Celestia Forums

Hello !

I have generated a new extended stars database for Celestia that should no more enter in conflict with Sun's neighborhood.

You can download them for the moment at :

http://perso.wanadoo.fr/celestia.stars/

V2.1 A : 1 016 385 stars (16.6 Mb)
V2.1 C : 2 072 873 stars (32.7 Mb)

Imprecision on data that comes from the Tycho catalogs and the current limits of the interpolation procedure are more noticeable for Sun’s neighborhood. To avoid any conflict with well known nearby stars, no Tycho stars have been added in this new release below 25 pc. Tycho stars with known parallaxes have been put farther when it has been established that error on their parallaxes could lead to stars with distances greater than 25 pc.

I will now work on the systematic integration of non-Hipparcos stars that come from the Gliese catalog. The goal is of course is to get an accurate representation of our neighborhood within 25 pc.

Let me know if you find other inconsistancies that could be corrected.

Best regards

Pascal

What happened to "B" ?

Hi Paul,

I think you mean V2.1 B (and not Sirius B) ?

It is a database with about 1 500 000 stars (between the A and C stars count) that I didn't put online, because I have little storage place on my personal web page.

Pascal

Works like a charm, thank you

Pascal, just incase you are not aware of this.
I found these stars when using your V2.1 C stars.dat with my mySQL database mod.

TYC 6623-145-1 (1001456623)
TYC 5996-1225-1 (1012255966)
TYC 7880-970-1 (1009707880)

They all have a parrallax of zero and a distance from Sol of about 3million light years.
The reason i found these was that my mod was crashing when trying to load these stars (this is my fault, I need to make the star reading code more like celestia's).
I checked them using celestia and they there but way beyond celestias rendering distance at 3million lightyears.

I used an 'ORDER BY parallax' and also an 'ORDER BY distfromsol DESC' to find these.
Im guessing they shouldn't be there.

cheers.
, and nice work, Im looking forward to the next version with more nearby stars.

Thank you Mark,

I will check the extraction procedure again to find out why these stars have been put at an infinite distance.

I wish I could use your very useful tool, but I didn't find any time yet to see if it can be compiled in OS X.

Best regards

Pascal

Pascal

The embedded mysql library will need to be compiled and linked to, (OS x isnt mentioned in the mySQL manual but im guessing its done in the same way as on linux. )
The generation process is pretty messy and uses two different programs (one can generate planets). Once i separate the star generation stage it should be easy to port.
I have a linux system though I havent compiled or had celestia running on it yet. Given time i should get this working under linux, but i cant guess how long that will take as I am no linux expert.

In the mean time, if you want me to do any particular queries for you, just ask.

Although I understand that the extended stars catalogs' distances aren't as reliable as those in the main Hipparcos catalog, I've found some that are quite wrong.

For example, I've been investigating the stars in M16, the open cluster which is associated with the Eagle Nebula. They are about 5000-7000LY away in the Sagitarius arm of the galaxy. I had hoped that the Tycho distance values would help to improve the accuracy of that distance. Unfortunately, I've found that all of the values for stars that are "obviously" members of M16 in the extended Stars.DB v2.1A and C are much too small.

For example, the cluster member stars HD 168075 and HD 168076 are classified by Simbad as type O+ and O5 respectively.

Stars DB v2.1A classifies them as type F3V at 518LY and F4V at 370LY: their distances are too small by about a factor of betwen 10 and 20.
I'm guessing that the stellar types are somehow being calculated from the distance values, rather than the other way around.

When I looked up these two stars in the Tycho database on the Web, neither of them have any trigonometric parallax mentioned. Presumably their distances were somehow calculated using other methods.

sigh.

added later:
Here's a Cel:// URL centered on M16. Essentially all of the stars in the field of view are members of the cluster.

cel://Follow/m16/2003-05-18T20:01:33.06408?x=Y1/0Rdd8eKx7DA&y=lpJoQrNjmkIB&z=lN53Aga5UEPL/////////w&ow=0.040008&ox=-0.003490&oy=0.995697&oz=-0.083520&select=m16&fov=0.135422&ts=1.000000&rf=66115&lm=0<d=0

and here's a picture of the region:

(As usual, this is a link to a higher resolution image. Also, there's a slight rotational offset between the photograph and where Celestia draws the stars.)

You are right Selden but it is a very special case. These stars are very hot and distant O and B stars.

The Tycho catalog for example for HD 168076 gives :

Bmag = 8.640 Vmag = 8.225 ====> B- V color index is equal to 0.415

The extraction procedure assumes then it is a class V star, and with the fitting function for class V stars, computes a visual magnitude equal to 2.948 for this star, leading to a distance of about 370 ly.

Main sequence stars with a color index of 0.415 correspond to a F4 spectral type.

What is going wrong ? The color index of 0.415 ? The mean error given by the Tycho catalog for B and V mag are equal to about 0.02.

O stars should have B-V color indexes of about -0.3
(according to the Handbook of Space Astronomy & Astrophysics)

A color index of 0.415 cannot correspond to 0 or B stars.

I think, that could be a consequence of the reddening of their light through the gaz of the nebula which lies in a HII region of the Galaxy.

What could be the solution ?

- to modify the extraction procedure, and for Tycho stars with a given spectral type, to rely on this field and no more on the color index ? But, if I remember well the results of the preliminary test I have made on the Tycho catalog, the spectral type is generally not given.

- to complete the extraction procedure with a list of stars (like these M16 stars) that should be handled in a special way ?

I think, the best solution would be to develop a star editor for the Celestia stars database. A star editor with the common possibilities of an editor : add, delete, modify, search for stars with certain criterias.

I once started the project in Cocoa because I am on OS X, but didn't find the time to finish it (and it will give only a Macintosh application).


But a powerfull editor could be a good solution, isn't it ?


Pascal

Pascal,

From what I've read, indeed the problem is that the stars are reddened by the nebulosity around them.

Maybe one method of partially automating the corrections would be to include information (spectral type in particular, perhaps) from some other catalogs for particular groups of stars, especially the stars in known associations. Of course, different catalogs use different presentation formats (different columns with different widths for the same information), which is a pain.

An editor would help, too, of course, but with 2 million stars, that's a lot of editing although perhaps not so much when working with just particular clusters.

The Messier objects might be a reasonable choice for the clusters to start with, however it's done.

Rigel wrote:I think, the best solution would be to develop a star editor for the Celestia stars database. A star editor with the common possibilities of an editor : add, delete, modify, search for stars with certain criterias.

If the tables are not compressed my MySQL add-on already has the capability of being able to perfom INSERT and UPDATE statements. It can then review the changes immediatley from within the modified Celestia. All it would need is some extra code so it could write the relavent columns to a stars.dat file. I could have a go at it if anyone is likely to make use of it.

Though I cant do it myself, porting it to OSX shouldn't be too hard either.

later: It should even be able to read new star data from text files.

Hey, marc, if you want and have time to compile in linux, just ask me. I'll be pleased to help you. Just inform me what version you have, as celestia source is a bit squeamish about the autoconf and KDE versions (I am getting mad at committing with its requirements to have both autoconf >=2.54 and no more than 2.53, for example ).

Jamarsa, that would be great. I might have a go at it this weekend, I'm running RedHat 7.2. I installed KDE but gnome is my primary desktop. Its been a while, but im pretty sure I installed everything else as well. Can you see any immediate problems with this setup?
The only problem is I've only got the crappy onboard video chipset (SIS630 I think) but it should be good enough for testing.

marc wrote:Jamarsa, that would be great. I might have a go at it this weekend, I'm running RedHat 7.2. I installed KDE but gnome is my primary desktop. Its been a while, but im pretty sure I installed everything else as well. Can you see any immediate problems with this setup?
The only problem is I've only got the crappy onboard video chipset (SIS630 I think) but it should be good enough for testing.

I think it's a little old version, because Celestia needs KDE 3.1, and unless you want to compile KDE from source, you wouldn't be able to find RPMs for that version. Try to upgrade to 8.0 or, at least, 7.3. I use also Gnome as my desktop, though I'm able to execute KDE programs thanks to the cross-compatibilty (and the presence of KDE base, of course ). You can obtain KDE for redhat at this link:

http://kde-redhat.sourceforge.net/

I'm pretty sure redhat offer them in their distribution too, (albeit they have a clear preference for Gnome).

Also you would have to upgrade your autoconf, automake, etc. Maybe it would easier to install the last RedHat (or Mandrake 9.1 )

I had a couple of years ago that SiS chipset and it worked flawlessly under Linux (Hey!! my brother Juanjo helped to maintain the Xfree86 driver!!), but as you know, it's a bit spartan card by today's standards. I don't know about the 3D acceleration, although I could ask

PS: You could try to make a GTK version of celestia, of course, and as it isn't being maintained (but it works, Christophe takes care of that), it could work for you in the present status of your machine.

Jamarsa, Marc,

If you're interested I can arrange for a daily tarball of the CVS to be uploaded to my website (with the autoconf/automake steps done).

You'll still have to compile but at least you won't have to deal with the autoconf mess.

Christophe wrote:Jamarsa, Marc,

If you're interested I can arrange for a daily tarball of the CVS to be uploaded to my website (with the autoconf/automake steps done).

You'll still have to compile but at least you won't have to deal with the autoconf mess.

Interesting deal!! But It would increase your site's traffic, and I don't need a daily update. Moreover, I'd prefer to have myself a clean autoconf/automake environment in which create the Makefiles. I gave up with the make -f Makefile.cvs for a couple of days, because I got rid of all the previous versions of autoconf prior to 2.57, only to find that it needed the 2.53 (autoheader) again!!

Thanks for the offer anyway!!

I'd very much like to see and use a stars.dat editor, but it'd have to be one that runs under windows.

It already runs under windows and you can edit stars with it via SQL statements once you modify the generation process. It will show the stars as edited in celesita. With a few changes it will be able output to stars.dat, but no-one has done any editing with it yet (not even me), and you need to know some SQL to be able to use it. I might leave this feature until i port it to linux.
http://www.celestiaproject.net/forum/viewtopic.php?t=2518

marc wrote:It already runs under windows and you can edit stars with it via SQL statements once you modify the generation process. It will show the stars as edited in celesita. With a few changes it will be able output to stars.dat, but no-one has done any editing with it yet (not even me), and you need to know some SQL to be able to use it. I might leave this feature until i port it to linux.
http://www.celestiaproject.net/forum/viewtopic.php?t=2518

Oooh. That's a bit bloody fantastic . I've going to have me some fun playing around with that...!!

Marc, I'm eager to see your SQL feature working in linux...