Conversation
| date_time::Date::Ordinal(year, 1))); | ||
| LOG(ERROR) << "Flowing to " << t; | ||
| auto const status = ephemeris_->FlowWithFixedStep(t, *instance); | ||
| if (!status.ok()) { |
There was a problem hiding this comment.
No, we do encounter an erroneous status (the orbiter runs into the ground at the end of 1971).
| elements.argument_of_periapsis, | ||
| mathematica::ExpressIn(Radian)); | ||
| } | ||
| // The elements e, i, and ω all vary quite a lot. |
There was a problem hiding this comment.
The numbers would be easier to interpret if, instead of min and max (or in addition), you looked at the relative error between measure and midpoint. It's like 75% for e and 17% for c₁.
There was a problem hiding this comment.
I don’t think the relative errors are meaningful here.
The point (c₁, c₂) lies within a bounded region within [0, 1] × [-3/5, 2/5] (depicted in [Лид61, p. 29], [LC62, p. 745]).
Relative errors clearly aren’t relevant for the angles, and for the eccentricity the fact that the orbit goes to infinity at e=1, not e=0, should make it clear that the relative error of e is, if anything, the reverse of what we care about.
There was a problem hiding this comment.
How do you demonstrate that something is constant if you have lots of measurements with noise? By looking at its statistical elements. If you are not happy with a cheesy relative error, please do a proper statistical analysis. (We have tools to do that in testing_utilities\statistics.hpp.)
There was a problem hiding this comment.
As discussed, what we care about is the absolute scale of the variations; they are not really noise, but they are very small compared to the size of the region given in Рис. 1.
|
retest this please |
This test showcases the Лидов–古在 mechanism, and thus fixes #1119.
The relevant elements vary as follows:
The eccentricity–inclination exchange is evidenced by the following graph:
The other conserved quantity (which relates all three elements) is also constant: the argument of the periapsis governs that exchange.