UmbralRaptor changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> … one of the other grad students just compared me to nomal O_o | <ferram4> I shall beat my problems to death with an engineer.
<kmath>
<twomiletower> behold, the ridiculous state of British railways: I CAN LITERALLY GET RETURN TICKETS TO NEW YORK CITY FOR THE LEICE… https://t.co/EhyhArhiI6
<UmbralRaptor>
Those are remarkably cheap plan tickets.
icefire has quit [Read error: -0x1: UNKNOWN ERROR CODE (0001)]
<Fiora>
whitequark... isn't widely known to be a cat?
<kmath>
<DrPhiltill> Am I the only person who is skeptical about this paper? Non-elliptical (similar to cardioid) orbits around stars? https://t.co/YUbu9OqJLj
regex has joined #kspacademia
Snoozee has joined #kspacademia
Snoozee is now known as Majiir
regex has quit [Remote host closed the connection]
<egg|work|egg>
+some_lambda to make it a pointer to a function
<Iskierka>
So Labradors come from Newfoundland and Newfoundlands come from both Newfoundland and Labrador
<Iskierka>
I think Canada pulled an iceland
<egg|work|egg>
TIL that Newfoundland is a type of dog
<egg|work|egg>
Iskierka: why iceland?
<Iskierka>
a big fluffy one that somehow comes from the same original breed as labradors
<Iskierka>
Iceland swapped names with greenland to discourage tourists
<Iskierka>
(More or less)
<egg|work|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a suitcase
<egg|work|egg>
UmbralRaptor: but then again it does say in that potential, not in a perturbed potential; I can't make sense of what the paper is trying to say
<egg|work|egg>
bofh: any idea?
<egg|work|egg>
yeah no, they don't seem to have anything else on their potential
<bofh>
egg|work|egg: sec
<egg|work|egg>
well if it's that it's obviously bunk, but then that's so obviously bunk I'm reluctant to interpret it as being that
<egg|work|egg>
The twitter thread mentions that things can look like that in a rotating frame, which is true
<kmath>
<DrPhiltill> Am I the only person who is skeptical about this paper? Non-elliptical (similar to cardioid) orbits around stars? https://t.co/YUbu9OqJLj
<BPlayer>
Hi! So, I was thinking recently. The thrust, gravity acceleration and drag experienced by a launch vehicle in KSP all are predictable and can be expressed by fairly simple functions. Now, couldn't one use calculus to calculate a whole launch using a single function with some tweakable parameters?
<BPlayer>
(I'm still under the impression of all the possibilities that open up once you realize that acceleration is the second derivative of position, and so on.)
<egg|work|egg>
whitequark: yay more cat pics \o/
<egg|work|egg>
"fairly simple functions" is unfair to Navier-Stokes, but yes, you can integrate a launch (after all that's just what KSP is doing, albeit with a shitty integrator)
<egg|work|egg>
(and also KSP puts a human in the loop that may give control inputs, so you need to replace that with some function of the phase space)
<BPlayer>
I am assuming no lift, which seems reasonable with a rocket.
<BPlayer>
So it does seem to boil down to simple functions...
<egg|work|egg>
well there's some manner of guidance (or aerodynamic forces turning your rocket), you're not flying straight up
<BPlayer>
Let me rephrase that as "I am assuming a rocket that is aerodynamically stable towards prograde"
<BPlayer>
Anyway, has anyone done such a thing before and uploaded the results to the internet?
<egg|work|egg>
but to answer your question it doesn't matter if the RHS is simple or complicated, it is in principle possible
<BPlayer>
And I mean not in principle but in practice. :P
<egg|work|egg>
well in practice you really want some sort of closed-loop guidance when launching a rocket
<BPlayer>
I was hoping to avoid that. It seems so complicated...
<egg|work|egg>
well on occasion rocket engineering is complicated :-p
<egg|work|egg>
BPlayer: ask lamont about guidance though, he's been working on that
<BPlayer>
You suggested that a while ago, IIRC... Where can I find him, again?
<egg|work|egg>
here :-p
<BPlayer>
Ah, woops. Must have missed his name at first glance. Sorry.
<egg|work|egg>
(now, *when* you can find him is a different question; he'll be there eventually)
<BPlayer>
That would be the next question. :P
<BPlayer>
Is it a faux pas to ;tell him my question?
<egg|work|egg>
no
<BPlayer>
Very well, then we have a plan. :D
<BPlayer>
;tell lamont Hi! People sent me to you when I asked about ascent guidance, as you were doing such stuff recently. The question would be what general approach is simple enough and sufficient for use in KSP and kOS, I guess...
<kmath>
BPlayer: I'll let lamont know when I see them
ferram4 has joined #kspacademia
<egg|work|egg>
!wpn котя and whitequark
* Qboid
gives котя and whitequark a Kozai Van Allen wrench
<UmbralRaptor>
… does the kozai mechanism apply to a wrench moving through a B field?
* UmbralRaptor
hopes BPlayer is not eggspecting an analytic solution.
<egg|work|egg>
UmbralRaptor: hah
<egg|work|egg>
yeah
<egg|work|egg>
UmbralRaptor: have you seen whitequark's latest catpics
<UmbralRaptor>
I think so. Kotya in a harness grooming, uh, however you transliterate the other one.
* UmbralRaptor
? ?
<egg|work|egg>
UmbralRaptor: ... I think you're swapping the cats again
<UmbralRaptor>
blarg
<UmbralRaptor>
I was fooled by the sizes.
<egg|work|egg>
котя is the larger one
<egg|work|egg>
UmbralRaptor: and there's also a photo of котя in a harness staring at a tree
<UmbralRaptor>
The black and white one looked larger >_>
<BPlayer>
UmbralRaptor: While I remember, shall I change anything with your TelescopeRaptor or is it fine?
<whitequark>
the black and white one is *tiny*
<whitequark>
it fits onto a palm
<UmbralRaptor>
!
<egg|work|egg>
!!
<egg|work|egg>
UmbralRaptor: FWIW their names transliterate to kotâ and èta hujnâ under ISO 9
<kmath>
<JossBlandHawtho> #IAU_Kavli Bob Abraham argues that the only life signature he cares about in exoplanet science is caffeine. Do you have good coffee?? Yeah.
<bofh>
UmbralRaptor: I can sympathize/agree with that fwiw.
<bofh>
UmbralRaptor: you mean эта хуйня?
<UmbralRaptor>
For the black and white cat, apparently.
<kmath>
<wellington_west> Good morning #Ottawa! We're not kitten around when we say @FelineCafe makes a great cuppa! https://t.co/OTyS6DZEuI
<UmbralRaptor>
BPlayer: which part? whitequark's cats?
<BPlayer>
I think so.
u_nuSLASHkm8 has joined #kspacademia
u_nuSLASHkm8 has left #kspacademia [#kspacademia]
<egg|zzz|egg>
!wpn Fiora
* Qboid
gives Fiora a superconducting trapezohedron
<egg|zzz|egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a [DATA EXPUNGED] tube
<egg|zzz|egg>
!wpn BPlayer
* Qboid
gives BPlayer a dorsal fish
egg|zzz|egg is now known as egg
* UmbralRaptor
watches [REDACTED]
<egg>
UmbralRaptor: I learned the other day that "tube" was used to mean television in american
<egg>
suddenly the name of the company I work for makes some amount of sense
* UmbralRaptor
hands egg a cathode ray tube.
* egg
keeps being confused by the fact that "quenching" = "trempe" and "tempering" = "revenu"
<egg>
also "trempe" means "dipping", which explains why you'd use it for "quenching" but that makes using an etymologically similar word for "tempering" extra confusing >_<
<UmbralRaptor>
quenching star formation by dipping matter into a supermassive black hole?
<UmbralRaptor>
<_<
* egg
is even more confused
<UmbralRaptor>
Oops
<egg>
I should make a journal for profile-guided optimization on the Cayley release
<lamont>
Bplayer still around?
<Qboid>
lamont: Pap left a message for you in #RO [19.07.2017 00:53:33]: "There is some strange occurance with new MJ when trying to launch into the Plane of a Target. It seems to work the first time, but then some value is not getting reset and if you do it again, it will launch automatically"
<lamont>
!tell Bplayer its not as simple as that, and you do want closed-loop guidance and it involves calculus of variations sorta graduate-level calculus
<Qboid>
lamont: I'll redirect this as soon as they are around.
<egg>
lamont: well in the abstract, you can take any RHS and just integrate it, which is one way of reading their question :-p
<egg>
(now if your closed-loop guidance which is part of the RHS involves simulating your trajectory, this is quite silly :-p)
<lamont>
well to plan an optimal trajectory you want to minimize the dV burned or maximize the ending mass
<egg>
yeah but the question wasn't directly an optimization question
<lamont>
you can shoot and integrate and then smear a genetic algorithm or hillclimbing or whatever onto it
<egg>
yeah, assuming the shooting doesn't involve much guidance
<egg>
so, ideally, a Lambda or Mu launch
<egg>
1-parameter family of initial conditions \o/
<lamont>
once you’re exoatmospheric you have a different guidance function than you do from just following a gravity turn arc as well
<egg>
hm yeah, not sure what Lambda/Mu did on the last stage
<egg>
yeah there's some attitude control fanciness on the upper stages
<lamont>
surveyor guidance (current MJ PEG guidance) uses sin pitch = A + Bt + G/a where A and B are free parameters and G is a centrifugal force term
<lamont>
i think optimal burns are tan pitch instead of sin pitch and they were optimizing for pitch angles near zero (think IGM uses tan pitch, but i’m not quite there yet)
<lamont>
and that is provably optimum for constant thrust. but once you start talking varying ISPs in the atmosphere that goes out the window. i understand its all solvable with primer vector theory and i think by doing piecewise integration with constant thrust accelleration pieces, but that’s just a label on an empty card in my brain that i need to learn all that at some point...
<bofh>
oh for fuck's sake, what do they call long double then?
<bofh>
binary80?
<egg>
bofh: an extended format associated with binary64
<egg>
bofh: e.g. this would allow for an extended format associated with single-precision float (binary32), which would then have at least 32 bits of mantissa
<egg>
(that's what I'm getting from looking at the standard at least; I may be misreading it and saying bullshit; Fiora, do the things I say above make sense?)
egg is now known as egg|nomz|egg
BPlayer has quit [Ping timeout: 180 seconds]
APlayer has joined #kspacademia
APlayer is now known as BPlayer
<SnoopJeDi>
I stumbled on the naming quirk pretty hard a few weeks ago while trying to help somebody with a numpy question :/
<BPlayer>
lamont: Okay, I saw what you've written, although I was not around when you wrote it.
<Qboid>
BPlayer: lamont left a message for you in #kspacademia [19.07.2017 16:56:55]: "its not as simple as that, and you do want closed-loop guidance and it involves calculus of variations sorta graduate-level calculus"
egg|nomz|egg is now known as egg
<BPlayer>
Well, I was hoping not to have to do closed loop guidance and calculus, but whatever. I can handle it if needed. What I really don't want to do is AI optimization. I want to implement this in kOS, and genetic algorithms would be beyond what I am ready to do for KSP.
<BPlayer>
I am really fine with an okay guidance that just gets me to orbit in a more or less cheap trajectory. I would imagine that a precalculated pitch vs. altitude table would do for the part below 20 or 30 km altitude and closed loop guidance from there on. But that leaves me with having to calculate a table and an algorithm for guidance.
<SnoopJeDi>
if it's for on-line guidance, a genetic algorithm would be waaaaaay too slow anyway
<SnoopJeDi>
unless you evolved an algorithm that tweaked variables in some control scheme I guess
<BPlayer>
Well, my eventual goal is a kOS GUI that launchzes you into a pre selected trajectory. A DIY Mechjeb, if you want.
<lamont>
BPlayer: you should play around with the guidance that i just wrote for MJ
<BPlayer>
Uh, if I know extremely basic C++ and some JS, do you think I will be able to understand the code and maths? If so, I would be glad to see it and try to understand it myself at first.
<egg>
the language isn't be the main complexity here
<lamont>
well the last link is in kOS code
<egg>
s/isn't be/isn't going to be/
<Qboid>
egg meant to say: the language isn't going to be the main complexity here
<lamont>
and i’m meaning to write up a blog page on it all, but that competes with code writing time, and is also a time consuming project on its own
<BPlayer>
lamont: Wow, thanks a lot! I'll have a look ASAP!
<SnoopJeDi>
Looking at the "Major Loop Algorithm" section, it's a bit thin, yea
<lamont>
the “simplified” nomenclature also winds up just making it more complicated
<SnoopJeDi>
yea, it's venerable to attempt to use simple language, but not if you're going to do it where it doesn't matter and then spew equations
<lamont>
f_r = sin pitch = A + C is the actual guidance law.
<lamont>
A = A + B dt; B = B; T = T - dt; is used every tick to update that guidance
<lamont>
if you know A, B, T precisely then that’s literally all you need to fly the rocket (single stage to insertion)
<lamont>
in order to update A, B, T then what you do is using the delta-h (angular momentum to-go) which is trivial — in order to update dV-to-go and T-to-go at the bottom there
<lamont>
then since you have an updated values of dV and T you have and updated b_0, b_1, c_0, c_1 constants
<lamont>
so you feed those into the matrix solution to solve for A,B which updates A,B
<lamont>
that process converges, so that once T does not change by more than 0.1 sec in a cycle of that algorithm then you know you’re converged
<lamont>
the wiki page is written literally backwards with the matrix solution step as “guide”
<kmath>
<wellerstein> OMG OMG. Edward Teller tried a run as a television show host in 1955. How did I not know this? These photos are AMA… https://t.co/rh5xiXZ0YA
<kmath>
<FioraAeterna> is there a way to get fusion 360 to give up after N seconds of solver failing instead of locking up for the rest of the life of the universe
Technicalfool has joined #kspacademia
<UmbralRaptor>
Increase Λ (well, w) so that the time to the big rip is N seconds.
<SnoopJeDi>
I vaguely remember pondering the same question when I used Inventor a lot
Technicalfool_ has quit [Ping timeout: 186 seconds]
<egg>
UmbralRaptor: oh TIL that my new product manager at work studied astrophysics
<egg>
and then some sort of spacecraft engineering thing iirc
* UmbralRaptor
pokes egg's manager with the Faber-Jackson relation.
<egg>
it's not my manager, it's my product manager
<egg>
independent reporting chains :-p
<UmbralRaptor>
o_O
<egg>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a war-surplus chromatograph