Digital Universe

[selden]

I'm not entirely sure if "loyal follower" is the term I'd use

My attitude is more "given this tool, how can I use it to portray an event that interests me?" For quite a while the events interesting to me have involved, among other things, spacecraft orbits and orientations. (cf. the "Across to Mars" addon.) As a result, my familiarity with Celestia usually makes it my visualization tool of choice. Use of Partiview, for example, would require me to write the software to do all of the orbital calculations. That's beyond my capabilities, unfortunately.

I have to admit that the complexity of the new Frame constructs is bothering me. Some of the necessary interactions among them are less than obvious to someone unfamiliar with them.

I think it is very important that the simplicity of the catalog formats not be compromised. It would be very unfortunate if Celestia's Addon interface were to develop into something unusably complex. Sadly, the Frame constructs seem to be headed in that direction, although the previous, less functional, commands are still available.

This concern resulted in my post to the developers' list which suggested that a macro preprocessor might be appropriate. Catalog authors could define and use libraries that define simplified declarations that expand into appropriate combinations of Frame constructs. They wouldn't have to wait for simplified definitions to be supported directly by Celestia.

Although they're readily available, I don't think the cpp or m4 macro preprocessors can meet the need. Floating point and transcendental function evaluation would seem to be necessary, e.g. to be able to create a LongLat definition that makes use of Frames. Translating from polar to cartesian coordinates would need those functions. Perhaps a preprocessor written in Lua could meet the need, but one hasn't been written yet.

[t00fri]

I'm not entirely sure if "loyal follower" is the term I'd use
...

Sorry, Selden,

your are perfectly right. I was implicitly referring to only one aspect of that statement...

Certainly, I was also aware of your repeated syntax critique that I fully share...I am just not willing to reiterate the whole frame construction stuff from the start, once Chris has completed already lots of respective code without previous discussions with experts about frames and their mutual maps...


Bye Fridger

[chris]

....
I think that SPICE support is extremely important. I'm interested in seeing Celestia adopted within the national and international space agencies, and SPICE support, reference frames, and flexible trajectory and rotation models are all essential features. I've had many requests for these things, and I happen to think that they fit better into Celestia than the particle cascade visualization. As neat as that animation is, why would it be integrated into Celestia than as a standalone application? Does it really gain anything by being part of a universe model?

...
--Chris

Chris,

one the one hand, I can very well see some of your driving motives that may justify this high amount of specialized activity towards "SPICE + frames" stuff, with particular emphasis on what space agencies are interested in...

As you may have noted, I have kept quiet since a while as to my personal opinions and preferences. But since you phrase things explicitly now, let me make some respective comments, nevertheless.

i) I think I was among the first in our community emphasizing again and again that my view about Celestia's purpose would be to make it a general visualization framework of scientific standards, rather than a specialized tool for e.g. education, world creators or gamers etc.

Hence a priori, an implementation of a SPICE interface had my full sympathies.

ii) However, as a theoretical physicist with much expertise in Astroparticle Physics, Cosmology and General Relativity (besides Astromechanics and all that, of course), I am increasingly sad to see where Celestia seems to be moving to, recently.

+++++++++++++++++++++++
From my perspective, there would be many other exciting areas related to a scientific visualization of the Universe, where Celestia could make a unique and conceptually pioneering contribution!
+++++++++++++++++++++++

As a physicist, I find this recent specialized work on custom spacecraft trajectories, rotating (spacecraft) frames etc largely "uninspiring", to say the least. From a physics point of view, these things are crystal clear to me and do not represent the slightest conceptual challenge (besides the considerable, respective coding efforts, of course!). On the other hand, I also can hardly see many people exploit these new degrees of freedom EXCEPT personel employed by space agencies...

I see much of these recent efforts as specialized yet FREE service work for space agencies, who always welcome oportunities to save money . We all have to get along professionally with limited budgets...

iii) Except for Selden as a loyal follower of your efforts, I can also see little enthusiasm & resonance in the rest of our community (including all other devs) for these new implementations, notably given the high price of a much complexified syntax etc.

Well, new syntax is even necessary in most cases. Everything that can be done with Celestia 1.4.1 ssc files can still be done in 1.5.0 without using reference frames. But I'd like to hear what the alternative is to the current implementation of frames. I'm sure there are some better ideas . . . if you have them, please share! I want to be able to place objects in body-fixed references frames, specify LVLH attitudes, use coordinate systems base on the relative positions of two bodies. What's the best way to do this? Or are suggesting that these things aren't necessary at all? Or that they would be useful, but that my coding efforts would be better directed elsewhere?

Since you normally do things first before asking around, I thought I had to express my views at that advanced state of affairs at least once. I am also confident that you will not be upset by my honest attempts of expressing some respective criticism.

No, I won't be upset. But I may raise a lot of counterarguments

--Chris

[Malenfant]

Well, I certainly have no interest in and see no point in being able to show particle showers in Celestia...

i) I think I was among the first in our community emphasizing again and again that my view about Celestia's purpose would be to make it a general visualization framework of scientific standards, rather than a specialized tool for e.g. education, world creators or gamers etc.

Funny, I could have sworn you were the one who was insisting all the time that it shouldn't be a scientific visualisation and educational tool. We've certainly had several arguments over this. I've always seen Celestia as being capable of doing either of these, and I'm confused as to why you think it would require "specialisation" to make it an educational tool. These are not mutually exclusive goals - not even remotely. But still, at the end of the day, Celestia can only ever be a visualisation tool at most, because you can't do original science with it.

As a physicist, I find this recent specialized work on custom spacecraft trajectories, rotating (spacecraft) frames etc largely "uninspiring", to say the least. From a physics point of view, these things are crystal clear to me and do not represent the slightest conceptual challenge (besides the considerable, respective coding efforts, of course!). On the other hand, I also can hardly see many people exploit these new degrees of freedom EXCEPT personel employed by space agencies...

I see much of these recent efforts as specialized yet FREE service work for space agencies, who always welcome oportunities to save money Wink . We all have to get along professionally with limited budgets...

You do realise that Celestia's purpose isn't to provide a "conceptual challenge" for you exclusively, right? You may not be interested in this stuff, but Chris is, and so are others. That alone is reason enough to continue with it.

And you also seem to be contradicting yourself - you want scientists to be able to use Celestia for visualisation, then you seem to be complaining that only "personnel employed by space agencies" will be using these features. So do you have something against spacecraft trajectory planners now? None of this is getting in the way of any 'pure astrophysics' visualisation that I'm presuming you're more interested in. And it seems that you're complaining about the educational uses of Celestia distracting from the scientific visualisation uses, while expecting the educational side to be toned down in favour of the scientific visualisation. Again, there's no need for "one or the other" - it can do both. Vincent's add-ons are a great boon for the educational side, and they've not detracted in any way from any scientific visualisation work.

Again though, the solution to all this is to work on what you want to see in Celestia. If that is what it takes to get you interested in it again then all the better.

[t00fri]

Chris,

But I'd like to hear what the alternative is to the current implementation of frames. I'm sure there are some better ideas . . . if you have them, please share! I want to be able to place objects in body-fixed references frames, specify LVLH attitudes, use coordinate systems base on the relative positions of two bodies. What's the best way to do this? Or are suggesting that these things aren't necessary at all? Or that they would be useful, but that my coding efforts would be better directed elsewhere?

Firstly, from my purely personal (!) preferences only, I would indeed prefer if you'd direct your coding efforts elsewhere. There are so many challenging issues related to 3D graphics that could experience a spectacular improvement given your unique know-how! Not to mention my old list of new challenges related to cosmology displays in Celestia...

Of course I also see your motivation for embarking into frames, rotations, SPICE and all that mundane stuff at the present time...

I also admit that frames deserve to be implemented systematically. Actually I think I even was the first of us -- already back in 2002-- emphasizing the need for systematic frames in our dev-"shopping list" . Yet, I was and I still am more interested in a flexible set of respective grids (with auto-adaptation) and cursor readouts for different standard frames etc.

Given the large amount of specific professional knowledge I have in the relevant field, I am obviously not the right kind of person who would enjoy to essentially "polish" your present frame implementation, after you have set up all the structures already at the code level. I think it's simply too late for significant contributions from my side. At this stage there is too much bias already.

Although I would also need to invest more research into this topic, let me nevertheless indicate some particular lines of attack that I find interesting/promising and hence would enjoy pursuing further.

Apparently, you have smoothly connected your incorporation of frames to the way they are handled in SPICE. That's fine, but all depends on the SCOPE one envisages for such a task. NASA and other space agencies have a quite narrow scope for frames, mostly related to an efficient description of spacecraft.

In my view this is among the least interesting possible applications of frames in Celestia. Again, I admit, this is personal. I would always set out to formulate frames to be ready for general relativity, gravitational lensing, red-shifting of galaxies etc, as required for astrophysical/cosmological visualizations that I find most exciting. Completely new aspects become relevant in this case.

+++++++++++++++++++++++++++++++++++++++++
One aspect that is entirely missing yet in your frame implementations is a discussion about the choice of optimal coordinates in the various frames.
+++++++++++++++++++++++++++++++++++++++++

You may know that this is a vast theoretical field, with many applications notably in the 3-body problem as applied to planetary orbiting. Let me briefly touch upon a few crucial issues in this context, hoping that you know what I am talking about.

I am referring to so-called action-angle variables.
These are ingenious coordinates, really, for our purposes.

Action-angle variables are the coordinates of choice for Hamiltonians without explicit time-dependence and notably for periodic or quasi-periodic systems! From standard (e.g. cartesian) coordinates, one may arrive at action-angle coordinates by exploting co-called canonical transformations. The trick is to make use of the constants of motion of the system ("actions")! The crucial property of the new set of conjugate phase-space variables (Q_i, P_i) is that

Code: Select all

Q_i = a_i * t + b_i  (generalized angles)
P_i = const!  (constants of motion = "actions") 


++++++++++++++++++
The constant conjugate momenta P_i may effectively be used for octree culling of such periodically moving objects and probably also to define 2-vector frames (Q,P). Moreover, these variables are great for doing perturbations around periodic orbits etc.
++++++++++++++++++

In case you are interested, there are many explicit applications/examples, of interest for us, like e.g. the "planar circular restricted 3-body problem (PCR3BP)" (Sun-Jupiter-planet) etc. I can even point you to some respective examples in the net.

So far for this.

Another issue concerns your syntax for frames. Here I am not ready yet to suggest /concrete/ improvements, but I am confident that I would be able to, after some further thinking. But it all means that I first should have a desire to "catch on" to that project

Bye Fridger

[t00fri]

Here is a very simple Caltech lecture (for people with a grasp of theoretical Hamiltonian mechanics), introducing action-angle variables and referring also to some standard planetary orbiting example:

http://www.its.caltech.edu/~mcc/Chaos_C ... tonian.pdf


Bye Fridger

[Malenfant]

Something that developers need to ask themselves is whether it's worth adding things like gravitational lensing and redshifting to Celestia in the first place. These sound like very complex things to code and to visualise and I'm wondering how much of the userbase is actually going to use it and care about it (then again I wonder how much of the userbase is going to use/care about the reference frame stuff too), and it might just not be worth the time and effort to do it.

Personally I'd rather see existing code refined and made more efficient (like the asteroid orbit rendering, octree culling, barycentres, atmospheres etc) than any more totally new code added. I think the frames are quite enough for now.

[Cham]

I agree with Fridger and Malenfant. However, I think that the general relativist thing would be great. And there are much more stuff (IMHO) which are more important than the reference frame stuff.

[Malenfant]

I guess my point is this - that there's all sorts of things out there that we could use Celestia to visualise, but the question is whether we should do that in the code itself. There's redshifting, relativistic effects, time dilation, gravitational lensing... but there's also accretion disks, realistic roche lobes, stellar flareing, interstellar dust, and so on. Where do we draw the line?

I think it's better to just focus on specific things and do them well, rather than try to render everything the universe can show us but do it poorly because efforts to do so are spread too thin.

(personally, I'd like to see stars beyond the 16k lightyear limit somehow. That's a refinement that's long overdue IMO).

[t00fri]

I guess my point is this - that there's all sorts of things out there that we could use Celestia to visualise, but the question is whether we should do that in the code itself. There's redshifting, relativistic effects, time dilation, gravitational lensing... but there's also accretion disks, realistic roche lobes, stellar flareing, interstellar dust, and so on. Where do we draw the line?
...

Hey guys,

whichever we pick from this list of fascinating phenomena, to me these are at the heart of astrophysics/cosmology and Celestia could make a unique "show" in visualizing them! In any case this appears to me more exciting than perfectionating the suspension of NASA spacecraft in the sky

Bye Fridger

[Cham]

whichever we pick from this list of fascinating phenomena, to me these are at the heart of astrophysics/cosmology and Celestia could make a unique "show" in visualizing them! In any case this appears to me more exciting than perfectionating the suspension of NASA spacecraft in the sky

I totally agree.

[Malenfant]

whichever we pick from this list of fascinating phenomena, to me these are at the heart of astrophysics/cosmology and Celestia could make a unique "show" in visualizing them! In any case this appears to me more exciting than perfectionating the suspension of NASA spacecraft in the sky

Bye Fridger

Sure, but you're rather deep in the field... what an expert finds "exciting" may not necessarily be that interesting to many others. And experts are often bored by what others think is exciting. Then again, I guess Chris is excited by things that push the graphic cards to the limit, or by things that help Celestia get used more in space agencies, or whatever. Everyone's different.

The good thing is that people can always develop the code in their own ways for their own versions of Celestia, but what should go into the 'official' version is up to Chris, and presumably he has to weigh up the utility of the code along with the ease at which it can be integrated.

Also, consider this - what would you actually use redshifting and lensing for? The Celestia camera can travel faster than light for one thing, and to get anywhere in the universe fast you'd have to travel at several multiples of c. Any redshifting of the view would take place only in the acceleration - after that, you may as well just show a Star Wars 'linear stars' hyperspace effect . And when are you really ever going to use gravitational lensing? Only when you're close enough to a black hole (or neutron star?), or if a galaxy happens to be in front of another one. While they'd be nice things to show, it seems to me to be an awful lot of effort to generate things that are hardly seen.

[ElChristou]

Something that developers need to ask themselves is whether it's worth adding things like gravitational lensing and redshifting to Celestia in the first place. These sound like very complex things to code and to visualise and I'm wondering how much of the userbase is actually going to use it and care about it (then again I wonder how much of the userbase is going to use/care about the reference frame stuff too), and it might just not be worth the time and effort to do it.

If Celestia is suppose to be the best visualization tool, all actual knowledge IMO should be incorporated; the use or not by the userbase of one or other feature is not really relevant. As example, if it's possible to visualize gravitational lensing via OpenGL, it should be implemented because actually it's a very important effect used in many case by astronomers.
Without talking of the educational value of the feature, we will simply descrive a natural effect and only this reason IMO is enough to implement it.

Personally I'd rather see existing code refined and made more efficient (like the asteroid orbit rendering, octree culling, barycentres, atmospheres etc) than any more totally new code added. I think the frames are quite enough for now.

I agree on this point, it would be nice for example to have 70% of work on new features, the rest for polishing and solve old problems...

[Cham]

I agree on this point, it would be nice for example to have 70% of work on new features, the rest for polishing and solve old problems...

I'll reverse that percentage : 70% of work on polishing and old bugs, and 30% on new features.

[t00fri]

Sure, but you're rather deep in the field... what an expert finds "exciting" may not necessarily be that interesting to many others. And experts are often bored by what others think is exciting.

But we also get confronted first hand with what the media are after. Don't forget that --being at a big international lab-- I know about their keen interest to invite us experts for TV shows etc about precisely those topics. They (presently) seem to induce lots of fascination also among the non-expert population..

Open any issue of Sky&Telescope and you'll see where the popular interest goes. When did you find last time an article about spacecraft orbits in S&T ?

Then again, I guess Chris is excited by things that push the graphic cards to the limit,...

So am I and I can guarantee respective challenges ...

Bye Fridger

[selden]

There are many galaxy clusters known (seen) to be producing lensing effects. It's not rare. The cluster that shows direct evidence for the gravitational lensing caused by "dark matter" is one well publicised example.

Deducing the masses of the various clusters and galaxies for Celestia where lensing has not yet been measured may be difficult.

And don't forget that some of the exo-planet searches use it to find candidate systems in our own galaxy: foreground stars lensing planets passing in front of background stars. Of course, being able to visualize that in Celestia would really be a challenge!

[t00fri]

I think in the Celestia dev-related community we should split our efforts twofold:

i) a "task force" of people mainly involved with R&D concerning "new horizons" for Celestia. I would be most interested to be part of that group!

ii) people who are mainly concerned with perfectionating the features that are already incorporated in Celestia.

Activity (i) clearly needs intensive discussion and real research before the coding can go to "production". It's a completely different activity from (ii). Both are very important. (i) not the least to warrant the long-time attractivity of Celestia!

Bye Fridger

[Malenfant]

If Celestia is suppose to be the best visualization tool, all actual knowledge IMO should be incorporated; the use or not by the userbase of one or other feature is not really relevant.

But this simply isn't practical.

Let's be realistic people, we can't possibly include EVERYTHING known about the universe. The glaring omission so far is that mass-based phenomena can't be included realistically for example, because mass isn't incorporated into Celestia and Chris has said that it can't be because then Celestia has to do a hell of a lot more calculations which would slow it down to a halt. Thus gravitational effects (including lensing), tidal evolution, roche lobes, etc simply aren't possible in Celestia. Heck, you can't even get Celestia to calulcate an orbital period on the fly because of this. So we HAVE to cull things somewhat. And unless Chris is planning to add support for Mass in Celestia then gravitational lensing and other effects won't be possible to show realistically anyway.

Adding star and galaxy catalogues is one thing, but being able to simulate every physical phenomenon in the universe in a single program is just ridiculous IMO.

[ElChristou]

...but being able to simulate every physical phenomenon in the universe in a single program is just ridiculous IMO...

True, but you are using the right word, simulate, not reproducing.
The problem is to find some tricks to "show" the effect without of course using all the reals equations... Only some brainstorming among the dev can tell us it's possible or not...