These are chat archives for esa/pagmo/expsin

10th
May 2015
Dario Izzo
@darioizzo
May 10 2015 18:32

Welcome to EXPSIN project channel!

Hey, this is the gitter.im channel for the EXPSIN project. As mentioned previously, all related discussions and ideas on the project should go here.
.. and be subjected to harsh, harsh criticism by the public. We plan to keep it as a future reference log as well so try to keep it on the topic ;)
In this project we are trying to add to PyKEP a lambert solver for exponential sinusoids. The reference paper is http://arc.aiaa.org/doi/abs/10.2514/1.21796?journalCode=jgcd
Dario Izzo
@darioizzo
May 10 2015 18:37
A first proof-of-concept showing the interface and all, was developed by Chris and can be found here https://docs.google.com/document/d/1d94cozzxBrxa-rdg7WY-Wm1sCB2QuFKHyPlfSDZNJtg/edit
Dario Izzo
@darioizzo
May 10 2015 18:46
  • The interface and naming should mimic the lambert_problem class already in PyKEP. So I suggest the name to be lambert_problem_expsin. The problem is solved upon construction and the border velocities extracted via the get_v1, get_v2 methods.
  • Would be nice to see if we can detected the maximum number of revolutions automatically (as done for the Lambert problem) but I am not sure it is practical in this case, so the m_multi_revs data member could, in the expsin case, contain the exact number of revolutions rather than the maximum number of revolutions.
Dario Izzo
@darioizzo
May 10 2015 18:58
  • The iterations (which would find the value for k2 that meet the tof request) will be one of the key factors. Which variable should we iterate upon? k2 directly? If so, can we write efficient Householder iterations? Here the problem is that sometimes the solution does not exist (also for 0 revs, unlike Lambert's problem) how do we detect these cases efficiently?