Feature request: star motions

[selden]

It's be really nice if Celestia could incorporate the proper motions of stars. For example, one would be able to find various stellar associations by watching them move among the others. Not to mention being able to watch the constellations change with time.

[MKruer]

Ouch! You know what you are asking for? Super computers clusters would have a hard time running Celestia. Yes it is possable but not realy feaisable at the moment. If trying to get the motion of planet (hyper accurate) brings even the fastest coptuters to a craw. With that information now your asking for between 100,000 to 2.5 million star motions to be included.

Let me know if im wrong Chris

I would put this under future objectives but dont be suprised if you dont see it for some time.

[Guest]

Well, remember that the calculations would only have to be done for the visible stars, and could be an option. Celestia already has at least two provisions for reducing the number of stars that are visible. And don't forget that today's desktop systems would have been considered supercomputers just a few years ago. Multiprocessor multi-gigahertz systems are almost affordable even now, so a multi-threaded solution might be appropriate.

I had been thinking about simple linear projections because the standard version of Celestia does not include any gravitational effects. Now that I think about it a little more, however, it seems reasonable to incorporate Keplerian orbits, either guestimated around the galaxy's center or explicitly defined in some cases.

[selden]

grumble. I keep forgetting that opening a new browser window and going to a link doesn't auto-login. That was me responding.

[billybob884]

well, then maybe could the binary stars at least orbit each other?

[Guest]

Well, remember that the calculations would only have to be done for the visible stars, and could be an option. Celestia already has at least two provisions for reducing the number of stars that are visible. And don't forget that today's desktop systems would have been considered supercomputers just a few years ago. Multiprocessor multi-gigahertz systems are almost affordable even now, so a multi-threaded solution might be appropriate.

I had been thinking about simple linear projections because the standard version of Celestia does not include any gravitational effects. Now that I think about it a little more, however, it seems reasonable to incorporate Keplerian orbits, either guestimated around the galaxy's center or explicitly defined in some cases.

Galactic orbits are not very Keplerian. Stars follow oscillating paths in the "vertical" direction normal to the galactic plane. Projecting Stellar motions over time using a linear model (constant 3D velocity) is perfectly possible even on single-processor (original) pentium-class machines. Such a linear model is no worse than a Keplerian model, since neither model takes into account the vertical oscillations that would be the most important error terms.

[selden]

Thanks for bringing up the topic. It's someting I've been wondering about.

It was my impression that the cyclic motion of stars perpendicular to the galactic plane was inferred from the ccncentration of mass in that plane.

My understanding is that the current theories on the distribution of dark matter suggest that there is no such planar concentration of mass, but rather that the galactic mass distribution is spherically symmetric.

(See, for example, the discussion at http://cosmology.berkeley.edu/Education/galrotcurve.html. Also, http://astro.esa.int/SA-general/Projects/GAIA_files/LATEX2HTML/node46.html mentions that a study using our favorite star catalog* indicates that most of the mass does have to be in the halo. There are also arguments that they might have chosen the wrong type of stars to do the study.)

From this I would conclude that there is no restoring force to pull stars back down into the "plane of the galaxy" since that plane doesn't really exist (at least not in the mass distribution). It's just an observational effect caused by the current distibution of bright stars.

It seems to me that "dark matter" certainly has to mess up a lot of the older theories about galactic dynamics. More precise astrometrics will help in this. Unfortunately NASA's FAME satellite lost funding. Hopefully ESA's GAIA will fly.

----
* - Hipparcos, of course

[erostosthenes]

star motions aren't just difficult to simulate, they're impossible with our current understanding of physics. if you take two gravitating bodies and ask for their motions, any physics 101 student can do it. take 3 gravitating bodies, and not even the fastest supercomputer ever built can do it (with any accuracy). the person who comes up with a diffeq for a 3 body problem will most certainly win a nobel prize.

[selden]

Which, of course, is why Celestia uses simple Keplerian orbits for just about everything that moves, and why astrophysicists make all sorts of simplifying assumptions when numerically simulating galactic dynamics.

Hey, I just wanna see the open clusters flow through the local stellar neighborhood, gradually dispersing. And watch Sol go off to meet Vega...

[HankR]

"If you take two gravitating bodies and ask for their motions, any physics 101 student can do it. take 3 gravitating bodies, and not even the fastest supercomputer ever built can do it (with any accuracy)."

Actually, it's quite possible to simulate 3-body motion with sufficient accuracy to, say, soft land a spacecraft on an asteroid...

- Hank

[selden]

And the people simulating galaxcies usually model systems of 100,000 "particles" or so. It's when you start modelling clusters of galaxies that things start to get reeely slow...

[Guest]

Thanks for bringing up the topic. It's someting I've been wondering about.

It was my impression that the cyclic motion of stars perpendicular to the galactic plane was inferred from the ccncentration of mass in that plane.

My understanding is that the current theories on the distribution of dark matter suggest that there is no such planar concentration of mass, but rather that the galactic mass distribution is spherically symmetric.

(See, for example, the discussion at http://cosmology.berkeley.edu/Education/galrotcurve.html. Also, http://astro.esa.int/SA-general/Projects/GAIA_files/LATEX2HTML/node46.html mentions that a study using our favorite star catalog* indicates that most of the mass does have to be in the halo. There are also arguments that they might have chosen the wrong type of stars to do the study.)

From this I would conclude that there is no restoring force to pull stars back down into the "plane of the galaxy" since that plane doesn't really exist (at least not in the mass distribution). It's just an observational effect caused by the current distibution of bright stars.

It seems to me that "dark matter" certainly has to mess up a lot of the older theories about galactic dynamics. More precise astrometrics will help in this. Unfortunately NASA's FAME satellite lost funding. Hopefully ESA's GAIA will fly.

----
* - Hipparcos, of course

There is a planar concentration of mass. It does not matter that it is embedded in a spherical mass distribution (the dark matter halo); the fact that there is a planar overdensity compared to the halo does matter. The galactic plane is not an artefact of bright stars; look at

http://antwrp.gsfc.nasa.gov/htmltest/jbonnell/www/multiw_sky.html

and you can see that the galactic plane exists in gas, dust, and stars. Pulsars outline the galactic plane nicely, as well.

The popular models of galactic structure at the moment are disk + bulge + halo models.

[selden]

Unfortunately, dark matter is not observable by traditional electromagnetic detection methods. As its title says, the images on that NASA page all are the results of various EM surveys, from gamma-ray to radio.

At the moment, the presence of dark matter can only be inferred from its gravitational effects. The first clue to its existance was that stellar orbital speeds do not fall off as 1/r^2 from the galactic center. (See http://www.pas.rochester.edu/~dmw/ast142/Lectures/Lect_16b.pdf and http://casswww.ucsd.edu/public/tutorial/DM.html for two discussions of it. http://www.astro.queensu.ca/~dursi/dm-tutorial/rot-vel.html has a cute interactive demo.)

Gravitational lensing effects are being used to probe dark matter distribution in various ways. (e.g. see http://astro.ic.ac.uk/Research/Extragal/ )

Rotational velocity surveys of galaxies seem to indicate that how much dark matter there is and how it is distributed varies somewhat from one galaxy to another. That's one reason why the planned high precision astrometric satellites are important. Precision measurements of the velocities of many stars are needed to get an idea of how dark matter is distriburted in our own galaxy, but mostly in our local neighborhood.

I'd forgotten that there already have been surveys of the velocities of stars above and below our own galactic plane which have been used to determine the period of the vertical motion. The period of the observed vertical oscillation (~30 MegaYears) seems to be shorter than originally expected. Apparently the sun has a vertical period of about 33MY, and a galactic revolution period of about 250MY: http://www.sigmaxi.org/amsci/articles/00articles/frischintro.html (Although I haven't found a direct reference to the original research.)

While the total mass density of our own galaxy's plane may be as much as 50% higher than is indicated by the observed gas, dust and stars, this result is rather controvercial. (See http://nedwww.ipac.caltech.edu/level5/Ashman2/Ashman2.html) This means that most of the dark matter affecting the galaxy's rotaional profile has to be in the halo.

Well, that's enough for now. I didn't intend to make this into a term paper

[Guest]

Unfortunately, dark matter is not observable by traditional electromagnetic detection methods. As its title says, the images on that NASA page all are the results of various EM surveys, from gamma-ray to radio.

You claimed that the galactic plane was an artefact of bright stars positions (although why bright stars should appear to lie in a plane if they're spherically distributed, I don't know), hence I provided that link. Gas, dust and stars are concentrated in a plane.

At the moment, the presence of dark matter can only be inferred from its gravitational effects. The first clue to its existance was that stellar orbital speeds do not fall off as 1/r^2 from the galactic center. (See http://www.pas.rochester.edu/~dmw/ast142/Lectures/Lect_16b.pdf and http://casswww.ucsd.edu/public/tutorial/DM.html for two discussions of it. http://www.astro.queensu.ca/~dursi/dm-tutorial/rot-vel.html has a cute interactive demo.)

Gravitational lensing effects are being used to probe dark matter distribution in various ways. (e.g. see http://astro.ic.ac.uk/Research/Extragal/ )

I'm aware of gravitational lensing and dark matter distributions. I don't see how any of this affects my point, which is that in the region of the galactic plane, you have:

total mass = spherical halo + plane

and everywhere else you have

total mass = spherical halo

The galactic plane is therefore an overdense region of the galaxy. Stars will oscillate about it because it's overdense. I'm not saying that the halo has no effect; that's obviously not true. I am saying that the galactic plane does have an effect on stellar trajectories, an important effect in most cases. In modelling stellar trajectories, this effect would have to be taken into account.

I'd forgotten that there already have been surveys of the velocities of stars above and below our own galactic plane which have been used to determine the period of the vertical motion. The period of the observed vertical oscillation (~30 MegaYears) seems to be shorter than originally expected. Apparently the sun has a vertical period of about 33MY, and a galactic revolution period of about 250MY: http://www.sigmaxi.org/amsci/articles/00articles/frischintro.html (Although I haven't found a direct reference to the original research.)

Note that the oscillation period is much shorter than the galactic revolution period. A Keplerian model would be insufficient.

[selden]

Sorry, I haven't been expressing myself very clearly, partially due to my own confusion.

I guess what I'm trying to say is that the fraction of the mass of the galaxy which currently is detectable by EM, either because something is emitting EM radiation or because it partially obscures something that is, is only a fraction of the galaxy's total mass. I had thought that the dark matter mass was much greater than the visible mass. That seems not to be the case.

My mistaken impression had been that the amount of (dark) mass in the halo was enough to make any concentration in the plane irrelevant: that there really wouldn't be any restoring forces. Hence many stars wouldn't be staying in the plane (unless their original orbits were coplanar); hence the current planar configuration would have to be just a fluke of the time of observation -- that after a while it'd look like an elliptical. (It still may, but not for that reason.)

Oh, well.

I still want to watch the stars move!

[alexis]

Whoa! I think your discussion has run a little wild here There is absolutely no way you can expect Celestia to make detailed emulations of gravitational interactions between galactic stars. As pointed out, you simply need too much computer power to that at any reasonable fps (remember, Celestia renders in realtime).

However, that's not the issue here. The stellar database is an infinitesimal fraction of all stars in the Milky way. Even the giant 2M catalogue by Rigel contains only 0.001% of the stars in the Milky Way. Of course, you could emulate a smooth gravitational potential of the galaxy and integrate stellar orbits over time, although that would also be a bit demanding. But that's not either the issue.

Rather, here are the top three obstacles to overcome in showing time-dependent stellar motion:

1) Data. We have radial velocities data only for a very tiny part of the Hipparcos sample. In most cases the proper motion (the spatial motion tangential to the line of sight from Earth) is well known while the radial velocity (parallell to the line of sight) is unknown. We thus simply don't have data to render the future or past positions of the stars in 3D. If Celesta would be restrained to an Earth-centred view of the universe in 2D there would be no big problem since the radial velocity then is unimportant for the stellar motion in a first approximation, and several 2D planetarium programs actually offer the user to change epoch in a limited time range and watch the stars of, e.g. the pharaos. For 3D, we may have to wait for comprehensive surveys, like the one GAIA will provide us in some 15 years.

2) Efficiency. I'm not sure how Celestia's stellar data structure is implemented, but I'm sure it's highly optimised to static, non-moving stars. To make an efficient data structure of moving stars would be a true challenge. There are too many stars to not have an efficient data structure to retrieve them from. This doesn't exclude the possibility that some clever guy (like Chris) comes up with a brilliant idea on how to do it.

3) Realism. Time runs swiftly when interstellar travelling, so that current stars in our local universe would rapidly diffuse out and be replaced by stars from other parts of the galaxy, stars which are completely unknown to us (particularly the fainter ones). And, on time scales of millions of years (one revolution around the galaxy is about 200 million years), the stellar evolution should also be taken into account for individual stars, further changing the local population. I therefore think the current solution is better from this perspective, since we at least have a representative population of stars of any time (on time scales of billion years) in our local universe.

I'm sorry if I sound a bit pessimistic (my intentions are to be realistic), I would also love to see stellar motions! Imagine a slowly rotating galaxy from some distance... well, it's allowed to dream.

/Alexis

[selden]

Alexis,

You're even more negative than I am I think it's better to concentrate on what can be done, even if it's not optimal. I try to limit my negativity to figuring out many the things that are going to cause problems, but then proceeding in spite of them, often finding that they aren't as bad as first imagined.

I'm certainly not proposing including gravity. Adding appropriate terms to Keplerian orbits would be more than adequate. Even simple linear approximations would be good enough for "short term" observations. And I'm not suggesting simulating the galaxy as a whole, either (maybe in another 5 years, though...). However, it seems to me that the realtime 3D nature of Celestia provides a much more useful tool for observing some effects than do the traditional 2D representaions of the stars around us.

1) grumble. You're right. I had rashly assumed the Hipparcos catalog itself included radial velocities. Well, there is the Hipparcos Input Catalog, which lists stars that they had intended to measure with Hipparcos. Hopefully it wouldn't be too hard to correlate its entries with the same stars in the Hipparcos and Tyco catalogs. Apparently it includes radial velocities for many of the stars it lists, although it only (!) contains 118,000 stars.

2) yup. Limiting the radius of interest might help.

3) I can't get too excited about this limitation: I'm interested in watching the ones that we have!

[Guest]

Whoa! I think your discussion has run a little wild here There is absolutely no way you can expect Celestia to make detailed emulations of gravitational interactions between galactic stars. As pointed out, you simply need too much computer power to that at any reasonable fps (remember, Celestia renders in realtime).

I don't think anyone here was seriously considering this. Numerical integration of the equations of motion for 10^11 stars wouldn't be fun for anyone.

However, that's not the issue here. The stellar database is an infinitesimal fraction of all stars in the Milky way. Even the giant 2M catalogue by Rigel contains only 0.001% of the stars in the Milky Way. Of course, you could emulate a smooth gravitational potential of the galaxy and integrate stellar orbits over time, although that would also be a bit demanding. But that's not either the issue.

Rather, here are the top three obstacles to overcome in showing time-dependent stellar motion:

1) Data. We have radial velocities data only for a very tiny part of the Hipparcos sample. In most cases the proper motion (the spatial motion tangential to the line of sight from Earth) is well known while the radial velocity (parallell to the line of sight) is unknown. We thus simply don't have data to render the future or past positions of the stars in 3D. If Celesta would be restrained to an Earth-centred view of the universe in 2D there would be no big problem since the radial velocity then is unimportant for the stellar motion in a first approximation, and several 2D planetarium programs actually offer the user to change epoch in a limited time range and watch the stars of, e.g. the pharaos. For 3D, we may have to wait for comprehensive surveys, like the one GAIA will provide us in some 15 years.

Most people (I suspect) are interested in seeing how the brighter (constellation) stars move over short (~10,000 year) periods. It would be possible to fade out stars depending on their velocity data quality (no RV, no PM, neither, both). I don't think it would be the least scientifically accurate feature on Celestia by far.

2) Efficiency. I'm not sure how Celestia's stellar data structure is implemented, but I'm sure it's highly optimised to static, non-moving stars. To make an efficient data structure of moving stars would be a true challenge. There are too many stars to not have an efficient data structure to retrieve them from. This doesn't exclude the possibility that some clever guy (like Chris) comes up with a brilliant idea on how to do it.

How about copying stars out of the display tree (or whatever's being used) to a temporary list containing only those that have complete velocity information, when "time travel" mode is on? A constant 3D velocity model can definitely be applied to 30,000 stars at about 5fps on a 1GHz Athlon, in my experience (that's without using anything in the way of 3D hardware assistance).

3) Realism. Time runs swiftly when interstellar travelling, so that current stars in our local universe would rapidly diffuse out and be replaced by stars from other parts of the galaxy, stars which are completely unknown to us (particularly the fainter ones). And, on time scales of millions of years (one revolution around the galaxy is about 200 million years), the stellar evolution should also be taken into account for individual stars, further changing the local population. I therefore think the current solution is better from this perspective, since we at least have a representative population of stars of any time (on time scales of billion years) in our local universe.

I agree that modelling galactic orbits over more than a few tens of thousands of years is not viable. On timescales shorter than that, I can't think of a serious reason why it wouldn't work for the bright stars.

[erostosthenes]

I agree that modelling galactic orbits over more than a few tens of thousands of years is not viable. On timescales shorter than that, I can't think of a serious reason why it wouldn't work for the bright stars.

cos the bright stars are the ones with the briefest lifetimes. a bright star like betelgeuse probably won't even be around for another 1000 years (it's already in its red giant asymptotic phase). it's fun to think about stellar motions for celestia, but the bottom line is that it just won't work, or it'll look ridiculous.

[alexis]

When I researched this some years ago, I found that only a few thousand stars have measured radial velocities. That's about as many stars as can be seen with an unaided eye a clear dark night, and the measured stars are not necessarily the brightest ones. The early ones (of spectral class O,B,A etc), that are among the most luminous, often rotate very rapidly which broadens the spectral lines and makes it difficult to measure their doppler shift accurately. If anyone happens to know of a larger database, I'd be happy to learn.

Of course, with only some thousand stars there would be no big problem to show their 3D motion in Celestia, or even to simulate the galactic gravitation over a few thousand years. But I'm not sure it would be that interesting, considering the vast majority of stars (with potentially high 3D velocities) would still be shown static. 5 fps doesn't sound very attractive to me.

/Alexis