raptop changed the topic of #principia to: READ THE FAQ: http://goo.gl/gMZF9H; The current version is Galileo. We currently target 1.5.1, 1.6.1, 1.7.x, 1.8.1, and 1.9.1. <scott_manley> anyone that doubts the wisdom of retrograde bop needs to get the hell out | https://xkcd.com/323/ | <egg> calculating the influence of lamont on Pluto is a bit silly… | <egg> also 4e16 m * 2^-52 is uncomfortably large
egg|anbo|egg has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
egg|anbo|egg has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
egg|anbo|egg has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
<discord-> l​amont. — the floating origin is only around one vessel, so the further away the other vessel is the more you get roundoff error with the physics engine trying to do its thing
<discord-> l​amont. — current versions of unity support multiple physics "scenes" so you could have multiple floating origins attached to each vessel and do physics in the frame of reference of each vessel, but with KSP written the way it is, you're going to be 1,000s of meters or kilometers away and roundoff error in the physics engine dealing with large numbers like that is going to be stupid. i expect PhysX is all s
<discord-> D​RVeyl. — I'd like to see where that precision is really driving the accuracy, tho. 1000 is 10 bits. 1M is 20. Still leaves a lot of fractional bits.
<discord-> F​alcon. — Question, is there an optimum travel time to Jupiter to save time on the gravity assist? Because i know the slower the transfer is to Jupiter, the more energy you will ultimately get from Jupiter if i recall correctly. But that is a super slow transfer compared to a year and a half transfer which is meant to cut down on time
<discord-> F​alcon. — so is there an optimum time for transfers to jupiter to save on post flyby time, but to still get a good amount of energy from Jupiter itself?
<discord-> D​RVeyl. — Tho I guess I need 10 more bits to be at mm accuracy, and so now I'm running out.
<discord-> D​RVeyl. — (But that says I'm running out at doing physics at distances of 1Mm from the origin. Unless the issue is in intermediate calculations, ie losing precision when I multiply?)
<discord-> D​RVeyl. — Tho I guess I need 10 more bits to be at mm accuracy, and so now I'm running out. Assuming this is the required positional precision. (edited)
<discord-> l​amont. — yeah machine epsilon of single precision floats is 1e-7
<kmath> YouTube - Unite 2013 - Building a new universe in Kerbal Space Program
<discord-> l​amont. — see also: "PhysX is crap"
<discord-> l​amont. — the unity vector class is also not very careful about overflow and stuff like `magnitude` and `normalize` can overflow near that machine eps. i suspect KSPs Vector3d will similarly overflow in those APIs if you push it.
Mike` has quit [Ping timeout: 194 seconds]
discord- has quit [Read error: No route to host]
oeuf has quit [Read error: Connection reset by peer]
uovo has quit [Read error: Connection reset by peer]
discord- has joined #principia
oeuf has joined #principia
bees has quit [Ping timeout: 204 seconds]
uovo has joined #principia
bees has joined #principia
Mike` has joined #principia
<discord-> m​oar ssto. — why would vector3d be different from other "types" of cartesian vectors?
UmbralRaptor has quit [Ping timeout: 204 seconds]
UmbralRaptop has joined #principia
UmbralRaptop has quit [Ping timeout: 204 seconds]
UmbralRaptop has joined #principia
<discord-> K​atniss. — Vector3D uses Double instead of single precision floats afaik
Blu3wolf has joined #principia
<discord-> S​tandecco. — So, let's move ksp to matlab
<discord-> S​tandecco. — Or to principia
<discord-> S​tandecco. — How hard would it be to implement keplerian orbits into Principia?
<discord-> m​oar ssto. — I hope actually we can get principia's integrator for ksp's off rail calculations.
<discord-> s​ichelgaita. — Re: accuracy. You indeed want mm accuracy, lest the vessels start to move in noticeable ways due to numerical errors. Now it's convenient to do integration in the barycentric frame, because just say no to apparent forces. To represent 1 mm at the aphelie of Pluto you need about 52 bits of mantissa. So double works, but barely.
<discord-> s​ichelgaita. — Re: accuracy. You indeed want mm accuracy, lest the vessels start to move in noticeable ways due to numerical errors. Now it's convenient to do integration in the barycentric frame, because just say no to apparent forces. To represent 1 mm at the aphelie of Pluto you need about 52 bits of mantissa. So `double` works, but barely. (edited)
<discord-> m​oar ssto. — ksp's intergator is wobbly and krakeny.
egg|anbo|egg has joined #principia
egg|cell|egg has quit [Ping timeout: 378 seconds]
egg|anbo|egg has quit [Remote host closed the connection]
egg|anbo|egg has joined #principia
<discord-> B​utcher. — I think 52-bit for representing your position is too much for double, as soon as you try to perform any operation on that it's going to explode.
<discord-> m​oar ssto. — I think the biggest problem with ksp's integrator is not floating point issue, it feels like the whole integrator system is broken.
<discord-> m​oar ssto. — The wobblyness might be floating point inaccuracy.
<discord-> m​oar ssto. — But the phantom forces that changes the energy of orbits with a particular pattern doesn't appear to be.
<UmbralRaptop> Stock uses basic Euler's method, right?
<discord-> m​oar ssto. — You can get kraken acceleration of something like couple dm/s
<discord-> m​oar ssto. — It definitely ruins your day.
<discord-> m​oar ssto. — > U​mbralRaptop. — Stock uses basic Euler's method, right?
<discord-> m​oar ssto. — @鹊桥IRC Someone say it's physX's symplectic euler.
<discord-> m​oar ssto. — But not directly calling it for position an velocity in cartesian.
* discord- e​gg. — stares at the backlog_
<discord-> e​gg. — a lot of that conversation is confused beyond the point of usefulness; this is admittedly not helped by the fact that the thing that KSP calls integrator or integration is not the actual integration of the ODEs involved, but simply the computation of the forces (the RHS of the ODE, in a sense)
<discord-> m​oar ssto. — So the actual integrator is not in ksp itsself
<discord-> m​oar ssto. — So the actual integrator is not in ksp itself? (edited)
<discord-> m​oar ssto. — > a lot of that conversation is confused beyond the point of usefulness; this is admittedly not helped by the fact that the thing that KSP calls integrator or integration is not the actual integration of the ODEs involved, but simply the computation of the forces (the RHS of the ODE, in a sense)
<discord-> m​oar ssto. — @egg For the actual ode, ksp is setting it up in body centered inertial or ship aligned frame?
<discord-> e​gg. — @moar ssto the actual physics engine operates in Unity’s World, to which KSP gives various meanings depending on the current situation, either rotating with the surface or nonrotating. The origin is always somewhat near the ship, as @lamont pointed out, so it is certainly not centred on any body.
<discord-> B​utcher. — @egg I noticed that, the integrate function is just `sumForces` essentially.
<discord-> e​gg. — yes
<discord-> e​gg. — terrible naming, a mainstay of KSP API design
<discord-> B​utcher. — Whatever it uses it doesn't seem like a sympletic integrator given the amount of kraken accellerations instroduced.
<discord-> B​utcher. — Whatever it uses it doesn't seem like a sympletic integrator given the amount of kraken accellerations introduced. (edited)
<discord-> B​utcher. — In particular whatever does surface collision response is terrible for energy conservation.
<discord-> e​gg. — well that is more the RHS not actually deriving from a Hamiltonian than anything to do with the integration
<discord-> e​gg. — collisions are devilish to start with admittedly
<discord-> B​utcher. — True. You tend to end up with the trade off between having intersecting objects and having sane response.
<discord-> B​utcher. — Since I suspect PhysX does a stepped move and collide.
egg|anbo|egg_ has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
<discord-> m​oar ssto. — > @moar ssto the actual physics engine operates in Unity’s World, to which KSP gives various meanings depending on the current situation, either rotating with the surface or nonrotating. The origin is always somewhat near the ship, as @lamont pointed out, so it is certainly not centred on any body.
<discord-> m​oar ssto. — @egg So do we have inertial frame within atmosheric bodies SOI above the atmosphere? Also, I guess centering the frame to body or to vessel shouldn't make any diffrenece in the position and velolicty calculated, since you are just flipping the sign of position, velocity and accelleration you are interested in, or I am totally wrong?
egg|anbo|egg has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
egg|anbo|egg has joined #principia
Blu3wolf has quit [Ping timeout: 189 seconds]
egg|anbo|egg_ has quit [Read error: Connection reset by peer]
Blu3wolf has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]
<discord-> B​utcher. — Ugh, this `other != nullptr` check is killing me.
<discord-> B​utcher. — Two crashes, turned on journalling, works perfectly.
<discord-> R​ingo. — How I can see the orbital parameters like inclination etc of a maneuver?
<discord-> B​utcher. — Mmm boat shaped orbit
egg|anbo|egg has joined #principia
<discord-> e​gg. — @moar ssto, the frame is not centred on the vessel. The frame is inertial, its origin (in position, and often in speed, depending on various bits of messy logic) is near the vessel. The frame is then moved around by KSP from time to time (sometimes at every frame), effectively changing the reference system in which the problem is expressed. This is known, for positions, as Floating Origin, and for vel
<discord-> e​gg. — @moar ssto, the frame is not centred on the vessel. The frame is inertial (or rotating with the planet at low enough altitudes), its origin (in position, and often in speed, depending on various bits of messy logic) is near the vessel. The frame is then moved around by KSP from time to time (sometimes at every frame), effectively changing the reference system in which the problem is expressed. This is
Blu3wolf has quit [Quit: Konversation terminated!]
Blu3wolf has joined #principia
Blu3wolf has quit [Quit: Konversation terminated!]
Jesin has quit [Quit: Leaving]
<discord-> l​amont. — wouldn't it be better to integrate gravity and external forces completely separate from PhysX and then do a coordinate transform into the rotating reference frame without gravity and have PhysX be responsible for moving the pile up around?
Jesin has joined #principia
<discord-> C​alvin_Maclure. — Huh... my save is apparently no longer letting me save?? The only thing I changed this week was updating Principia to Gallois. Anyone else??
<discord-> C​alvin_Maclure. — I also could not get engine rated burn time data when middle clicking in the VAB, for some reason.
<discord-> C​alvin_Maclure. — Not sure if both related, or anything to do with latest Pr release
<discord-> l​pg. — those both sound like krash shenanigans
<discord-> C​alvin_Maclure. — Yeah
<discord-> C​alvin_Maclure. — My take as well
<discord-> l​pg. — disabling saves and disabling testflight are both things it does
<discord-> C​alvin_Maclure. — but the only update was Pr...
Jesin has quit [Quit: Leaving]
<discord-> l​pg. — no update needed, it's just a thing it _does_ sometimes
<discord-> C​alvin_Maclure. — right
<discord-> C​alvin_Maclure. — ok
<discord-> l​pg. — alt-f12 menu to re-enable saves, testflight gets enabled via its own UI
<discord-> C​alvin_Maclure. — got it. Thanks, cheers
Jesin has joined #principia
<discord-> l​amont. — wouldn't it be better to integrate gravity and external forces completely separate from PhysX and then do a coordinate transform into the rotating reference frame without gravity and have PhysX be responsible for moving the pieces in the pile up around? (edited)
<discord-> l​amont. — wouldn't it be better to integrate gravity and external forces completely separate from PhysX and then do a coordinate transform into the rotating reference frame without gravity and have PhysX be responsible for moving the pieces in the pile up around? (this idea may fall apart once you have to deal with collisions into the ground, although i wonder if the external integrator could just smash the
<discord-> e​gg. — @lamont you are describing how Principia works, yes; we consider the pile ups as point masses, and PhysX deals with the pile ups themselves as best it can. Some fudgery is required with rotation which PhysX does not even attempt to do properly.
<discord-> e​gg. — @blowfish I did the thing https://github.com/KSP-RO/SolverEngines/pull/44
<discord-> b​lowfish. — oh cool, thanks
<discord-> b​lowfish. — I can try to release tonight
<discord-> b​lowfish. — it's been a while
<discord-> e​gg. — @blowfish @ferram4 so that we can finally recreate https://en.wikipedia.org/wiki/2003_Baghdad_DHL_attempted_shootdown_incident in KSP
<discord-> s​ichelgaita. — ☝️ This is insane.
<discord-> K​atniss. — This is pretty insane too https://en.wikipedia.org/wiki/Korean_Air_Lines_Flight_007
<discord-> b​lowfish. — I know you can control without yaw. but _only_ yaw?
<discord-> b​lowfish. — oh, I guess there's some pitch moment based on the overall throttle
<discord-> K​atniss. — you can control pitch via throttles too. There was an autoland feature texted at some point, that could land using only engines.
<discord-> K​atniss. — you can control pitch via throttles too. There was an autoland feature tested at some point, that could land using only engines. (edited)
<discord-> K​atniss. — you can control pitch via throttles too. There was an autoland feature tested at some point, that could land using only engines. Didn't go anywhere tho, was deemed unnecessary. (edited)
<discord-> K​atniss. — you can control pitch via throttles too. There was an autoland feature tested at some point, that could land using only engines. Didn't go anywhere tho, was deemed unnecessary.
<discord-> K​atniss. —
<discord-> K​atniss. — there was also United 232 accident (edited)
<discord-> S​tandecco. — still incredible that the landing even happened
<discord-> S​tandecco. — they had a damaged wing, no hydraulics and a leaking wing fuel
<discord-> S​tandecco. — cannot get any harder than this without being completely impossible
<discord-> S​tandecco. — they had a damaged wing, no hydraulics and were leaking fuel (edited)
<discord-> b​ofh453. — "In May 2006, Captain Éric Genotte, together with Armand Jacob, an Airbus experimental test pilot, gave a presentation to the Toulouse branch of the Royal Aeronautical Society titled "Landing an A300 Successfully Without Flight Controls"." i want to find the slides of this presentation.
<discord-> l​amont. — i want to see how much bleach had to get used on their underwear after that
<discord-> C​alvin_Maclure. — What's the simplest way to figure your optimal-ish launch for a good TLI with Principia?
<discord-> C​alvin_Maclure. — There's a KOS script that was given to me, but other than that? You can't use MJ's ''relative inclination'' which is what I did in my pre-Principia days
<discord-> l​pg. — without kos, the simplest I've found is to go polar(ish), orienting your orbit to intersect the moon's plane ~half a week (~45 degrees) in front of the moon
<discord-> C​alvin_Maclure. — Polar?? Hmm... though of that, but that's cutting into my DV a bit too much.
<discord-> l​pg. — really what you want is to get close to an orbital plane perpendicular to the moon's plane. depending on time of year, that can be much less than polar
<discord-> l​pg. — I'm still working on my technique for timing a due-east launch, and wouldn't be able to explain it well. And I can still only reliably get it right twice a month anyway
<discord-> l​pg. — I really wish there existed a way to draw a great-circle ground track onto map view, based on a given surface position and heading. My visualization skills are just not up to imagining it
<discord-> C​alvin_Maclure. — Yeah. I mean, I get the concept, but looking to figure it out in a more methodical manner
<discord-> C​alvin_Maclure. — > I really wish there existed a way to draw a great-circle ground track onto map view, based on a given surface position and heading. My visualization skills are just not up to imagining it
<discord-> C​alvin_Maclure. — @lpg I mean... that sounds like a good mod idea... or update 😉
<discord-> l​amont. — you need the node of the transfer orbit with the moon's plane to line up with the moon after your transfer period, so advance the moon's orbit by ... 3 days? ... then you want the 28.608 degree inclination plane which contains that vector.
<discord-> C​alvin_Maclure. — > you need the node of the transfer orbit with the moon's plane to line up with the moon after your transfer period, so advance the moon's orbit by ... 3 days? ... then you want the 28.608 degree inclination plane which contains that vector.
<discord-> C​alvin_Maclure. — @lamont Well... yeah... but eyeballing that is... well crap
<discord-> l​amont. — from the desired plane you can find the LAN
<discord-> l​amont. — right but don't, figure it out with math
<discord-> C​alvin_Maclure. — > right but don't, figure it out with math
<discord-> C​alvin_Maclure. — @lamont Huh?
<discord-> l​pg. — math is what we pay you and egg for
<discord-> C​alvin_Maclure. — Damn right imma do it the math way! I'd have to hang my mech eng bachelors degree in shame if I didn't! lol
<discord-> l​amont. — you can take the osculating elements and advance it by the transfer time easily enough, particularly from within KSP you can use uh GetStateVectorsFromUT() or whatever it is and that'll be certainly better than eyeball.
<discord-> C​alvin_Maclure. — Also, please note this:
<discord-> l​amont. — the pos then becomes your node line, your plane must include that and be at 28.608 degrees
<discord-> l​amont. — so `dot([0, -1, 0], hf) = cosd(28.608)`
<discord-> C​alvin_Maclure. — All I know is I have to nail this next launch. I've crunched the numbers and my rocket is on the pad with having JUST unlocked and equipped lunar comms with literally just over 5 days till Lunar flyby & impact contracts both due to end!
<discord-> C​alvin_Maclure. — I... I don't even know how I managed to cut so ridiculously close!
<discord-> l​amont. — then `dot(hf, pos) = 0`
<discord-> l​amont. — so `dot([0, -1, 0], hf.normalized) = cosd(28.608)` (edited)
<discord-> l​amont. — i think there's gotta be an easier way to do this that i've seen to select the right plane/angular momentum vector give a line and an inclination
<discord-> C​alvin_Maclure. — Well, KOS + script seems to be the easiest way
egg|anbo|egg_ has joined #principia
egg|anbo|egg has quit [Ping timeout: 204 seconds]
<discord-> C​alvin_Maclure. — Also, how... does one get MJ to execute a Principia maneuver?
<discord-> l​pg. — one gets a recent dev build and one clicks the new button
<discord-> C​alvin_Maclure. — So Galois no good?
<discord-> l​pg. — of MJ
<discord-> C​alvin_Maclure. — Ah...
<discord-> l​pg. — then one discovers it doesn't take ignition time into account, one curses, and one goes back to the manual way
<discord-> C​alvin_Maclure. — lol!
<discord-> C​alvin_Maclure. — Yeah... that's what I figured.
<discord-> C​alvin_Maclure. — It's just that... when that kick stage starts hitting +10G acceleration values...
<discord-> l​pg. — (or the kOS way, of course)
<discord-> l​pg. — Better Time Warp is my new best friend
<discord-> C​alvin_Maclure. — 1/4 time speeds?
<discord-> l​pg. — 0.1x
<discord-> C​alvin_Maclure. — Ooh! That's right!
<discord-> C​alvin_Maclure. — Yeah, I mean, with THAT, sure I'll do manual
<discord-> C​alvin_Maclure. — What's the way to activate that again???
<discord-> C​alvin_Maclure. — Also, wouldn't displaying patched conics to visualize the Moon's orbital plane be good enough?
<discord-> C​alvin_Maclure. — Seems like it would make an imprecise method a little more precise
<discord-> C​alvin_Maclure. — Although the moon's orbit seems completely uninclined??
egg|anbo|egg has joined #principia
egg|anbo|egg_ has quit [Ping timeout: 194 seconds]
<discord-> B​utcher. — @Calvin_Maclure what's the issue with just using kos?
<discord-> C​alvin_Maclure. — There isn't one, really. I'm just not one to be satisfied with not understanding the underlying math or physics. That's all.
<discord-> C​alvin_Maclure. — I don't like to just relly on something withou understanding how it works and why.
<discord-> S​tonesmile. — Writing the scripts your self would take care of the 'understanding' part
<discord-> B​utcher. — Which script are you using?
<discord-> C​alvin_Maclure. — Yeah. But then again, I've also purchased SMAD and like tonread through applicable chapters, or compliment with online reading. I just don't like to code...
<discord-> C​alvin_Maclure. — > Which script are you using?
<discord-> C​alvin_Maclure. — @Butcher someone gave me one of yours, I think."LunarLaunchOrbits.ks"
<discord-> B​utcher. — Yes that's mine.
<discord-> B​utcher. — The maths is pretty simple.
<discord-> C​alvin_Maclure. — Where... do I place the script?? New to KOs
<discord-> B​utcher. — Ships/script I think.
<discord-> C​alvin_Maclure. — Guess I'll have to restart the game then
<discord-> S​tonesmile. — Not needed I think
<discord-> C​alvin_Maclure. — Wait... where do I place the actual script?? Do I put the actual file in the kOS folder??
<discord-> C​alvin_Maclure. — Ships/scripts is the place??
<discord-> C​alvin_Maclure. — So next to "boot"?
<discord-> S​tonesmile. — No, ksp_folder/ships/scripts
<discord-> C​alvin_Maclure. — Yes, that's what I mean. Then in there, there's a "boot'' folder.
<discord-> C​alvin_Maclure. — So nested next to that
<discord-> C​alvin_Maclure. — ?
<discord-> S​tonesmile. — Correct, wrote before I saw that message
<discord-> C​alvin_Maclure. — Got it. Thanks.
<discord-> C​alvin_Maclure. — No restart needed?
<discord-> S​tonesmile. — To acces that script from in-game you type `switch to 0.` and `run <script>.`
<discord-> B​utcher. — No restart, it picks up changes live.
<discord-> B​utcher. — I often edit scripts while in flight.
<discord-> B​utcher. — The maths is: predict where moon will be in 4 days (using osculating elements), form lands containing that point, the earth and the launch pad. Launch into plane.
<discord-> B​utcher. — The maths is: predict where moon will be in 4 days (using osculating elements), form plane containing that point, the earth and the launch pad. Launch into plane. (edited)
<discord-> C​alvin_Maclure. — What it needs to be.
<discord-> B​utcher. — You'll then be in an orbital plane that intercepts the moon in 4 days.
<discord-> B​utcher. — The prediction is obviously not perfect but it's close enough.
<discord-> C​alvin_Maclure. — Now if my first time flying LR-105 and 89s just don't quit on me, we'll be gtg!
<discord-> B​utcher. — It had revolutionised my lunar flights, no more waiting for windows or getting into an orbit that needs a huge plane change.
<discord-> C​alvin_Maclure. — So... can you then just pretty much launch at any time?? Or did I misunderstand that last statement?
<discord-> S​tonesmile. — Correct, just jaunch into the inclination that the script tells you
<discord-> S​tonesmile. — Correct, just launch into the inclination that the script tells you (edited)
<discord-> C​alvin_Maclure. — Got it.
<discord-> C​alvin_Maclure. — Sweet
<discord-> B​utcher. — Yes, ideally you launch right after running the script.
<discord-> C​alvin_Maclure. — So: 1- get to pad, 2- run script, 3- launch right after, 4- launch into designated kOS plane
<discord-> B​utcher. — Yes.
<discord-> C​alvin_Maclure. — Fabulous kOS script is fabulous
<discord-> B​utcher. — Given how simple it is I'm surprised no one has used this before.
<discord-> C​alvin_Maclure. — Probably just a question of not knowing this thing was around!
<discord-> C​alvin_Maclure. — That and maybe some kOS intimidation
<discord-> B​utcher. — I mean the technique.
<discord-> C​alvin_Maclure. — Ah!
<discord-> B​utcher. — I only wrote the script a few weeks ago.
<discord-> B​utcher. — But the maths is very straightforward so I would have thought someone would have tried it.
<discord-> C​alvin_Maclure. — How did you figure out the process, out of curiosity?
<discord-> B​utcher. — I'm not sure, I think I just thought what's the best way to align to get an easy flight to the moon and then thought about using a plane from three points,since that's a standard method of obtaining a plane.
<discord-> C​alvin_Maclure. — Fair enough
<discord-> B​utcher. — And it worked.
<discord-> B​utcher. — I'm playing at vafb so I don't get anywhere near the moon's plane itself.
<discord-> C​alvin_Maclure. — Why V? KSC too mainstream?
<discord-> B​utcher. — Fancied a change. Was planning to do a lot of polar direct ascent.
<discord-> B​utcher. — Then I wrote this script and now I don't do much direct ascent.
<discord-> C​alvin_Maclure. — 🤣
<discord-> B​utcher. — It's a neat trick, but the launch windows can be 20 days apart.
Jesin has quit [Quit: Leaving]