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
egg|cell|egg has quit [Remote host closed the connection]
egg|cell|egg has joined #principia
<discord->
newtype_. — JNSQ looks cool, is it murder on CPU with so many bodies?
<discord->
Acer_Saccharum. — JNSQ actually doesn't have much of a performance impact in my experience
<discord->
Acer_Saccharum. — And I don't have a monster PC that can just run anything for reference
<discord->
Acer_Saccharum. — And I don't have a monster PC that can just run anything at 144fps for reference (edited)
<discord->
newtype_. — neat might try it, enjoying MKS but stock system doesnt have many places for functional outposts for refueling and stuff
<discord->
Acer_Saccharum. — What will hurt your CPU is any asteroids that spawn with JNSQ
<discord->
newtype_. — can i disable asteroids i dont even care about them
<discord->
Acer_Saccharum. — Principia's performance scales roughly with the amount of vessels
<discord->
Acer_Saccharum. — Yes you probably should
<discord->
newtype_. — i have comm stuff off because i want more functional vessels not like half my cpu going to a comm network
<discord->
Acer_Saccharum. — Just don't install the custom asteroids config, and delete the actual asteroid part from gamedata/squad/.../potatoroid
<discord->
newtype_. — neat ty
<discord->
Acer_Saccharum. — At least I think that's what the asteroid part is called
<discord->
newtype_. — (this mod is so cool)
<discord->
_felixu_. — > can i disable asteroids i dont even care about them
<discord->
_felixu_. — Holy crap can I do this in RSS? *I. Don't. Even. Look. At. Them.*
<discord->
newtype_. — been playing since .2 (on and off) and have never wanted to visit an asteroid
<discord->
_felixu_. — No but seriously, where can I find the option in settings? I imagine it would benefit my performance too
<discord->
lpg. — you can't. you delete the file/folder Acer mentioned
<discord->
_felixu_. — Just saw that, that's brilliant. I think I'll do just that
Mike` has quit [Ping timeout: 190 seconds]
Mike` has joined #principia
raptop has quit [Ping timeout: 190 seconds]
egg|laptop|egg_ has joined #principia
egg|cell|egg has quit [Ping timeout: 189 seconds]
egg|laptop|egg has quit [Ping timeout: 378 seconds]
<discord->
Damien. — @egg have you got a link to that design document where you listed all the interesting orbit types for the missions you were working on with Sir Mortimer?
<discord->
egg. — @Sir Mortimer should that be in your KC repo?
<discord->
Sir Mortimer. — Yes, was too lazy to move it
<discord->
Damien. — today in cat news: I had to smash a hole in my bathroom vent to release a trapped bird that got in from outside. Cat is sulking that she couldn't have the bird, after alerting us to its location in the first place
<discord->
Damien. — also I'm blaming you for me buying a gyreggscope
<discord->
Damien. — precession is cool so I wanted one
<discord->
Damien. — @Pap do you still have that link to the principiacontracts mod you updated?
<_whitenotifier-d13c>
[Principia] eggrobin commented on issue #2519: Spin-up on reentry under physics warp with FAR - https://git.io/JfBiD
<discord->
egg. — @Damien commented; basically, with Principia things will behave differently (because they are wrong in stock), but while your sounding rockets might spin up on uncontrolled reentry (as real ones do), they should not spin their way up to 500 rpm
<discord->
Damien. — sounds good
<discord->
lpg. — > This will, in particular, have the effect of making timewarp equivalent to non-timewarp as long as the vessel remains rigid and free of torque.
<discord->
lpg. — satisfying indeed
<discord->
Damien. — what was the difference in time/non-timewarp before?
<discord->
egg. — in non-timewarp, we were trusting PhysX and doing something which we assumed was going to be a minor correction for model errors (fuel-induced spin/despin not being taken into account, etc.)
<discord->
egg. — it turns out that PhysX was not to be trusted, and so code that was written with the intention of being a small correction was called upon to, effectively, solve a very finnicky differential equation (Euler’s equation for rigid body rotation)
<discord->
Damien. — > *PhysX is crap.*
<discord->
egg. — so it would try its best to do the same thing as the much fancier timewarp code, but do so in a very brutish manner
<discord->
Damien. — new principia motto
<discord->
egg. — now that we know what we are up against, we just use the fancy timewarp code (and we even have a rationale for why that makes sense)
<discord->
egg. — (as long as you don’t physics warp, because that changes the timestep and you are not allowed to do that to a symplectic method ; so each physics warp or dewarp will introduce errors)
<discord->
egg. — (but non-physics warp-dewarp is fine, since we are using the same mechanism and exactly computing the warp evolution)
<discord->
Damien. — that's good news
<discord->
Damien. — the physx thing was sounding like a showstopper at first
<discord->
egg. — no it was a major milestone, suddenly we understood what we had been trying to fight all that time
<discord->
egg. — prior to it we were about to give up, because we had no idea what was going on
<discord->
Damien. — lets be fair, if you two couldn't solve it, no one could
<discord->
egg. — at this point we have heavy enough artillery that working around PhysX is easy
<discord->
egg. — @ferram4 also, this means we will effectively be treating FAR (among other things) as the right-hand-side of a fancy integrator with a splitting of the equation of motion into Euler’s equation and ~aerodynamic magic~, which is conceptually cool
<discord->
Damien. — *conceptually cool* being the main driving force in principia project development
<discord->
egg. — very much so
<discord->
Damien. — *starts making plans to have UI development sound 'conceptually cool'*
<discord->
ferram4. — ...I see I have a bunch of stuff to read
<discord->
ferram4. — Also, hi
<raptop>
A wild ferram4 appears!
<discord->
Sir Mortimer. — > (The interested reader can look at the definition of the polhode and the herpolhode in Wikipedia in French, German or Spanish; the English article is lame.)
<discord->
Sir Mortimer. —
<discord->
Sir Mortimer. — I’m still laughing 😂
<discord->
ferram4. — @egg So for someone who's a little out of the loop from not staring at this sorta stuff for awhile, what's the source of the oscillations in all of those graphs?
<discord->
egg. — the purple graph is probably OK, the yellow graph is what happens if I try to be too smart in the worst way possible
<discord->
egg. — the blue graph is stock, which probably moves unphysically
<discord->
egg. — (that particular example is stock stock, no FAR)
<discord->
ferram4. — I meant more along the lines of, "what's happening that causes the angular frequency and such to oscillate all over the place even in the more accurate graphs"
<discord->
egg. — well, the booster flops around in the breeze
<discord->
ferram4. — Like, what's the physical source of that behavior
<discord->
ferram4. — Oh, just trying to find a stable orientation?
<discord->
egg. — yeah, and when spun up the righting forces lead to further oscillation rather than just putting it straight in the airstream
<discord->
ferram4. — Interesting
<discord->
ferram4. — Go home PhysX, you're shitfaced
egg|laptop|egg has joined #principia
egg|laptop|egg_ has quit [Ping timeout: 378 seconds]
<_whitenotifier-d13c>
[Principia] eggrobin edited a comment on issue #2519: Spin-up on reentry under physics warp with FAR - https://git.io/JfBiD
<discord->
Damien. — > This is not how physics works in this universe.
<discord->
Damien. — please educate the physX team in exactly this tone 🙂
egg|laptop|egg has quit [Remote host closed the connection]
egg|laptop|egg has joined #principia
egg|laptop|egg has quit [Read error: Connection reset by peer]
<discord->
lpg. — I've shared the issue's link with a friend at nVidia. physx team may hear about it
<discord->
lpg. — I've shared the issue's link with a friend at nvidia. physx team may hear about it (edited)
<discord->
Damien. — Haha
<discord->
Damien. — Don't you dare edit it egg
<discord->
egg. — I can’t imagine this being fixed ; it is a reasonably well-known property of PhysX (you have other engines that advertise themselves as doing that right, explicitly contrasting with it) and lots of PhysX users will rely on the broken behaviour
<discord->
Andrew Chen. — Out of curiosity, would this "behavior" cause axially symmetric satellites to have phantom torque exerted by the apogee kick stage, and/or break spin-stabilized upper stages
<discord->
egg. — I don’t know ? I don’t think so ? I don’t think it is a reasonable idea to attribute any random bug to that. If anything, the PhysX thing makes rotation artificially stable
<discord->
Damien. — Could it be that persistent rotation handled it more simplistically and unrealistically. What we're experiencing now could be how it should be
<discord->
egg. — yeah persistent rotation was wrong, that we knew, in exactly the same way that PhysX is (except that it treated the whole vessel as PhysX treats a single part)
<discord->
lpg. — friend's response: "Reported it to their VP"
<discord->
Damien. — I've noticed spin stabilised stuff still moves along the other axises (axes?) But I'm guessing that's the intermediate axis theorem at work
<discord->
egg. — well, for a single part, the issue is that PhysX does not give you the intermediate axis theorem
egg has quit [Read error: Connection reset by peer]
<discord->
egg. — it kind of gives it to you if you have multiple parts (because ultimately the equations of rotational motion for rigid bodies follow from the equations of linear motion for point masses; if you build a thing out of bits, the (properly integrated) motion of bits will cause it to exhibit the джанибеков effect
egg has joined #principia
<discord->
egg. — @Damien now bear in mind that you can also just have an angular velocity that is not aligned with an axis of symmetry of the vessel; that will look a it wobbly, but that is purely geometric wobbliness, not physics
<discord->
lamont. — doesn't physx know the inertia tensor of a part?
<discord->
egg. — (the physics would be the movement of that angular velocity with respect to the vessel, following the polhode, and in space, following the herpolhode; but you have to know what to look for, or to take an example that is glaring, such as the rover body along its second axis)
<discord->
egg. — @lamont yup
<discord->
egg. — but it does not know about Euler’s equation
<discord->
lamont. — yeah unity ribidbodies have inertia tensors
<discord->
lamont. — so in principle this is fixable?
<discord->
egg. — yup, they just don’t use them properly
<discord->
lamont. — 'k
* discord-
lamont. — hops on board the "PhysX is crap" bandwagon_
<discord->
egg. — yeah; how well things would behave with a simple fix (cartesian symplectic integration of Euler’s equations) is not immediately obvious to me
<discord->
egg. — there is a non-simple fix, which is the sort of thing we do, which is far more robust, but this took us a year of development and finding sign errors in papers of (and reporting bugs to) domain experts (and a bit of actual maths of our own to get a full solution)
<discord->
egg. — that is to exactly integrate free-rigidbody rotation
<discord->
egg. — conceptually, the same thing you do when you integrate linear motion : q += v Δt is an exact solution of the motion of a body moving at v Δt in the absence of forces
<discord->
egg. — conceptually, the same thing you do when you integrate linear motion : q += v Δt is an exact solution of the motion of a body moving at v for Δt in the absence of forces (edited)
<discord->
egg. — and the simplest splitting consists in doing p += forces(q) Δt; q += p/m Δt
<discord->
egg. — where the += forces is the bit that you do in the game logic, usually
<discord->
egg. — that side stays essentially the same, L += torques(attitude, etc.) Δt
<discord->
egg. — but the q += v Δt becomes very hairy, and involves elliptic integrals
<discord->
lamont. — ah, i think you've posted Mr. Fuk's references in the pastd
<discord->
lamont. — ah, i think you've posted Mr. Fuk's references in the past (edited)
<discord->
egg. — ah, of course, I run into issues when the vessel is almost symmetrical but a bit wobbly, because the principal axes are singular and I use them as a reference…
<discord->
egg. — not quite out of the woods yet
<discord->
egg. — I should correct based on the axes of a part, not the principal axes