19:03 <lamont> Pontryagin hates me

19:12 <egg|zzz|egg> lamont: how so?

19:12 <lamont> optimizer is just being weird

19:13 <egg|zzz|egg> !wpn -add:adj Понтрягин

19:13 <Qboid> egg|zzz|egg: Adjective added!

19:13 <egg|zzz|egg> !wpn lamont

19:13 <lamont> it may actually be Levenberg and/or Marquardt that hate me

19:14 <egg|zzz|egg> lamont: also somebody in china is interested in your MJ work: #1871

19:14 <Qboid> [#1871] title: Support with a mod that could do maneuver nodes automatically | Another request, now in RO, it's seriously such a hard thing to adjust orbit in a high accuracy with hand, and now, I didn't find a mod that could do maneuver nodes automatically, which is hard when y... | https://github.com/mockingbirdnest/Principia/issues/1871

19:15 <lamont> yep, i’m in the middle of a big refactor to support maneuver nodes

19:20 <egg|zzz|egg> lamont: well it would have to have a dedicated principia manoeuvre path, you can't work through the stock nodes with principia, as discussed

19:21 <egg|zzz|egg> lamont: what do you need to optimize in the principia case?

19:22 <lamont> yeah but to solve stock maneuver nodes i need to solve handling coast phases and specifically the coast-burn-coast rendezvous problem

19:22 <GH> [Principia] lamont-granquist commented on issue #1871: to a certain extent interplanetary transfers will always require tweaking flybys with corrective burns. 0.01 m/s accuracy is most likely completely unreasonable. at some point the amount that orbits hop around in KSP due to floating point error will get magnified by each flyby and will always become wildly incorrect.... https://git.io/fbpcB

19:26 <egg|zzz|egg> and yeah, TCMs TCMs TCMs

19:27 <egg|zzz|egg> (but having a good node executor would be cool)

19:27 <lamont> Turner Classic Movies?

19:27 <egg|zzz|egg> (manoeuvre executor, we don't use the word node in principia)

19:27 <egg|zzz|egg> TCM?

19:27 <Qboid> egg|zzz|egg: [TCM] => Trajectory Correction Manœuvre

19:29 <lamont> yeah, at some point Lyapunov has his way with your careful trajectory planning

19:50 <lamont> how can i start a root finding problem that already matches to 3.77261133121446E-11 and have the optimizer wander away the point where its at 14.4401056475246 when it gives up?