NASA OS Software: The General Mission Analysis Tool (GMAT)

[ElChristou]

Tx Selden for more info.

Indeed if Celestia and further NASA/ESA project will collaborate in the future, this question is quite cruxial, no? Any way to normalize this point?

Seems SPICE and GMAT use JPL DE405... noob question, are this long term choices?

As Celestia was already questioned on the VSOP87 choice, is the use of JPL DE405 be an interesting option? (instead of VSOP2000 for example)

[t00fri]

Tx Selden for more info.

Indeed if Celestia and further NASA/ESA project will collaborate in the future, this question is quite cruxial, no? Any way to normalize this point?

Seems SPICE and GMAT use JPL DE405... noob question, are this long term choices?

As Celestia was already questioned on the VSOP87 choice, is the use of JPL DE405 be an interesting option? (instead of VSOP2000 for example)

The main issue is the moon's ephemeris anyhow. It's not very good in VSOP87, while VSOP2000 had the improvement of moon's position high up on their shopping list.

Bye Fridger

[ElChristou]

Tx Selden for more info.

Indeed if Celestia and further NASA/ESA project will collaborate in the future, this question is quite cruxial, no? Any way to normalize this point?

Seems SPICE and GMAT use JPL DE405... noob question, are this long term choices?

As Celestia was already questioned on the VSOP87 choice, is the use of JPL DE405 be an interesting option? (instead of VSOP2000 for example)

The main issue is the moon's ephemeris anyhow. It's not very good in VSOP87, while VSOP2000 had the improvement of moon's position high up on their shopping list.

Bye Fridger

Do you mean it's possible to mix the ephemeris? or your personal choice tend to VSOP2000 in replacement of the actual? (then for info (apart the moon position) why this choice?)

[t00fri]

The main issue is the moon's ephemeris anyhow. It's not very good in VSOP87, while VSOP2000 had the improvement of moon's position high up on their shopping list.

Bye Fridger

Do you mean it's possible to mix the ephemeris? or your personal choice tend to VSOP2000 in replacement of the actual? (then for info (apart the moon position) why this choice?)

Well I don't know whether this would be a good solution.

Some time ago I just went in detail through a number of respective scientific papers and conference talks that discussed those matters.

In a nutshell, the message seemed to be that the essential improvement between VSOP87 and VSOP2000 is the moon's emphemeris.

Bye Fridger

[ElChristou]

Ok, so the question now is VSOP2000 vs JPL DE405, which is better?

(Fridger, not a question for you but for who can answer)

I'm asking because if there is no arguments against JPL DE405, then the fact that SPICE is already part of Celestia, it's use seems judicious (also good for the eventual collaboration with GMAT, but also good for the end users, avoiding too much datas to download in case of professional "addons" using SPICE/GMAT)...

[t00fri]

Ok, so the question now is VSOP2000 vs JPL DE405, which is better?

(Fridger, not a question for you but for who can answer)

I'm asking because if there is no arguments against JPL DE405, then the fact that SPICE is already part of Celestia, it's use seems judicious (also good for the eventual collaboration with GMAT, but also good for the end users, avoiding too much datas to download in case of professional "addons" using SPICE/GMAT)...

++++++++++++++++++++++
From a practical, performance point of view it's not only a matter of "better"=more accurate. Celestia is a real-time simulation and thus must be FAST! So one has to critically examine & compare performance as well, which is a pretty "BIG" job .
++++++++++++++++++++++


Bye Fridger

[selden]

What do you mean by "better"?
More accurate or easier to use?

DE406 should be the most accurate, but the DE40x ephemerides use large data files, which most people will be unwilling to use just because of the size and download time. (There is about a 1 meter difference in the position of the Moon between DE405 and DE406 according to http://www.willbell.com/software/jpl.htm)

If you just need something for the duration of the Apollo 11 mission, I would expect that the DE40x ephemeris files would be relatively small.

[ElChristou]

What do you mean by "better"?
More accurate or easier to use?

DE406 should be the most accurate, but the DE40x ephemerides use large data files, which most people will be unwilling to use just because of the size and download time. (There is about a 1 meter difference in the position of the Moon between DE405 and DE406 according to http://www.willbell.com/software/jpl.htm)

If you just need something for the duration of the Apollo 11 mission, I would expect that the DE40x ephemeris files would be relatively small.

Nope Selden, the Apollo stuff is not my preoccupation here, I'm asking for a long term choice in a global optic.

By better, I mean... better... can be about accuracy, easy of use or as Fridger point out, speed... or else...

So following the above, what are the possible candidate for the eventual replacement of VSOP87?

- VSOP2000 for a better moon,
- JPL DE405 seems accurate and is used by SPICE

Another one?

Noob question: Comparing the use of ephemeris (reactance within Celestia) is a question of building using one or another? Is there already any good "How to" on the topic?

[selden]

Chris,

I don't know what Chris Laurel is considering for the ephemeris that is built into Celestia, but I do know that DE405 is much too big. It is hundreds of megabytes.

Instructions for using JPL ephemeris files with Celestia v1.5.0 are available in the WikiBook at http://en.wikibooks.org/wiki/Celestia/JPL_Ephemerides

[ElChristou]

Chris,

I don't know what Chris Laurel is considering for the ephemeris that is built into Celestia, but I do know that DE405 is much too big. It is hundreds of megabytes.

Instructions for using JPL ephemeris files with Celestia v1.5.0 are available in the WikiBook at http://en.wikibooks.org/wiki/Celestia/JPL_Ephemerides

Ok, so adios DE405!

Tx for the link.

[t00fri]

Chris,

I don't know what Chris Laurel is considering for the ephemeris that is built into Celestia, but I do know that DE405 is much too big. It is hundreds of megabytes.

Instructions for using JPL ephemeris files with Celestia v1.5.0 are available in the WikiBook at http://en.wikibooks.org/wiki/Celestia/JPL_Ephemerides

Ok, so adios DE405!

Tx for the link.

Unlike a number of the writers in this thread, I guess I REALLY understand the physics and the pros & cons of VSOP87. I also know quite a bit about JPL40x, for example.

But why should we loose our time with this NOW?

In 2-3 months Chris L. will suddenly reappear and announce his preference "Ex Cathedra" so to speak (if we still remember this discussion until then, hi hi )

A number of years ago, when Chris and I where busy testing the VSOP87 accuracy, I really went deep into these issues...

Bye Fridger

[ElChristou]

Unlike a number of the writers in this thread...

right!

SO VSOP87 is still today the best choice for Celestia...

Couldn't you said that earlier?

(Indeed from the several questioning of VSOP87 in the past, I thought a change had to be made...)

[steven hughes]

I tracked down enough information to get a pretty good first cut of the outbound Apollo 11 trajectory modeled in GMAT. The picture below illustrates the free return trajectory. I haven't modeled the maneuvers at the moon, so you see the trajectory return to Earth naturally. This was a design feature in case they needed to abort the mission.

Darrel is going to take the trajectory data and create an xyz file for Celestia. Then we'll see if there is an Ephemeris issue.

[t00fri]

Unlike a number of the writers in this thread...

right!

SO VSOP87 is still today the best choice for Celestia...

Chris, that was NOT implied in my post, since there are a number of quite different constraints to observe.

To repeat: accuracy for the planets, accuracy for the moon (which is a special issue!) , easyness in handling AND speed considerations. And more recently, there is the fact that SPICE and GMAT do NOT use VSOP87.

Couldn't you said that earlier?

NO, see above.

Bye Fridger

[ElChristou]

Unlike a number of the writers in this thread...

right!

SO VSOP87 is still today the best choice for Celestia...

Chris, that was NOT implied in my post, since there are a number of quite different constraints to observe.

To repeat: accuracy for the planets, accuracy for the moon (which is a special issue!) , easyness in handling AND speed considerations. And more recently, there is the fact that SPICE and GMAT do NOT use VSOP87.

Couldn't you said that earlier?

NO, see above.

Bye Fridger

Humm.. For once I don't follow you well

You said above that you have deeply study the topic a few years ago, resulting in the choice of VSOP87 (taking in account all possible factors).

You said that it will be a lost of time to enter again in the topic, so I conclude the VSOP87 is still ok to the date! (again, taking in account all possible factors)

And now you seems to say that... nope...

I'm a bit lost...

[t00fri]

Humm.. For once I don't follow you well

You said above that you have deeply study the topic a few years ago, resulting in the choice of VSOP87 (taking in account all possible factors).

You said that it will be a lost of time to enter again in the topic, so I conclude the VSOP87 is still ok to the date! (again, taking in account all possible factors)

And now you seems to say that... nope...

I'm a bit lost...

Sorry Chris,

a TOTAL misunderstanding!

VSOP87 was decided about and implemented by Chris L. very early in time (~5 years ago?). There were still a few bad bugs remaining, one of the worst I have spotted. After that initial period, we both spent lots of time and research into checking and improving things. During this stage and during my long previous association with the XEphem project (Elwood Downey), I just dug deeply into the basics of VSOP87. And after all, I am an experienced theoretical physicist .

Knowing Chris L. since 5.5 years . I just cynically stated that whatever we are chatting about VSOP87 NOW, will have little effect on Chris L's decision, once he will reappear. That's the way things are.

So why loosing time...?

But meanwhile also new points of view have emerged: SPICE and GMAT for example. Both NOT using VSOP87. That's why the matters should be critically reconsidered.

So altogether at this point, there is no simple YES/NO statement possible.

That's all.

Bye Fridger

[ElChristou]

...Knowing Chris L. since 5.5 years . I just cynically stated that whatever we are chatting about VSOP87 NOW, will have little effect on Chris L's decision, once he will reappear. That's the way things are...

That, I understood from your previous post, ...bad memories!

Really it's too bad nothing can even be discussed meanwhile Chris is out...

[djcinsb]

...
To repeat: accuracy for the planets, accuracy for the moon (which is a special issue!) , easyness in handling AND speed considerations. And more recently, there is the fact that SPICE and GMAT do NOT use VSOP87.
...
Bye Fridger

Do you remember roughly how different the VSOP87 and VSOP2000 were for moon position, and how much they differ from the DE40x files? For typical GMAT folks, I think the view will usually include most or all of the body being orbited, so for visualization an error of a few kilometers would not matter (as long as the user knows it is there) -- even if the Earth fills my 1900 pixel across screen, a single pixel is over 6 km wide. A difference of tens or hundreds of kilometers would be a different matter. (Of course, precision in the numerics used to control the spacecraft is a different issue!)

Meanwhile, here's a tweaked version on Steve's Apollo free return mission, in Celestia:



and a different view as the return to Earth starts:

[ElChristou]

...Meanwhile, here's a tweaked version on Steve's Apollo free return mission, in Celestia:

...

and a different view as the return to Earth starts:

...

What about those hard edges in the trajectory? not enough data?

[djcinsb]

...Meanwhile, here's a tweaked version on Steve's Apollo free return mission, in Celestia:

...

and a different view as the return to Earth starts:

...

What about those hard edges in the trajectory? not enough data?

Yes. During the transfer orbit, the data points are about 45 minutes apart. We can set things to generate more data; but for now we're still tuning some things up for this trajectory. The default setting on GMAT's propagators is to let the integrator adapt step, and to use the largest step it can take while still achieving a specified tolerance. That leads to quicker runs (:D), but does produce more jagged trajectories(:cry:). For this case, the propagator was set to achieve accuracy of 1 part in 1e11 per step, which allows a fairly large step when the forces are not changing much -- like between the Earth and Moon.

Since GMAT is typically used to search for solutions for targeting or optimization problems, it can run LOTS of trajectories, so the speed of the run is pretty important. I tend to make the tolerance a bit tighter (being paranoid about getting inaccurate results), but here we're just running with a default settings. We can also set a smaller maximum step to get a smoother trajectory, of course.

Once I have a few tools in place to automate the data processing I'm currently doing by hand -- here I'm talking about the epoch conversion from GMAT's modified Julian format into full Julian -- I'll make a few runs that refine things better. (Yeah, I could hack something together to do the conversion quickly, but I want to use this as an opportunity to write a small GMAT programming tutorial.)