Astronomy / Space Locating a real star in ED?

Guywired

Guywired
Genar-Hofoen said:
If M. Brookes was to come back to me and say "yes we actually perform calculations on every one of those 400 billion stars systems at our database end", I'd be very surprised ;)
Click to expand...

lol, yeah, I doubt it too.

Pretty certain they do that on a system level, not a galactic level.

Still, it would have made for a pretty entertaining behind the scenes video.
 

Michael Brookes

Game Director
NeilF said:
I know there's "2MASS....." ones in ED... So I'll search for it on that reference rather than the Kepler ID...

Ta
Click to expand...

There is, but generally only for populating various star clusters and special cases. Most of the 2MASS data isn't really detailed enough for accurate positioning, although that's true for most distant catalogues.

Michael
 

Para Handy

Para Handy
Michael Brookes said:
There is, but generally only for populating various star clusters and special cases. Most of the 2MASS data isn't really detailed enough for accurate positioning, although that's true for most distant catalogues.

Michael
Click to expand...

So presumably you use galactic co-ordinates to insert systems to an accurate position?
 

Alec Turner

Alec Turner
Last edited:

Genar-Hofoen

Genar-Hofoen
Michael Brookes said:
At the simplest level yes, but also accounts for age, life cycle, and a bunch of other stuff.

Michael
Click to expand...

Thanks for the answers - interesting.

You know what would be fantastic - an FDEV-created youtube video on exactly how your galaxy was created from scratch.

Did you start off with a 'whole', non-spiral galaxy then 'hollow' it out into a spiral, or was it 'grown' into a spiral?

Thoughts like that have crossed my mind when I've been pondering making an OpenGL representation of a spiral galaxy for Captain's Log. I started twitching shortly after thinking about that endeavor.. "How the heck would I go about doing something like that in OpenGL!?"

Hats off to you guys - I don't know how you pulled it off, but I bet it involved lots of people with maths, physics, and database/IT degrees :)
 

aymerix

aymerix
I like very much the astronomy (and Elite even more;). This question is fascinating. Have you (or can plan it) a "Meet the dev" on: how were added stars in Elite, the management of mass, etc?
 
Last edited:

edd

edd
The coordinate transformation isn't too bad - although I've not actually got my code working properly yet.

The units of the coordinates are obviously in light years. Sgr A* is at a unit vector of (0,0,1) - well close enough, it's unnormalised vector is (25, -21, 25900), and you can look up the Sgr A* RA and Dec. So you just need to determine the rotation matrix and do a bit of unit conversion (the astronomical to cartesian is textbook). My matrix finding appears to be buggy right now though.
 

AnthorNet

AnthorNet
Michael Brookes said:
There is, but generally only for populating various star clusters and special cases. Most of the 2MASS data isn't really detailed enough for accurate positioning, although that's true for most distant catalogues.

Michael
Click to expand...

which would explain why the 2MASS appear twice in game :)
 

NeilF

NeilF
I'd like to find 2MASS J19280644+3736023 (KIC 2305543) but I just don't think this is in the game :(

I can find systems beginning with "2MASS J193"... But not "2MASS J192" :(
 

Athan

Athan
Genar-Hofoen said:
Thanks for the answers - interesting.

You know what would be fantastic - an FDEV-created youtube video on exactly how your galaxy was created from scratch.

Did you start off with a 'whole', non-spiral galaxy then 'hollow' it out into a spiral, or was it 'grown' into a spiral?

Thoughts like that have crossed my mind when I've been pondering making an OpenGL representation of a spiral galaxy for Captain's Log. I started twitching shortly after thinking about that endeavor.. "How the heck would I go about doing something like that in OpenGL!?"

Hats off to you guys - I don't know how you pulled it off, but I bet it involved lots of people with maths, physics, and database/IT degrees :)
Click to expand...
I believe it starts with a matter density function to generate the general shape, and then something for the current position of the arms (which are related to the current position of a travelling density *wave* within the galaxy, with the brighter arms not being where there's more mass/stars per se, but instead where the density has been increased 'recently' by this wave causing more/newer (brighter) star formation). After that it's just some seeded PRNG rolls for what will be in a given volume (sector) of the galaxy within those parameters. Any overrides would get applied before the lowest level of this, as Michael says, using up some of the available mass.

Thus adding any new 'override'/real stars to the game will almost certainly cause some PG star/systems to disappear.

The older Hipparcos data that was largely used for distances is only of any real accuracy out to 1000-1500ly - beyond that the error in the measurement starts to be larger than the measured value.

Adding in all of the more accurate Gaia data now available would cause a HUGE increase in the numbers of overrides, greatly distorting the PG systems 1000-2000ly around Sol, and cause the game client to bloat a lot (all of the overrides data is directly in the client, yes?).

But, yes, a video, akin to that AWS one about the backend, about all of this would be most welcome.
 

edd

edd
Hmm. Calculating the coordinate transformation between astronomical [spherical] and E: D [cartesian] is faintly more non-trivial than I thought. I'm assuming Sgr A* is in basically the right place, as it's so close to being exactly in the z direction. Having that gives a rotation from one vector to another about a specific axis - basically you have a great circle between them giving the axis and angle for that arc. Then a second star rotated about the axis you have for the vector for the first one (basically rotating the coordinate system about Sgr A*) should fix it and give you the single rotation you actually need from multiplying the two rotation matrices, but doing that for a couple of fairly simple cases and checking for the third (and handling some not unexpected rounding errors as best as one might reasonably manage) doesn't seem to give satisfyingly consistent results...

Something for the morning I think.
 

Athan

Athan
edd said:
Hmm. Calculating the coordinate transformation between astronomical [spherical] and E: D [cartesian] is faintly more non-trivial than I thought. I'm assuming Sgr A* is in basically the right place, as it's so close to being exactly in the z direction. Having that gives a rotation from one vector to another about a specific axis - basically you have a great circle between them giving the axis and angle for that arc. Then a second star rotated about the axis you have for the vector for the first one (basically rotating the coordinate system about Sgr A*) should fix it and give you the single rotation you actually need from multiplying the two rotation matrices, but doing that for a couple of fairly simple cases and checking for the third (and handling some not unexpected rounding errors as best as one might reasonably manage) doesn't seem to give satisfyingly consistent results...

Something for the morning I think.
Click to expand...
You might want to start with this old post of mine about the forward transformation, and trying to match thing sup to what ED has. You basically need to do the inverse.
 

edd

edd
Thanks. I don't think the forward vs inverse is an issue - I'm confident in transforming astronomical systems to a Cartesian, and then it's just finding the rotation between two Cartesian systems and that's the same problem whichever way. I'm likely to try picking three reasonably spread out Hiparcos systems to try again on when I next look at it.

I also check my calculated transforms on the first two systems which work, so it's baffling when the third fails!

edit: just occurred to me I may have a handedness problem that might fix it...
 
Last edited:

madbilly

M
Alec Turner said:
If anyone's really interested in this stuff than Jackie Silver and Esvandiary figured out and described ED's galactic co-ordinate system (and how it gets translated to/from system names although not how it maps to real galactic co-ordinates) quite some time ago (Esvandiary even gave an excellent talk on it at last years Lavecon).

https://forums.frontier.co.uk/showthread.php/196297-RV-Sonnenkreis-Decoding-Universal-Cartographics
Click to expand...
I can highly recommend Jackie and Esvandiary's thread on this. It's long, but it helps to read from start to finish - most questions you have will have already been asked and answered. There's a slideshow somewhere in there as well, maybe linked from a sig.

It's complicated (at least I found it so), but when the penny drops you'll think you've been daft for not understanding it sooner!
 

goemon

goemon
edd said:
Thanks. I don't think the forward vs inverse is an issue - I'm confident in transforming astronomical systems to a Cartesian, and then it's just finding the rotation between two Cartesian systems and that's the same problem whichever way. I'm likely to try picking three reasonably spread out Hiparcos systems to try again on when I next look at it.

I also check my calculated transforms on the first two systems which work, so it's baffling when the third fails!

edit: just occurred to me I may have a handedness problem that might fix it...
Click to expand...

I'm really flattered for you trying to do this ... i have done the math from ra/dec once myself ... and i have asked several helpfull people in some other cases, but none has come up with a formula ... i hope you can make it!
 

Para Handy

Para Handy
edd said:
Hmm. Calculating the coordinate transformation between astronomical [spherical] and E: D [cartesian] is faintly more non-trivial than I thought. I'm assuming Sgr A* is in basically the right place, as it's so close to being exactly in the z direction. Having that gives a rotation from one vector to another about a specific axis - basically you have a great circle between them giving the axis and angle for that arc. Then a second star rotated about the axis you have for the vector for the first one (basically rotating the coordinate system about Sgr A*) should fix it and give you the single rotation you actually need from multiplying the two rotation matrices, but doing that for a couple of fairly simple cases and checking for the third (and handling some not unexpected rounding errors as best as one might reasonably manage) doesn't seem to give satisfyingly consistent results...

Something for the morning I think.
Click to expand...

I had enough trouble with spherical trig for my navigation courses (many moons ago {excuse the pun}) to know that I have absolutely no inclination to even think of doing this. :)
 

Faybs

Faybs
Michael Brookes said:
We don't - however the mass would eat up all of the generated systems for a large volume, bar any manual overrides.

Michael
Click to expand...

It may be a good idea to start thinking of a process to add new discoveries to the game... the computational muscle being thrown at real life galactic mapping keeps on going up, and ED still has most of its projected lifetime ahead of it. This is a game for space nerds, and people are going to keep asking for it :)
 
You must log in or register to reply here.
Back
Top Bottom