NathanKell changed the topic of #RO to: Welcome to the discussion channel for the Realism Overhaul (meta)mod for KSP! Realism Overhaul Main Thread https://goo.gl/wH7Dzb ! RO Spreadsheet http://goo.gl/Oem3g0 ! Code of Conduct http://goo.gl/wOSv2M ! | [15:01] <soundnfury> Straight Eight Stronk (and) RP-0/1 is basically "Space Agency Spreadsheet Simulator" with a rocket-flying minigame
<Starman4308>
My greatest foes attempting to create a kOS script for landing on Phobos/Deimos are RCS overcorrections and sheer, overwhelming boredom.
<Bornholio_>
lower RCS thrust and balance using RCS build aid
<Bornholio_>
for isntance , Mercury pod at full, bounces all over on renetry, uses all HTP Quickly, at 25% it settles into nice gentle retrograde and barely uses any HT{
<Starman4308>
I've fixed the RCS overcorrections for the most part, and the boredom has been solved by using reverse RCS to accelerate me to -15 m/sec until I get close.
<Starman4308>
(kOS has a mechanism to multiply the RCS output for its cooked steering)
aradapilot has quit [Remote host closed the connection]
aradapilot has joined #RO
<lamont>
heh i barely use the docking autopilot normally with MJ
<lamont>
but the whole maneuver node system doesn’t interface with principia well, and trying to use osculating orbits for hitting 200m targets from half an orbit away in principa is going to fail because of physics and stuff…
<egg>
lamont: but now you have an endpoint! :D
<lamont>
also “congratulations you have been made a developer of the principia functionality port of mechjeb!”
<lamont>
i do
<lamont>
i should play with the damn thing now
<egg>
lamont: eggsactly
<lamont>
Clifford?
<egg>
lamont: also I like that you use RIGHT SINGLE QUOTATION MARK etc. correctly
<egg>
lamont: ?
<egg>
lamont: the release that will include your endpoint is Coxeter
<lamont>
ah so master?
<lamont>
“at first i was but a learner, now i am on the master branch…”
<lamont>
egg: i made progress at two-point-boundary-value-problems
<lamont>
and it turns out matlab has a levenburg-marquardt solver as a choice for fsolve() implementation and that code runs fine with it, and alglib that is already in mechjeb has an L-M optimizer
<lamont>
and i’ve got C# code for rkf45 kicking around
<lamont>
so put it together and weld it to mechjeb and we get proper grown-up trajectory optimization
<Bornholio_>
.cheer
Bornholio_ is now known as Bornholio
<egg>
so you'd need the integration of some slightly odd ODE for that?
<egg>
and I guess some sort of metric to optimize on the trajectory?
<lamont>
yeah that’s what that does
<lamont>
you’ve got a normal initial value ODE problem that the integrator deals with — but you know only half your starting conditions and half your ending conditions (ish) and you have metric to optimize
<egg>
lamont: yeah but I guess at least the integration and metric computation should be done in principia
<lamont>
then mumble mumble pontrayagin mumble weirstrauss-erdman corner conditions mumble mumble and you can turn the TPBVP into a root finding problem
<egg>
that's a good spec
<egg>
I like it
<egg>
lamont: can you specify your ODE in a way that doesn't require reading too much matlab in stackoverflow though
<lamont>
yeah i need to play with it more
<egg>
(also your metric ideally, though that's likely reference-frame dependent in some fancy way that needs to be generalized for principia fanciness
<egg>
)
<lamont>
typically you optimize time, or final mass
<lamont>
you can also minimize the integral of the dV over the burn
<lamont>
but those should be reference frame independent
<lamont>
i see green tests on master
<lamont>
the suborbital hop problem is probably solvable via similar techniques but the end condtions, transversality conditions and the metric is going to be different...
<lamont>
that’s the problem of trying to completely burn a rocket that won’t make it to orbit but deliver it to, say 150km with maximum velocity
<lamont>
i see the usual red tests
<egg>
lamont: uh, surely you're also optimizing towards some target trajectory?
<lamont>
the rest of the parameters can be entirely free if you don’t care about inclination / LAN / etc. you’re just trying to deliver a rocket to 150km with as much horizontal velocity as possible.
TM1978m has joined #RO
<egg>
lamont: I mean for optimizing to a trajectory (which I thought was the goal with that fancy optimizer? how else am I going to get MJ to eggsecute my weak stability boundary transfers :-p)
probus_ has quit [Quit: Leaving]
Majiir is now known as Snoozee
<awang>
Anyone want to try out a 1.4.1 TF dll?
<awang>
I need to see whether 1.4.1 DLLs compiled on mac work on Windows
BasharMilesTeg_ has joined #RO
<lamont>
yeah but final alt = 150km and velocity = max actually will select a particular best trajectory. you can certainly specify all 6 keplerian elements and do the equivalent of an intercept (*cough* docking autopilot *cough* hohmann transfer *cough*)
<lamont>
that is actually much easier
<lamont>
no transversality conditions
BasharMilesTeg has quit [Ping timeout: 186 seconds]
<awang>
Bornholio: Got a link to the 1.4.1 golden spreadsheet?
SpecimenSpiff has joined #RO
<SpecimenSpiff>
o/
SpecimenSpiff has quit [Quit: Web client closed]
Olympic1 has quit [Ping timeout: 186 seconds]
<soundnfury>
awang: yes I'm on linux (Debian to be specific)
TM1978m has quit [Remote host closed the connection]
blowfish has joined #RO
blowfish has quit [Remote host closed the connection]
blowfish has joined #RO
ProjectThoth has quit [Ping timeout: 383 seconds]
KevinStarwaster has joined #RO
Luciano has quit [Ping timeout: 198 seconds]
Starwaster has quit [Ping timeout: 186 seconds]
TM1978m has joined #RO
Technicalfool_ has joined #RO
Technicalfool has quit [Ping timeout: 186 seconds]
Mike` has quit [Ping timeout: 198 seconds]
BasharMilesTeg has joined #RO
BasharMilesTeg_ has quit [Ping timeout: 186 seconds]
Mike` has joined #RO
BasharMilesTeg_ has joined #RO
BasharMilesTeg has quit [Ping timeout: 186 seconds]
ProjectThoth has joined #RO
Mike` has quit [Ping timeout: 186 seconds]
Mike` has joined #RO
blowfish has quit [Quit: Leaving]
Senshi has joined #RO
TM1978m has quit [Remote host closed the connection]
Shoe17 has joined #RO
ProjectThoth has quit [Ping timeout: 383 seconds]
<egg>
lamont: yes, but I mean for the general case you don't want to guide to a Keplerian orbit
<egg>
lamont: e.g. WSB transfers
<egg>
lamont: so you'd have to have a way to compute the residual wrt to a trajectory in a chosen reference frame
Olympic1 has joined #RO
schnobs has joined #RO
<schnobs>
o/
Maxsimal has joined #RO
ProjectThoth has joined #RO
Olympic1 has quit [Ping timeout: 186 seconds]
<schnobs>
Hmmm. So we carefully align our space centers so the runway points due east, the rotate all rockets by 90 degrees.
<Qboid>
BadRocketsCo: awang left a message for you in #RO [14.03.2018 12:32:42]: "For gravity assists, there's the KSP Trajectory Optimization Tool. Technically not a mod though"
<BadRocketsCo>
Oh yeah. That's what I thought of
<awang>
acharles: Thanks!
<awang>
That tells me that things compiled on mac will *not* work on Windows
<awang>
:(
<awang>
Seems Unity is looking for the mac-specific modules
<awang>
egg: ^
<Mike`>
theres also flyby finder, BadRocketsCo
<awang>
Sarbian: ^
<awang>
Assuming I didn't screw things up, at least
<awang>
I find it odd that there doesn't appear to be any references to the split DLLs on the forums
<awang>
At least through google
<egg>
awang: ow
<egg>
awang: but the reverse is fine, right?
ProjectThoth has joined #RO
VanDisaster has quit [Ping timeout: 190 seconds]
VanDisaster has joined #RO
VanD has joined #RO
VanDisaster has quit [Ping timeout: 198 seconds]
VanD has quit [Client Quit]
VanDisaster has joined #RO
BadRocketsCo has quit [Ping timeout: 383 seconds]
VanDisaster has quit [Ping timeout: 190 seconds]
VanDisaster has joined #RO
schnobs has joined #RO
<schnobs>
o/
Probus has joined #RO
<Probus>
o/
<schnobs>
Procedural Spaceplane Tank seems a bit off.
<schnobs>
a) it has size limits
<schnobs>
(said limits refer to techs that no longer exist)
<schnobs>
b) utilization defaults to 86 and can be cranked up to 100.
<schnobs>
I know how to deal with a), no clue about b).
<schnobs>
also, repost because weekand and possibly a new audience: