Project Orbiter Galaxy

[jedidia]

Lots of them have been spotted, and they are in alright. The problem here is with the generator, which doesn't generate any at the moment. I'm reworking the table a bit, but I have virtually no references to go on, so this will be quite some guesswork...

 

[dumbo2007]

ok....i thought you were displaying stars straight off a static table that was available online and had position + mass data - but I wonder why a generator would not generate them - I mean do you not generate stars based on some age or mass constraints ? just curious...

 

[jedidia]

ok....i thought you were displaying stars straight off a static table that was available online and had position + mass data

That's one part of it (allthough mass data isn't part of the catalogue... the masses for most stars are still unknown. And the catalogue isn't online, it's a csv ion my hard drive). The other is a generator producing the rest of the galaxy.

but I wonder why a generator would not generate them - I mean do you not generate stars based on some age or mass constraints ? just curious...

As I said, the major reason seems to be wrong data in my reference table.

It goes like this: If a Star has enough mass to become a red giant and is in the right age, it's mass gets reduced by 40% (pretty humdrum estimate, but it's the best rule of thumb I found so far) and the code seeks for a Giant with apropriate mass in the table. But the mass of an M9III in this table is over 4 solar masses, which is way too high (it should be around 0.5), so the star is returned unprocessed. Since I'm very lazy with error handling in my code it took me a while to find this out, but after a few tests this indeed seems to be the reason.
On the checks I've run so far there were enough stars generated that had the right age (which at first I assumed as the cause, because the duration a red giant is around is rather small compared to the age of the galaxy), but their mass is beneath the minimal mass noted in the reference table.

Currently I'm trying to find reliable data to fix the table, but information on the subject is rather rare in the net... I'll end up guessing, I'm afraid.

 

[Linguofreak]

Yeah, that table (and any table in general) isn't really great for masses for anything that's not on the main sequence. The main sequence is tight enough that you can estimate mass fairly closely from spectral class, but off the main sequence stars change temperature and luminosity very rapidly and in different ways.

If I could have a copy of the executable I could check the way it generates things against my own astronomical knowledge. (Hopefully it will run under Wine).

EDIT:

The reason you can't really say "A K8II or a M9III has such and such a mass" is that different masses of stars leave the main sequence from different points and in different ways, and so two giant stars with roughly the same spectral class may have different ages, masses, and ways that they got to the spectral class they have now. Worse, quite a number of giant stars are pulsating variables of some sort, so that they may significantly (though often fairly predictably) change their positions on the HR diagram within a period of days. Mira variables can undergo changes in luminosity of ~10 magnitudes (5 magnitudes is a 100-fold brightening/dimming, so 10 is 10,000 fold!), and Type 1 Cepheids can go between F and K within days.

A few links:

http://www.daviddarling.info/encyclopedia/A/AGB.html (Nice picture of the HR track that stars will take after leaving the Main seq.)
http://www.daviddarling.info/encyclopedia/C/Cepheid_variable.html
http://www.daviddarling.info/encyclopedia/M/Mira_variable.html
http://www.daviddarling.info/encyclopedia/P/pulsating_variable.html

 

[jedidia]

If I could have a copy of the executable I could check the way it generates things against my own astronomical knowledge. (Hopefully it will run under Wine).

As soon as I have it running at a reasonable level of functionality, I will put a public demo out. Currently it doesn't make much sense, since I am aware of most problems with the current algorithm. Once it's finished and working the way I imagine it to, any input on how to make it better and more accurate will be welcome, but first I have to get the basics running.
Plus on saturday I leave for my "little buisness trip" which will take until the end of march, so it will be a bit delayed. I hope I find the time to work on it every now and then during that time so that the demo will be ready somewhen during the next two months.

Thanks for the links, though!

---------- Post added at 06:51 PM ---------- Previous post was at 06:09 PM ----------

Edit:

and so two giant stars with roughly the same spectral class may have different ages, masses, and ways that they got to the spectral class they have now.

What you're saying is that mass is not directly related to spectral class in a giant star. That would leave me the option of bringing a bit of randomness into this, but I'd rather find out about how the initial conditions are connected to the giant "produced". I spent a bit of time searching for data, so far nothing. You wouldn't know where to get that kind of information, would you? Slowly but inevitably the time I googled for this project grows more than the time I actually coded...

 

[Linguofreak]

As soon as I have it running at a reasonable level of functionality, I will put a public demo out. Currently it doesn't make much sense, since I am aware of most problems with the current algorithm. Once it's finished and working the way I imagine it to, any input on how to make it better and more accurate will be welcome, but first I have to get the basics running.
Plus on saturday I leave for my "little buisness trip" which will take until the end of march, so it will be a bit delayed. I hope I find the time to work on it every now and then during that time so that the demo will be ready somewhen during the next two months.

Thanks for the links, though!

---------- Post added at 06:51 PM ---------- Previous post was at 06:09 PM ----------

Edit:



What you're saying is that mass is not directly related to spectral class in a giant star. That would leave me the option of bringing a bit of randomness into this, but I'd rather find out about how the initial conditions are connected to the giant "produced". I spent a bit of time searching for data, so far nothing. You wouldn't know where to get that kind of information, would you? Slowly but inevitably the time I googled for this project grows more than the time I actually coded...

Well, for a given set of initial conditions, the type of giant depends on how long the star has been off the main sequence and what it's burning at the moment. There's basically a whole track for every mass. The English Wikipedia article on the Asymptotic Giant Branch has a picture with tracks for four different sizes of star, but you'll probably want something spaced a bit closer than that. It should give you a general idea, though.

As far as the lower limit for becoming a red giant, which you had been talking about: We don't know exactly where it is, because the stars in that mass range haven't left the main sequence yet.

 

[jedidia]

Interesting, I missed that picture until now. Majorly because I currently don't load any pictures in firefox (the son of our landlady currently has semester hollidays, which brought our little connection here far beyond the monthly download-limit...) and secondly because I didn't bother with the asymptotic branch too much yet (figured it would be easier to do the "regular ones" first.

As things look, my best bet will be to put an HR diagramm into the programm, then predefining some paths stars of different masses are going to take on it and work the rest out from there. If I found a few more examples for different masses that shouldn't be too much trouble.

---------- Post added at 03:07 PM ---------- Previous post was at 09:43 AM ----------

alright, I think I found enough data to pull it off... Now I just need to find the time.

 

[jedidia]

Orbiter Galaxy is definitaley being delayed quite some time... firstly because of a lot of other work that wants to be taken care of, and secondly because I have to do a complete re-write of the evolutionary model. Currently occupied with teaching the program to interpret and manipulate an HRD, but there will be a lot more work to it. I'm not exactly sure when I will have the demo finished, since after this is done I'll also have to evolve the systems (system creation works only for main sequence stars, so they have to be evolved with the stars), especially with the truckload of work that got dumped before my dor and will keep me rather busy until the middle of april, and there's no telling what summer will bring. But I sure hope to have the demo finished before summer.

 

[tgep]

No worries. Creating Galaxies aint like baking a cake. The best things in life are worth waiting for. I've got my new Orbiter " sci-fi only " clean install up and running like a champ so I'm all ready for testing when you are.

 

[dumbo2007]

yep...keep going !!

 

[jedidia]

Interesting, what kind of data files are those? I'd expected a probability tables and stellar classes table, nothing big.

Stellar class table is already in, the trouble is with evolving the stars. Everything that goes out of the main sequence gets nearly unpredictable. You can't tell the mass of an M6III just because you know it's an M6III (as you could for stars in the main sequence). So my evolution model that just assigned a certain mass to a certain class of stars was wildly unprecise.

Instead, I'm now switching to generating the star in the main sequence, and then following the path such a star would take along the HR-diagramm when it leaves the main sequence, according to it's age, to determine what becomes of it. That first let me with creating a logarithmical HRD that my programm could interpret, and now it means vectorizing a few dozen paths that stars of different masses take along the HRD. That doesn't sound like much, but since all the data I have available is in form of pictures describing those paths, I have to read them out and convert them to units that make sense on my HRD by hand (read: measure out pixels, enter into pocket calculator, convert, then write result into data file). The data file will probably not be much bigger than some kilobytes in the end, but converting that data is a tedious task that takes some time...

 

[Artlav]

Instead, I'm now switching to generating the star in the main sequence, and then following the path such a star would take along the HR-diagramm when it leaves the main sequence, according to it's age, to determine what becomes of it.

Ok, for what i understood, you end up with stars that have mass, age and class. How much more accurate that is than assigning same kinds of masses and classes to the stars just so? And relative to what reference accuracy?

I've been working on a galaxy generation too, http://s-way.sourceforge.net/main.eng.html , and just getting them randomly along the distribution probabilities and class traits gives a plausible set of stars. Or i think so.

 

[jedidia]

Ok, for what i understood, you end up with stars that have mass, age and class.

No, I just ended up with stars that had mass (based on an estimate of the mass-loss the star underwent), but had absolutely no Idea what class it would be.

I've been working on a galaxy generation too,

I know, and it's certainly a great thing you guys did with spaceways. By the way, you once told me that you'd send me your texture generation libraries. I hope the offer still stands, because I'll need them as soon as I get finished with the generator.

and just getting them randomly along the distribution probabilities and class traits gives a plausible set of stars. Or i think so.

It's a reasonable way to go. One problem is that I absolutely wanted to use stargen for system generation, because I never saw a generator giving so realistic and detailed results. However, it forced me to start with main sequence stars. I cannot generate a realistic system for an M6III. I have to know what that M6III was when it was in its main sequence in order to create a system for it, then evolve the system along with the star. If I would have gone via probability table, I'd still have ended up with lots of giants and no possibility to determine what they once where (a problem I am still having with the real stars imported from the catalogue, of course, I'll have to do a bit of random guesswork with those...).

 

[Artlav]

By the way, you once told me that you'd send me your texture generation libraries. I hope the offer still stands, because I'll need them as soon as I get finished with the generator.

Yes, that part of it kind of went out of focus. There are 5 classes and a dozen types total, some of which are quite weird, not exactly a complete set. Anyway, i can wrap it out in a matter of hours, so just ask when you're ready.

However, it forced me to start with main sequence stars. I cannot generate a realistic system for an M6III. I have to know what that M6III was when it was in its main sequence in order to create a system for it, then evolve the system along with the star. If I would have gone via probability table, I'd still have ended up with lots of giants and no possibility to determine what they once where (a problem I am still having with the real stars imported from the catalogue, of course, I'll have to do a bit of random guesswork with those...).

Why determine what they once were? Do you model galaxy in 4D, with time also adjustable?

 

[jedidia]

Yes, that part of it kind of went out of focus. There are 5 classes and a dozen types total, some of which are quite weird, not exactly a complete set. Anyway, i can wrap it out in a matter of hours, so just ask when you're ready.

great. It will still take a little while, I'll give you a call.

Why determine what they once were? Do you model galaxy in 4D, with time also adjustable?

Well, theoretically I could do that with my current engine, but that was not the motivation behind it. I just can't rig stargen to output systems for anything else than main sequence (I already had to extend it to get believable systems on the whole range of main sequence stars), and I didn't feel like writing an inferior generator just to throw out systems for giants. So I decided to create the systems for main sequence stars and then evolve them with the star to get results as believable as possible. It might be a bit overkill, but after all it is not THAT much more work (I thought. I got a nasty surprise when I noticed that I can't determine what my stars are actually becoming. It's still not that much more work, it's just darn boring work...). This is a programm that focuses solely on the creation of a moderately believable galaxy, after all, so I felt like putting a bit more scientific detail into it than can be found in most other generators.

 

[Linguofreak]

I just had a thought:

If you have access to a Linux or MacOS system that you have permission to compile, install, and run programs on, you could try this program:

http://chandra.as.arizona.edu/~dave/tycho-intro.html
http://chandra.as.arizona.edu/~dave/tycho7-10.gz

(The site is poorly laid out and the documentation and website are a version old (I forget where I found out that version 7 even existed) and not very good, but I did manage to compile and run it a while ago).

You can give it initial parameters (though the setup is a bit tricky), and then run it on those parameters for as long as you wish, or until it crashes or otherwise exits (it will stop if it determines, for instance, that a supernova has occurred, and you often have to fiddle with a run to make it not crash on the input data).

 

[jedidia]

hmmm... if anything, I'd be tempted to just take the code and plug it into my generator. But unfortunately I don't know fortran, and I guess the math would be overkill for any cpu to handle when it has to generate and evolve 10'000 stars in the blink of a second. It would get too slow.

I would have access to a macOS, allthough an old one, but it might take me a lot longer building the stuff and understanding how to use it, and in the end I'd still have to log the data I need by hand. I only have a dozen or so paths left to vectorize, after that the legwork will be done (allthough, I wouldn't mind having some more data of HR-paths of stars between 0.5 and 1 solar masses, I couldn't really find much data in that cathegory). Since you already seem to have expierience with it, what does the program output? does it output the path a star takes along the HR-diagramm, with timesteps? if yes, I think it would be worth the trouble.

 

[jedidia]

after a bit more research into the matter I learned that there should be a database existing that has quite detailed tracks of stars ranging from 0.8 to 120 solar masses. Unfortunately I am only able to find bits and pieces of it. If anyone happens upon a complete version of that database, please tell.

---------- Post added 02-28-10 at 10:05 AM ---------- Previous post was 02-27-10 at 12:08 PM ----------

compiled a detailed database of evolutionary tracks from 0.5 to 7 solar masses so far. Still looking for data on higher-mass stars.

 

[Linguofreak]

hmmm... if anything, I'd be tempted to just take the code and plug it into my generator. But unfortunately I don't know fortran, and I guess the math would be overkill for any cpu to handle when it has to generate and evolve 10'000 stars in the blink of a second. It would get too slow.

It wouldn't be any good for that. It takes a good while as the only program running to give a track for a single star, but, if you lack a preexisting database, it would probably be the fastest way to make one.

I would have access to a macOS, allthough an old one,

It would have to be OS X, at least. Before that MacOS wasn't unix-like.

 

[jedidia]

finally found the extended database up to 120 solar masses, but it's in a different format than the other Database, and it's practically undocumented. But as soon as I can make sense of it and compile it together with the rest of the data I'll be ready to finally go on coding. I think implementing the database won't be too much trouble, but there's still an algorithm for evolving the systems along with the stars waiting and a bit of GUI-work. Looking at the setbacks I have had so far I won't promise any release dates anymore... :shifty: