New star generator for other galaxies... (work in progress)

[Cham]

And finally, I've added 200 giants in the core. Here's a view of the galactic center (see the nice color match with the sprites representation !), and the HR diagram of the 200 supplementary stars (non-main sequence) :

core.jpg

giants.jpg

The little square on the upper-right part of the HR plot is Betelgeuse, for comparison.

I didn't found any exagerated stars yet. Actually, the size and luminosity appear to be very "conservative" for all of them, so I think this is it !

Take note that I'm not adding any white dwarfs, pulsars or other "exotic" stars in there. Maybe much later.

[t00fri]

Cham,

for comparison, see the video from the paper by FA in my edited answer to Selden.

http://rqgravity.net/images/spiralmotions/gss.avi
(Depending on your line speed you might have to download it first.)

In this complete simulation the whole galaxy model only uses 4500 stars and still arrives at a viable dynamical MilkyWay model including time dependence and the central bar! The green component represents ingoing/outgoing gas.

Honestly, I don't understand the physics motivation for your quasi uniform disk star distribution (50000 stars!)



Fridger

[Cham]

Honestly, I don't understand the physics motivation for your quasi uniform disk star distribution

Fridger,

this video is a gross exageration. The star points are way too big in your video.

Why should I bother to do any galactic arms with less than 100k stars !? The stars are NOT VISIBLE enough to reveal ANY global structure. Adding arms that fit with the sprites model would be way too much work for nothing !! It's just not worth it !

Stars in Celestia aren't gross points like in your video. And please, don't forget that the picture above was used to show the distribution. It's NOT like that in Celestia, with proper AbsMagn ! Stars aren't that visible ! GEEZ !

Again, you are OFF-Topic !

[t00fri]

Honestly, I don't understand the physics motivation for your quasi uniform disk star distribution

Fridger,

this video is a gross exageration. The star points are way too big in your video.

Why should I bother to do arms with less than 100k stars !? The stars are NOT VISIBLE enough to reveal ANY structure.

The star size is clearly too big and the stars are not HR distributed. But the general physics looks intriguingly close. This video needs careful examination in order to note the subtle migration of stars during the animation, i.e. as function of time.

The whole point about the FA simulation is that already with 4500 stars, we can see the gross features correctly. In your case 50000 uniformly distributed stars don't look much like a spiral galaxy yet. Of course, in the FA case one needs to add many more stars and sprites for the gas before it could be entirely realistic.

Again, you are OFF-Topic !

Then it's better you talk to yourself in this thread.

Fridger

[Cham]

Fridger,

read again my previous post. Stars in Celestia aren't huge points. They aren't visible enough to reveal any structure when there are less than, say, a million of stars. So why bother about galactic arms !?

If you did tried my STC files, you'll clearly understand that arms aren't important at all for any reasonable STC files in Celestia.

[Hungry4info]

Yes, I need more examples.

And K010053
Spectral: K0II
Radius 0.35 R_sol

And K01026
Spectral: K0V
Radius: 0.19 R_sol

And K0102
Spectral: K0I-a
Radius: 0.39 R_sol

And_C K111491
Spectral: K1V
Radius: 2,200 R_sol

[Cham]

Yes, I need more examples.

And K010053
Spectral: K0II
Radius 0.35 R_sol

And K01026
Spectral: K0V
Radius: 0.19 R_sol

And K0102
Spectral: K0I-a
Radius: 0.39 R_sol

And_C K111491
Spectral: K1V
Radius: 2,200 R_sol

EDIT : Well, yes, they are too small. But try the new files below.

I've updated the files. Much better now, I think.

Same link as before :

For everybody, here's again the link to the files (updated) :

http://fsgregs.startlogic.com/Public_Fi ... 1stars.zip (2.6 MB)

EDIT : Arrgh ! The K stars do have a problem with their size. They are way too small on average. There's something fishy here and I'll have to check again my HR curve.

[Cham]

Now, there's something I don't understand. The K stars have apparently a too small radius, while still on the main sequence branch. Here's an example from my last files published in the message above :

fishy.jpg

What is happening here ?

[t00fri]

Cham

the problem with your stars being too dim e.g. in M31 has a simple explanation. In the Celestia 1.6.x distribution the apparent magnitude for stars is limited to <=12.0 (automag=on).

For your Andromeda stars and a typical distance of 50kpc after "GoTo", an upper limit of app.mag <=12.0 is entirely inadequate. In Celestia.Sci I have extended the limit to
app.mag <= 15.0. Therefore in Celestia.Sci, I can see most of your stars quite brightly . Here is a neat screenshot:
Click on the attachment to see your stars:

and15.jpg

Fridger

[Cham]

Okay, it's my HR curve which isn't precise enough in the K region. The approximate curve is going slightly under the real one, which explains the small radius in Celestia. Apparently, Celestia's calculations are very sensible to a small change in the curve parameters. I'll have to redefine it.

O well ....

[selden]

The STC files contain AbsMag values which are too dim, which causes Celestia to calculate radii which are too small.
For main sequence K stars, the AbsMag should be between roughly 5 for K0V and 10 for K9V, for example.
This is from eye-balling an HR diagram. I'm sure there are more accurate values available somewhere.

[Cham]

Very nice pict, Fridger.

Could you post a similar pict from face-on ?

[t00fri]

Cham,

here is another nice shot of the sideways MilkyWay with many beautiful disc and halo stars, that are entirely unrelated to YOUR stars. The stars you see here are not ficticious, but real stars, yet from the familiar 2million stars set (CML). All star parameters (RA, DEC, color class, app.mag) are precise except the distance values, as you surely know.

clicking essential!

mw15.jpg

Fridger

Edit:
With a few lines of C/C++/Fortran/Perl code the binary 2M stars can be read in and the realistic HR-diagram data then be printed out. One can even restrict the used stars by angular cuts to the ones that are associated to the MilkyWay disc. Any standard plot program then gives an accurate and realistic HR-diagram.

With Maple/Mathematica one then can easily do a least-square fit of the HR-data to some suitable analytical function, which then serves as a highly realistic probability distribution for colored star generation.

[Cham]

With a few lines of C/C++/Fortran/Perl code the binary 2M stars can be read in and the realistic HR-diagram data then be printed out. One can even restrict the used stars by angular cuts to the ones that are associated to the MilkyWay disc. Any standard plot program then gives an accurate and realistic HR-diagram.

With Maple/Mathematica one then can easily do a least-square fit of the HR-data to some suitable analytical function, which then serves as a highly realistic probability distribution for colored star generation.

Unfortunately, I'm unable to do that by myself. I may find an approximate HR curve, better than the one I defined, but not the way you suggest. I'm using a polynomial of degree 4, with three points fixed : the two extremities, and a third one associated to the sun as a reference. The curve is a smooth function AbsMagn(Spectral Type).

May I ask you to find it for me, using your method ?

If only I could extract the equation behind the nice smooth curve embeded in this applet :

HRapp.jpg

[starguy84]

I'm surprised this discussion has gone this far without any mention of mass functions (intial or current). Basically, to achieve a realistic representation of stars, you need the H-R diagram for the properties of stars, AND a mass function to relate the number of stars at each bin- very many M, very few O. Likewise, the O stars should trace the gas and spiral arms because they are so short-lived they rarely go far from the gas clouds that formed them; K and M dwarf stars should be found everywhere.

50,000 stars aren't enough to give a reasonable distribution for an entire galaxy (as you demonstrate when your average distance between stars is 300+ ly). Are you going for overall distribution, or do you want to just pick out the 50,000 brightest stars (which would all be O dwarfs, B dwarfs, M supergiants, K giants?

[Cham]

I'm surprised this discussion has gone this far without any mention of mass functions (intial or current). Basically, to achieve a realistic representation of stars, you need the H-R diagram for the properties of stars, AND a mass function to relate the number of stars at each bin- very many M, very few O. Likewise, the O stars should trace the gas and spiral arms because they are so short-lived they rarely go far from the gas clouds that formed them; K and M dwarf stars should be found everywhere.

50,000 stars aren't enough to give a reasonable distribution for an entire galaxy (as you demonstrate when your average distance between stars is 300+ ly). Are you going for overall distribution, or do you want to just pick out the 50,000 brightest stars (which would all be O dwarfs, B dwarfs, M supergiants, K giants?

Very interesting comments, starguy84. Thanks.

As for the mass function, I did incuded it very roughly (it's a first order approximation) : I simply pondered the statistic counts to favor the small masses, and make the O stars less likely (see the HR diagram I posted above).

I may add a more accurate distribution on the various spectral types, but I need some analytic function to do it. This shouldn't be difficult to do.

And there's also a choice to make, as you suggested : what do we want to represent in the galaxy ? The most luminous stars (the most visible ones) ? Or just a random sample of all the stars ? I prefer the second choice and this is what I try to do. In this case, I should take the mass function into account. However, with less than 100 000 stars scattered on the whole galaxy, it really doesn't matter how I define the distribution. The population density is so low that it can't represent anything at all !

The distribution has a very high pedagogical value, though : the very strong feel of the vastness of space and the huge volume of the Andromeda galaxy. This is so striking while navigating through the galaxy, it's almost vertiginous ! Strangely, it's the first time I have this impression in Celestia, probably because the default stars in the Milky Way are all packed into a relatively small region of the MW.

So, wathever the distribution (realistic or not), this experiment is really valuable !

[chris]

And there's also a choice to make, as you suggested : what do we want to represent in the galaxy ? The most luminous stars (the most visible ones) ? Or just a random sample of all the stars ? I prefer the second choice and this is what I try to do. In this case, I should take the mass function into account. However, with less than 100 000 stars scattered on the whole galaxy, it really doesn't matter how I define the distribution. The population density is so low that it can't represent anything at all !

I see a potentially serious problem with using a realistic mass function: the stars that will be visible at large distances and that show the spiral structure of the galaxy are also extremely rare. For each bright supergiant or OB star, there will be tens of thousands of red dwarf stars that will only be visible to a viewer a few light years away from them. In practical terms, you'll have a catalog filled with stars that will never be seen by anyone.

Unless you're generating a catalog with millions of stars, I think it would be more useful to generate just the bright ones.

--Chris

[Cham]

Very important point, Chris. Thanks !

So no mass function.

Now, if I limit myself to visible stars only (O and B), the distribution could end to be boring.

I could publish two versions however : one for the visible stars only, and a second one to show the full Main Sequence (which was my goal at first). The user could then select and use the version he/she prefer (or even both versions, while not representative of the global population).

Currently, with the defaut database in the Milky Way, it isn't always obvious to show to students what stars are part of the main sequence and which one isn't, while traveling through the Milky Way. Especially since there is no HR diagram (by default) in Celestia. With my project for M31, I intended to show only the main sequence and ALL its representatives stars (from the frequent low mass up to the rare large mass). I added some K, M giants stars (not on the main sequence) into the core, though, through a third STC file.

[Cham]

Chris,

what would be the impact on frame rate if we use a database of, say, 1 million stars in M31 ? Will it have an impact while traveling far away from M31 ? And what if we travel through the MW (while M31 is still visible in the background) ?

Of course, I could do the experiment to see by myself if there's such an impact, but I would like to know before compiling a large database.

And there's also the possibility that I populate some other galaxies in the local group (up to 10 MLY !)...

Once my code is finished (very soon !), I'll be able to populate any part of the Celestia universe (within its 10 MLY limit, of course), with any number of stars.

[Cham]

Update to the HR main sequence curve ! Wew !

Here's a graphic showing two curves made with Mathematica and several data points from three sources. The vertical axis is the AbsMagn (with a wrong sign because it's pretty hard to invert this axis in Mathematica). The horizontal axis is a linear variable from 0 to 1 associated to the spectral classes (0 for the O0 class, and 1 for the M9 class).

HRCurve.jpg

The light-blue curve was the one I used in my previous compilations (the STC files published on the previous pages). It's clearly a bad approximation of the main sequence !

The blue dot near the middle is our Sun.

The red and green dots are data from two independant sources. The purple dots are coming from a more precise graphic (a scanned page from an astrophysics handbook). The red curve is a polynomial of degree 5 which interpolates these points. I'll compile some new STC files using that curve.

It's now very clear why my K stars were too small, and some A and F stars were too large !

Once the new STC files are compiled with the red curve, I'll show the HR plot from within Celestia, using my LUA HR tool...