juliaspacemissiondesign/ephemerides.jl

A Modern Binary Ephemeris Reader for Julia, in Julia.

ephemerides.jl's Issues

Small loss of accuracy in Chebyshev interpolation

Starting from the highest Chebyshev coefficients and moving to the lowest improves the accuracy because adding from biggest to smallest floating-point numbers is less accurate than adding from smallest to biggest?. Implementing it however, would require an extra for-loop on every interpolation, is it worth the extra time?

See the MWE in issue #10.

Improve precompilation times

Implement package precompilation calls.

Loss of precision in time argument

When computing the MOON_PA orientation, a loss of precision in the time arguments arises when converting from TDB seconds since J2000 to TDB days since J2000.

CALCEPH and SPICE instead have exact results. MWE:

using Ephemerides 
using CalcephEphemeris
using JSMDInterfaces.Ephemeris
using Tempo 
using SPICE
using ReferenceFrameRotations

kernel =  "test/assets/moon_pa_de440_200625.bpc"

ephj = EphemerisProvider(kernel);
ephc = CalcephProvider(kernel);

et = rand(0.0:1e8)
et = 1.0

yj, yc = zeros(3), zeros(3)
ephem_orient!(yj, ephj, DJ2000, et/86400.0, 31008, 1, 0)
ephem_orient!(yc, ephc, DJ2000, et/86400.0, 31008, 1, 0)

yj2 = ephem_rotation3(ephj, 1, 31008, et)

A = DCM(pxform("J2000", "MOON_PA", et))
angs = dcm_to_angle(A, :ZXZ)

yj = mod.(yj, π)
yj2 = mod.(yj2, π)
yc = mod.(ys, π)
ys = mod.([angs.a1, angs.a2, angs.a3], π)

yj-yc
yj2-yc 

yj-ys
yj2-ys

yc-ys

# yj2 is more precise than yj!
yj2 - ys 
yj2 - yj

Time handling as two floating 64-bits variables

Evaluate whether implementing time as two floating-point 64-bits variables is worth and quantify the accuracy gain.

Use StaticArraysCore instead of StaticArrays

Missing docs in SPK Type 17 Header

Complete the description of the SPKSegmentHeader17 fields.

Update docstrings for SPK segment caches

The implementation of a public interface to retrieve the segment data (see #26) will require the user to know how that data is organized inside the Julia structures. Although the description of the segment caches will still be a low-level API, all the caches should be documented nonetheless.

TagBot trigger issue

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

If you'd like for me to do this for you, comment TagBot fix on this issue.
I'll open a PR within a few hours, please be patient!

Optimise record coefficients look-up time

For segments with unequal time intervals that have epoch directories, check whether it is worth to store the last loaded segment time boundaries, so that if they match the user requested time, you do not have to read through all the epochs to find the desired segment index.

Add support for SPK type 15

Remove comment warning during package loading

A warning is thrown each time the package is loaded because one of the kernels in the artifacts does not have the EOT in the comment. Should temporarily disable the logging during precompilation to avoid showing that each time.

Fix missing EOT bytes in Test Artifacts

Some of the kernels in the artifacts used in the tests (but not in precompilation) are missing the EOT byte in the SPK comment.

Fix segment times with overlapping boundaries

The get_segment_boundaries function doesn't merge the times when the start time of a descriptor equals the end time of another descriptor. Indeed, calling:

desclist = [
    Ephemerides.DAFSegmentDescriptor(2, 1.0, 1.6, 31008, 1, -1, 1, 1)
    Ephemerides.DAFSegmentDescriptor(2, 1.6, 2.0, 31008, 1, -1, 1, 1)
]

ts, te = Ephemerides.get_segment_boundaries(desclist)

Returns:

ts = [1.0, 1.6]
te = [1.6, 2.0]

To-Do:

• Fix aformentioned bug
• Add tests on these routines

Add ability to generate binary kernels

Add the possibility to create and manipulate binary SPK/PCK ephemeris kernels.

Add support for SPK type 5

Segment Cache Public Access and Handling

When segment caches are retrieved in the spk_vector functions, a copy of the original structure is made, such that when the segment cache is later inspected, it appears empty.

High-level interface functions to access segment data (i.e., the polynomial coefficients) should also be implemented.

Add tests for hyperbolic orbits on SPK type 5

Re-order segment priorities to speed up transform if-elses

Add support for SPK type 17

Add support for SPK type 14

Implement support for SPK type 14 segments, that handle unequally-spaced Chebyshev polynomials. For tests, use the binary kernel available in CALCEPH test directory.

Add support for INPOP v2.0 ephemeris files

Add support for INPOP v2.0 binary files, whose format is described here.

Assert SPK 15 test tolerances

Relative test tolerances on SPK Type 15 have been lowered to 1e-11 because they yield bigger differences for propagation times > ~3000s. A check is required to establish whether this difference is only due to numerical issues or because of some bugs in the algorithms