross-sectional area).  That's a collision cross-section of 0.001.

That means that even in that ultra-super-hyper-crowded scenario, you could send a relativistic ship barrelling through that swarm and it would still have only one chance in 1000 of hitting anything, even with the highly non-trivial physics-tick problem solved.

In short, this is not something that players need to worry about, at all.  Even trying to collide on purpose would be super difficult in that scenario, and even if you did somehow pull off that one-in-a-thousand shot of colliding in that many-thousands-times-more-crowded-than-actual-KSP-gameplay scenario... so what?  "Ships ded."  Okay.  So?

And to achieve that, you'd have to completely rejigger the physics calculations, making them much more complicated and probably more computationally expensive.  Bearing in mind that what Star Theory is trying to do is to make KSP run faster, not "much slower".

Huge cost, negligible benefit-- I just can't see this happening.

29 minutes ago, DStaal said:

Computing intersections of lines based on two points each is a relatively well-known problem, so it wouldn't take that much.

It's a whole lot harder than simply computing intersections of static lines.  Even simply computing the lines is a difficult O(N²) complexity problem when you're dealing with large numbers of objects, and in this case it's not just about finding intersections, it's also about timing; the objects are moving, and may be subject to physics forces along the way.

I'm asserting that this is a much more thorny programming problem than you appear to think it is.  Heck, look at all the problems that plain vanilla KSP has run into with physics over the years, and that's without having to deal with this stuff at all, and that's with quite a few really smart people working really hard on it for quite a bit of time.  Physics simulation is hard, and extremely bug-prone, even when you make it as bare-bones simplified as you can.

[nejc]

Practical effects of relativity would not become important at these speeds, but optical effects on the skybox would, as these effects become prominent faster. Stars behind would readshift and expand from each other, stars in front would concentrate together and blueshift. No effect on gameplay but a neat optical effect nonetheless, if they choose to implement it.

If we could reach higher speeds using antimatter, however, they could add a simple fix without having to simulate relativity entirety, by just making the life support get used up slower, thus simulating one simple consequence of time dilation.

Edited September 13, 2019 by nejc

[kerbiloid]

18 hours ago, Snark said:

Space is big.  Collisions simply don't happen (especially when everything not in the physics bubble is on rails), unless you've somehow hyper-crowded the space, which generally doesn't happen in KSP.  Suppose you've got literally 1 million spacecraft, and every single one of them is big (say, a 1000 m² cross-sectional area), all crammed into a relatively small area of space (say, 1 million square kilometers in cross-sectional area).  That's a collision cross-section of 0.001.

The space is big, but full of bottlenecks. Like sky is big, but planes collide because they are following routes connecting points.

P.S.
A relativistic collision would look like a very short bright flash in night sky.
P.P.S.
... and the cooling hull of your ship glowing red after being heated by Xrays, if you didn't take part in the collision.

Edited September 13, 2019 by kerbiloid

[Snark]

6 hours ago, kerbiloid said:

Like sky is big, but planes collide because they are following routes connecting points.

And the sky-- the part that airplanes fly in, anyway-- is a bottleneck many, many orders of magnitude smaller than space.

And there are a lot more airplanes flying in the real world than the typical KSP player has active around any given planet at any given time.

And, unlike KSP, the real-world sky doesn't run everything on rails outside of a tiny "physics bubble", which means that collision probabilities are an O(N²) thing rather than O(N), which is a huge difference.

And yes, airplane collisions happen, but they're also really rare.

So basically you're taking an event that's super rare to start with, then making it thousands of times less likely due to smaller numbers of objects, then make it again thousands of times less likely due to the larger volume of space involved, then again hundreds or thousands of times less likely due to the "physics bubble" limitation... that's not going to be a problem that's common enough to be worth spending any significant dev time on, let alone likely destabilizing the game and impacting performance significantly.

6 hours ago, kerbiloid said:

P.S.
A relativistic collision would look like a very short bright flash in night sky.
P.P.S.
... and the cooling hull of your ship glowing red after being heated by Xrays, if you didn't take part in the collision.

Except that that could basically never happen in KSP because if you didn't take part in the collision, there would be no collision because things outside your physics bubble are on rails and it's physically impossible for them to collide.

Either you collide with something-- in which case you're just destroyed and dead-- or else there's no collision.

[kerbiloid]

5 minutes ago, Snark said:

Except that that could basically never happen in KSP because if you didn't take part in the collision, there would be no collision because things outside your physics bubble are on rails and it's physically impossible for them to collide.

That's clear. I was just trying to realise, what do one expects to see on a relativistic collision.

P.S.
Spaceports and stations are the bottlenecks.

Edited September 13, 2019 by kerbiloid

[Snark]

6 hours ago, nejc said:

Practical effects of relativity would not become important at these speeds, but optical effects on the skybox would, as these effects become prominent faster. Stars behind would readshift and expand from each other, stars in frond would concentrate together and blueshift. No effect on gameplay but a neat optical effect nonetheless, if they choose to implement it.

If they wanted to simulate Lorentz contraction by applying some graphical filter to the skybox, now that would be something that should be fairly low-risk and hopefully not too expensive to implement.

It wouldn't be free, though, and the question remains whether it would be a big enough effect to be worth the bother.  A ship going 5% lightspeed would experience only a very small redshift / blueshift / geometric distortion.  The player might not even notice it, even if it's implemented.  So they'd have to determine whether the benefit would be worth the cost to implement.  Maybe it would be worth it-- I honestly don't know.  But it's certainly not a slam-dunk that it would.

2 minutes ago, kerbiloid said:

Spaceports and stations are the bottlenecks.

Oh, absolutely.  And certainly a player can accidentally crash into their space stations and such-- that can and does happen, it's an important part of the game.

But that doesn't happen at relativistic speeds.  A player who's trying to dock at a significant fraction of lightspeed has... um... worse problems to worry about. 

And, more to the point, that works with traditional tick-based physics models, already works just fine in KSP 1, and I can't imagine it would be any worse in KSP 2.  It ought to work just fine.

[nejc]

3 minutes ago, Snark said:

...  A ship going 5% lightspeed would experience only a very small redshift / blueshift / geometric distortion.  The player might not even notice it, even if it's implemented. ...

A Slower Speed of Light gives an idea as to how noticeable it would be at small fractions of c. For me personally it is enough to desire it.

[kerbiloid]

25 minutes ago, Snark said:

But that doesn't happen at relativistic speeds.  A player who's trying to dock at a significant fraction of lightspeed has...

... missed with relativistic "jump".
He wants to decelerate from his 0.5 c down to 0 as close to the station as possible, to not spend several months taxiing from his zero-speed point to the station.
Preparing the relativistic (brachistochronic or else) flight, he targets the station beacon ("Target: Orbital station Alpha").
The modern, highly accurate autopilot calculates his trajectory...
... and the pilot misses the alarm clock and suddenly brakes down to 0.1 c instead of 0 being precisely aimed at the destination station and the cloud of crafts performing docking operations.

Upd.
Especially if his SAS automatically turns him to the "Target".

Edited September 13, 2019 by kerbiloid

[kerbiloid]

Another case of the relativistic collision: relativistic projectiles, like in "The rose and the worm" novel.

An alien invasion fleet of several ships is approaching to the Solar System at 0.5 c, aiming at the Earth.
Right before braking they release a cloud of 100 kg heavy graphite rods.
The ships start decelerating, while the cloud of rods keeps moving to the Earth at 0.5 c.

Long before the fleet arrival, the rod cloud reaches the Earth, and every rod hit releases ~ (100 * (0.5 * 3*10⁸)² / 2 / 4.2*10¹⁵) ~ 260 Mt of energy, so they devastate the Earth.

So, we can aim a station with SAS, decouple impactors and decelerate. Some of impactors will hit the station, takng out the opponent's base before our arrival.

***

Though, I don't see why the relativistic collision should differ in implementation from any explosion of enough high yield.
Say, a nuke totally releases energy in ~10^-7..10^-6 s. A photon can pass just a hundred meters, a relativistic impactor as well.
So, the question is just in massive explosions at all, relativism plays no role here.

Edited September 14, 2019 by kerbiloid