search

astropy / coordinates-benchmark

Accuracy Benchmarks of different coordinate conversion codes. Maintainers: @astrojuanlu, @astrofrog, and @cdeil
https://www.astropy.org/coordinates-benchmark/summary.html
11 stars 12 forks source link

Add ephemerides benchmarks #14

Open cdeil opened 11 years ago

cdeil commented 11 years ago

Some day we want to have astropy support coordinate computations for solar system bodies like the sun, moon and planets (called ephemerides): https://groups.google.com/d/topic/astropy-dev/jjH0dtdByxk/discussion

When this effort starts we should add checks against other tools: http://rhodesmill.org/pyephem/quick.html#bodies http://ssd.jpl.nasa.gov/?ephemerides

eteq commented 11 years ago

+1 to this!

astrojuanlu commented 11 years ago

Prior to the benchmarks, is there an issue or PR in the core astropy repo to track progress in adding ephemerides computations?

cdeil commented 11 years ago

I'm don't know exactly what is going on with coordinates and ephemerides in astropy. I see you have started something here.

@eteq will know what the status is and if you are interested in contributing to astropy. Possibly a discussion on the astropy-dev mailing list followed by an astropy-api proposal.

astrojuanlu commented 11 years ago

Well, actually in that project you pointed out there's already some bare bones ephemerides calculations based on some numerical formulas which compute the osculating orbital elements. It worked for me because I had to perform some medium-precision computations, but I haven't compared them with, say, the JPL ephemerides. Other than that, it might be preferable to have RA/dec ephemerides for example? I have followed only one course on Orbital Mechanics so I might probably miss some important details.

FWIW, I collected some of the relevant projects and ideas in this issue.

eteq commented 11 years ago

@Juanlu001 - there have not been any explicit plans to include ephemerides in astropy yet, mainly because we're working on re-vamping coordinates for 0.3, and of course anything with ephemerides will need to make extensive use of coordinates. But it's definitely something we will want in astropy at some point hopefully in the not-too-distant future. Feel free to make an issue to do this in the main astropy repo, if you like.

(Also, if you don't know about this, @Juanlu001, note that you don't need to give full links to issues on github - if you just type Pybonacci/poliastro#1, github is clever enough to recognize the link automatically: Pybonacci/poliastro#1 and it also creates indications in the issue that it has been commented on here)

There has been some discussion of this on astropy-dev in rather vague terms - the main points to consider are that the JPL ephemerides should be downloaded using the astropy download/caching tools, the result definitely should be astropy.coordinates.<something> objects, and where relevant, the SOFA routines (soon to be ERFA) for earth/sun/moon location and the major planets should be used instead of "home-grown" approximations.

astrojuanlu commented 11 years ago

Thanks for the information and the clarification. I have not been following dev-related discussions but I see in the repo issues that you are working hard on coordinates, and on the other hand right now I am basically gathering fundamental knowledge about the matter, so as of today I think I will wait a while to open that issue.

Anyway, I will go on learning and keep an eye on this just in case I can be of any help when the time comes.

Kudos for the great work, astropy is my favorite Python project in terms of organization and code quality by far :)

(And thanks for the tip about issues links!)