raptop changed the topic of #principia to: READ THE FAQ: http://goo.gl/gMZF9H; The current version is Fréchet. We currently target 1.5.1, 1.6.1, and 1.7.x. <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
Wetmelon has quit [Ping timeout: 378 seconds]
Mike` has quit [Ping timeout: 204 seconds]
Mike` has joined #principia
Raidernick_ has joined #principia
Raidernick has quit [Ping timeout: 204 seconds]
Wetmelon has joined #principia
Wetmelon has quit [Ping timeout: 190 seconds]
UmbralRaptop has quit [Ping timeout: 204 seconds]
egg|laptop|egg has joined #principia
UmbralRaptop has joined #principia
egg|laptop|egg has quit [Remote host closed the connection]
egg|laptop|egg has joined #principia
egg|laptop|egg has quit [Remote host closed the connection]
egg|laptop|egg has joined #principia
egg|laptop|egg has quit [Remote host closed the connection]
egg|laptop|egg has joined #principia
egg|laptop|egg has quit [Remote host closed the connection]
<discord-_> d​kavolis. — @egg Have you tried building line meshes for trajectories instead of drawing them as GL lines every frame? I haven't played with principia in quite a while but it used to lag a lot with very long flight plans/histories in map view.
egg|laptop|egg has joined #principia
<discord-_> e​gg. — define quite a while
<discord-_> d​kavolis. — 1.4 probably
<discord-_> e​gg. — we improved that by an order of magnitude in Чебышёв (August 2017)
<discord-_> e​gg. — but mostly you may be confused about what the costly thing was
<discord-_> e​gg. — after Чебышёв a major source of slowness was the computation of predictions, rather than their plotting
<discord-_> e​gg. — This was addressed in fano
<discord-_> e​gg. — This was addressed in Fano (edited)
<discord-_> e​gg. — (by computing predictions asynchronously)
<discord-_> d​kavolis. — How often do you recompute the predictions if nothing has changed?
<discord-_> d​kavolis. — Also, if I increase the plan length and max steps per segment without a maneuver, the fps tanks to low single digits.
<discord-_> S​ir Mortimer. — I now have 8, 9, 12, 14, 32 and 64 faces. this is getting out of control.
<discord-_> e​gg. — @dkavolis max steps is literally "how much work should we do"
<discord-_> e​gg. — if you tell us to do more computation, we will do more computation
<discord-_> d​kavolis. — But do you have to keep repeating all of it if conditions haven't changed?
<discord-_> e​gg. — for the predictions, it is asynchronous, so displaying it does not block things (we keep recomputing, and when we have a new one start recomputing again, but you never wait for it)
<discord-_> e​gg. — for the flight plans, we do not recompute unless you tweak the flight plan settings
<discord-_> e​gg. — but more flight plan does mean more things to render, and that is not just drawing on the screen: it is reference frame changes to compute the trajectory in the proper frame
<discord-_> d​kavolis. — An empty flight plan with a large number of steps basically halts KSP until you delete it
<discord-_> e​gg. — yes, that is what max steps is for
<discord-_> e​gg. — you should not increase it unless you need it
<discord-_> e​gg. — and for an empty flight plan you probably don’t actually care about seeing more of it
<discord-_> e​gg. — the specifics will depend on your current trajectory
<discord-_> e​gg. — but if you ask us to render months of LEO, that will be slow
<discord-_> e​gg. — that is also not the point of the flight planner
<discord-_> e​gg. — (the orbit analyser might provide more usable information for that)
<discord-_> e​gg. — note that months of LEO planning is much more expensive (both to compute, to process, and to render) than months of interplanetary plan, because the flight plan works with an adaptive stepsize
<discord-_> d​kavolis. — A 200d interplanetary plan in stock is still a big fps hit, ~78 to ~30
<discord-_> e​gg. — yes
<discord-_> e​gg. — though that will depend on your tolerance
<discord-_> H​aukifile. — Any ideas on what might cause principia crashes when staging rockets or jettisoning fairings? I'm on 1.8.1 and frobenius
<discord-_> H​aukifile. — https://pastebin.com/qV2VMm19
<discord-_> H​aukifile. — all the crash logs show similar logs
<discord-_> H​aukifile. — all the crash logs look the same (edited)
<discord-_> e​gg. — fixed in the next version
<discord-_> e​gg. — are you doing something new? faster vessels? new solar system? something?
<discord-_> e​gg. — this is an ancient bug and people keep coming across it now
<discord-_> S​ir Mortimer. — this transit mission turns out to be quite fun 😛
<discord-_> H​aukifile. — interesting. I'm not doing anything special, just staging rockets in LEO.
<_whitenotifier-d13c> [Principia] eggrobin opened issue #2528: Crash when pushing a trivial manœuvre beyond the end of a trivial flight plan - https://git.io/Jfe5d
<discord-_> H​aukifile. — I've had the issue semiconsistently since making orbit
<discord-_> H​aukifile. — Can't remember if I encountered it before, but ofc there's less staging before getting to orbit
<discord-_> H​aukifile. — I don't think it happens in atmosphere, for example my first stage seems to always separate just fine
<discord-_> e​gg. — AFK dinner
egg|laptop|egg has quit [Remote host closed the connection]
egg|laptop|egg has joined #principia
egg|laptop|egg has quit [Ping timeout: 190 seconds]
<discord-_> n​eph. — egg is making an omlette
<discord-_> n​eph. — egg is making an omelette (edited)
<discord-_> S​tandecco. — that'd be cannibalism
<discord-_> d​kavolis. — It takes ~7ms for principia to draw 1349 lines for ~200d flight plan in stock, a mesh would likely be faster
egg|laptop|egg has joined #principia
<discord-_> e​gg. — are you timing just the C# time?
<discord-_> d​kavolis. — Just the whole render function
<discord-_> e​gg. — I would expect most of the work to be in the C++, where we do the change of reference frame and the downsampling
<discord-_> e​gg. — thing is we are not drawing a fixed set of lines
<discord-_> e​gg. — we have a continuous trajectory, and we choose which lines we show based on where the camera is
<discord-_> d​kavolis. — Are you not caching trajectories?
<discord-_> e​gg. — so as to draw as few lines as possible, but still match closely the continuous trajectory
<discord-_> e​gg. — We are not caching the rendered thing, no, because it keeps changing
<discord-_> e​gg. — The trajectory is stored
<discord-_> e​gg. — but that is not simply a sequence of positions
<discord-_> e​gg. — (or rather, it is in representation—although we intend to change that soon due to other performance issues—but it carries special meaning and encodes a continuous function of time)
<discord-_> d​kavolis. — Could you not build the sampled trajectory in world space and let GPU handle rendering?
<discord-_> e​gg. — re. we intend to change that, see https://github.com/mockingbirdnest/Principia/issues/2400, we will likely use a representation based on frequency analysis
<discord-_> e​gg. — @dkavolis that would result in a very inefficient thing in world space, because your camera might be close to the trajectory
<discord-_> e​gg. — there is *a lot* of data in that continuous function
<discord-_> e​gg. — sampling it to get something that looks good from all angles is very messy
<discord-_> e​gg. — (an likely even messier once we go to fancier trajectory representations)
<discord-_> e​gg. — ultimately, there likely is room for improvement, but nothing easy; and we would rather spend a year doing things that we find more interesting (including addressing other, more pressing, performance concerns) than that
<discord-_> P​teropodidae. — What are the biggest performance issues in Principia right now?
<discord-_> e​gg. — #2400 is the one we had been looking at (though right now unbreaking the whole rotating rockets thing is a priority)
<discord-_> e​gg. — (I think we now understand the rotation issue though, so there is that)
<discord-_> e​gg. — (not sure we have a solution)
<discord-_> e​gg. — #2400 is the one for which the plan is to do something a bit like
<discord-_> e​gg. — > Jean Chapront and Vũ Dương Tuyền (1984), *A new compact representation of ephemerides: application to the motion of Pluto, the Sun and the Galilean satellites of Jupiter*.
<discord-_> d​kavolis. — Is there a reason you're not saving current system state like an initial state for the game start?
<discord-_> e​gg. — well, we need to start from the beginning to get histories
<discord-_> d​kavolis. — Would be nice to have an option to truncate histories to the past few years
<discord-_> d​kavolis. — Not everyone needs the entire history
<discord-_> e​gg. — yeah, but the thing is we used to have that setting, and people then set it to 0 to make thing go fast, and then didn’t have it when they wanted it  :-p
<discord-_> d​kavolis. — That's why we can't have nice things
<discord-_> e​gg. — (in particular there is a Reach video like that where we would really like to see the ISEE-3 history but it is set to 0)
<discord-_> e​gg. — but in the longer term I would expect to be able to store years of history cheaply
<discord-_> e​gg. — using the methods described in the papers whereon the cat sleeps
<discord-_> e​gg. — in the meantime they are workarounds, though a bit annoying (you can keep a vessel but cut its history by going to it, hack gravity, unhack gravity)
<discord-_> e​gg. — in the meantime there are workarounds, though a bit annoying (you can keep a vessel but cut its history by going to it, hack gravity, unhack gravity) (edited)
<discord-_> d​kavolis. — Maybe have an option per vessel to not keep its history? With default to keep
<discord-_> d​kavolis. — Once you're done, you can toggle it
<discord-_> e​gg. — we are not really looking for quick fixes
<discord-_> e​gg. — there are enough workarounds available that it is usable
<discord-_> H​aukifile. — any ideas on how to fix the crashes? outside of just attempting a reinstall
<discord-_> e​gg. — we’d rather spend time looking at the fun papers
<discord-_> e​gg. — @Haukifile move the moon between the Earth and the Sun, at which point the next version of Principia will get released with the bugfix
<discord-_> e​gg. — (or build from head, though this is a bit nontrivial)
<discord-_> e​gg. — (some documentation may be out of date)
<discord-_> H​aukifile. — I can build it, but what do you mean with "from head"?
<discord-_> e​gg. — from the current state of master
<discord-_> H​aukifile. — oh, cool
<discord-_> H​aukifile. — thanks
<discord-_> e​gg. — (note that master may be arbitrarily broken between releases)
<discord-_> e​gg. — (so you may be trading crashes for other crashes)
<discord-_> H​aukifile. — well, at least there's a chance of working, right now it's crashing so consistently that it's basically unplayable
<discord-_> e​gg. — this is very odd, since this used to be a hard-to-reproduce bug
<discord-_> e​gg. — any new parts or new things from your usual setup?
<discord-_> H​aukifile. — It's been crashing since I updated to 1.8.1, I was on 1.7.3 until last week and don't think I encountered any crashes there
<discord-_> H​aukifile. — nothing special in the builds, but I can upload the crafts if it helps
<discord-_> e​gg. — hm interesting that it changes with the version
<discord-_> H​aukifile. — it's mostly a similar setup , but not necessarily exactly the same mods, so there might be something else causing it
<discord-_> e​gg. — basically things getting accelerated a lot immediately after they decouple would be the cause
<discord-_> H​aukifile. — that makes sense. I had some cases where jettisoning fairing and igniting ullage motors at the same stage crashed the game, but if I jettisoned manually and then burned, it didn't crash
<discord-_> H​aukifile. — but right now it crashes every time i decouple my lunar impactor from the srb
<discord-_> B​utcher. — @egg that paper seems to have put your cat to sleep.
<discord-_> e​gg. — yes
<discord-_> e​gg. — it is snoring as we speak
<discord-_> B​utcher. — 😸
<UmbralRaptop> 🐈
<_whitenotifier-d13c> [Principia] eggrobin labeled pull request #2527: Avoid exceptions on very long flight plans - https://git.io/JvjZE
<_whitenotifier-d13c> [Principia] pleroy closed pull request #2527: Avoid exceptions on very long flight plans - https://git.io/JvjZE
<_whitenotifier-d13c> [Principia] pleroy pushed 2 commits to master [+0/-0/±2] https://git.io/JfehD
<_whitenotifier-d13c> [Principia] pleroy bc5d7e1 - Avoid exceptions on very long flight plans.
<_whitenotifier-d13c> [Principia] pleroy 9067d8c - Merge pull request #2527 from pleroy/LongFlightPlan Avoid exceptions on very long flight plans
egg|laptop|egg has quit [Remote host closed the connection]
Wetmelon has joined #principia
egg|laptop|egg has joined #principia
<discord-_> P​teropodidae. — Is discussion of arbitrary math topics on-topic here?
<discord-_> n​eph. — I don't believe there is an off-topic here
<discord-_> n​eph. — So about that rotation around a temporal axis...
<discord-_> e​gg. — yeah this channel is not very topical
<discord-_> P​teropodidae. — Summary: Someone else on a different server posted an image saying one way to wrap your head around higher dimensions is to imagine the 4th and higher dimensions as temporal dimensions
<discord-_> P​teropodidae. — And my gut reaction was "that's not right", but I'm not really sure what a more rigorous refutation would be
<discord-_> e​gg. — that can sometimes be useful, but not as much as you might like
<discord-_> P​teropodidae. — Eventually thought that one reason I might not have liked it was because some spatial operations may not intuitively apply to temporal dimensions
<discord-_> e​gg. — you end up looking at slices
<discord-_> P​teropodidae. — Such as rotation
<discord-_> n​eph. — How do you generalize rotation? It's clearly something that has to do with two spatial dimensions, but I can't say I've ever thought very deeply about it as a concept besides the obvious ways it can be represented as a permutation group
<discord-_> e​gg. — and you end up having a preferred direction
<discord-_> e​gg. — @neph first I would say there is plenty enough to do trying to understand 3d rotation
<discord-_> e​gg. — re. time, having a preferred direction is massively weird
<discord-_> n​eph. — How do you generalize rotation? It's clearly something that has to do with a minimum of two spatial dimensions, but I can't say I've ever thought very deeply about it as a concept besides the obvious ways it can be represented as a permutation group (edited)
<discord-_> e​gg. — permutation group? I don't think that means what you think it means
<discord-_> e​gg. — (a permutation group is a finite group, spatial rotations form a Lie group, i.e. a group to which you can do calculusy things)
<discord-_> n​eph. — It's been a while since I thought about algebras much
<discord-_> e​gg. — they live in a nontrivial 3d manifold, which is already extremely hard to wrap one's mind around :-p
<discord-_> n​eph. — Is a permutation group a group whose elements are permutations of a set? If so then no, it's not what I thought it meant
<discord-_> n​eph. — aha
<discord-_> n​eph. — I think we might be talking past one another a bit, you're referring to the set of rotations of an object in a given space, but I'm still stuck on what the most general way you can define "rotating" is
<discord-_> n​eph. — eg: rotations about a point clearly exist in a 2+1 universe
<discord-_> n​eph. — Do they exist in a 1+2?
<discord-_> e​gg. — @neph orientation-preserving isometry.
<discord-_> e​gg. — that is what a rotation is.
<discord-_> n​eph. — Forgive me for ignoring the well understood in favor of the ill-defined and hypothetical
<discord-_> e​gg. — there is nothing ill-defined or hypothetical here
<discord-_> e​gg. — you are in the realm of maths, even though you pretend to speak of physics
<discord-_> n​eph. — How would you define orientation?
<discord-_> e​gg. — (the sane generalization of the cross product)
<discord-_> n​eph. — I remember exterior products but not well enough...
<discord-_> P​teropodidae. — ....So if I wanted to learn the stuff you guys are talking about, what classes would I want to take?
egg|laptop|egg has quit [Read error: Connection reset by peer]
<discord-_> n​eph. — An undergrad topo seminar gave me just enough info to pretend I can follow egg
<discord-_> e​gg. — @Pteropodidae to be able to understand spatial rotations and rigid motions in 3D, I would recommend taking a differential geometry course
<discord-_> e​gg. — I think that is a graduate maths class, so then work your way up the chain of prerequisites :-p
<discord-_> P​teropodidae. — Yeeaaaahhh, that might be a bit hard
<discord-_> P​teropodidae. — Looks like it's reachable by undergrads, but it's probably going to be hard to fit into my schedule
<discord-_> e​gg. — maybe I should write a thing that tries to get there without going too far astray someday
<discord-_> e​gg. — if I find a source of infinite time somewhere
<discord-_> S​ir Mortimer. — > An undergrad topo seminar gave me just enough info to pretend I can follow egg
<discord-_> S​ir Mortimer. — @neph oh that’s easy. You just have to loose all fear of Greek letters 🙂
<discord-_> e​gg. — 𒀝𒅗𒁕𒀀𒄿𒉌𒅅𒁉 then
<discord-_> e​gg. — muahahaha
<discord-_> S​ir Mortimer. — I even named a variable α in code, which I wouldn’t have dared before
<discord-_> S​ir Mortimer. — Also I added hundreds of hardcoded lines of numbers
<discord-_> n​eph. — fish with chain link armor microphone fallen over sheet music stand dropped knives a family of dropped knives jacked up guitar neck office building dizzy seagulls really jacked up guitar neck?
<discord-_> S​ir Mortimer. — Things you learn from looking into principia code 😉
<discord-_> n​eph. — So what I'm thinking is you'd have to define some metric over both spatial and temporal dimensions to start talking about isometry in mixed dimensions
<discord-_> P​teropodidae. — > if I find a source of infinite time somewhere
<discord-_> P​teropodidae. — I think I had one hidden behind the grill in the shed somewhere...
<discord-_> e​gg. — @neph a sort of, say, pseudo-Riemannian metric? :-p
<discord-_> n​eph. — /me runs in fear
<discord-_> e​gg. — (you have such a thing, in flat spacetime it is the Minkowski metric)
<discord-_> n​eph. — \me runs in fear (edited)
<discord-_> e​gg. — but then distances are not necessarily positive
<discord-_> n​eph. — \me runs in fear 😠 (edited)
<discord-_> S​ir Mortimer. — As anyone with a basic understanding of maths can see what these numbers obviously are <https://github.com/Kerbalism/KerbalismContracts/blob/8cc6b8b20f68cc7aa7d44d82627b1f8bf63d27c7/src/KerbalismContracts/SubRequirements/ObserveBody.cs#L256>
<discord-_> e​gg. — (see timelike vs. spacelike)
<discord-_> n​eph. — Of course
<discord-_> A​mpersand. — complex valued spacelike momentum vector
<discord-_> n​eph. — So the history of a particle in a 1-1 rotation around some temporally fixed point would be closed loop
<discord-_> P​teropodidae. — @Sir Mortimer o_O
<discord-_> n​eph. — Because it cannot be contained within its light cone?
<discord-_> n​eph. — I feel like I'm still stuck on the idea of circular motion when I should be thinking more generally
<discord-_> e​gg. — well, you said rotation
<discord-_> e​gg. — not rotational motion
<discord-_> e​gg. — rotation is not a motion
<discord-_> e​gg. — it is merely a linear transformation
<discord-_> e​gg. — an instantaneous mapping of some basis to another
<discord-_> e​gg. — no time-dependent anything here
<discord-_> n​eph. — How is rotation expressed on the minkowski manifold?
<discord-_> S​ir Mortimer. — I learned to code at a time when Unicode was an obscure draft. it took many years for computer languages to accept the fact that most of the world doesn’t exclusively use the us-ascii character set, and since that finally happened, beauties like this here are possible
<discord-_> e​gg. — @neph read https://en.wikipedia.org/wiki/Poincar%C3%A9_group
<discord-_> n​eph. — oh
<discord-_> e​gg. — https://en.wikipedia.org/wiki/Lorentz_group is the bit that doesn’t move the origin (like rotations—or rather orthogonal maps—within the Euclidean group)
<discord-_> n​eph. — I see
<discord-_> n​eph. — So it is well defined
<discord-_> n​eph. — Very well studied, to boot
<discord-_> n​eph. — For a 2+2 with temporal dimensions t and T would the analogous group be all isometries preserving t^2+T^2-x^2-y^2?