Newtonian Aurora - Galactic Map Question

[PTTG]

Since there's no evidence that there actually are fewer red dwarfs, doesn't it seem reasonable to add presumed stars? That isn't to say they'd be random- at least not in the sense of changing in every game- just that they are stars we expect to be there that we haven't proven.

It just seems to me that having a uniform set of realistic stars is more "real" than having only a few proven stars that are proven to exist.

[Steve Walmsley (OP)]

Since there's no evidence that there actually are fewer red dwarfs, doesn't it seem reasonable to add presumed stars? That isn't to say they'd be random- at least not in the sense of changing in every game- just that they are stars we expect to be there that we haven't proven.

It just seems to me that having a uniform set of realistic stars is more "real" than having only a few proven stars that are proven to exist.

Yes, that is a good point. I will also look at finding more up to date information. The Hipparchus data is from 1998 so more red dwarfs stars will have been found since then. Unfortunately, it is a major task identifying them from many different sources whereas the Hipparchus data for 120,000 stars was available in an Access database.

Steve

[fcharton]

Hi,

The distribution of stars remains constant because although there are more stars as you move further out, they are spread over a wider area.    This isn't a guess - this is based on generating the maps.   

But wouldn't the clustering problem remain the same? Real stars at distance R belong to a sphere with surface 4 PI R^2.  On a 2D map, they get drawn over a circle 2 PI R long.  If you double the distance, the surface of the sphere is multiplied by four, but all systems at this distance will be drawn over a circle only twice as long.  Unless the density of stars falls drastically as one gets away from the "central system", you will have clustering.

This has two consequences.    If star density is constant, the further you move away from earth, the more clustered the stars should appear on the 2D map (note this might be attenuated when using real stars, because star catalogues are segregated by magnitude: as you move away from Earth, smaller systems don’t get listed).    Also, 2D projection greatly reduces the distance between two systems chosen at random.  In my opinion, this is the main flaw in your “string method”.  You are keeping the correct distances to Sol, but shrinking all the rest.   

There is no simple solution: 3D space is just much larger than 2D space.   

How are you modeling the actual positions of stars, in 2D or 3D (meaning you have bearing and elevation, or just bearing)? If your actual data model is 2D, I would suggest the following approach for real stars generation:

Generate random stars in 2D, from a density map (something that measures the average number of systems in every area of the map).    Then pick one as Sol, sort all others by distance to it, and name them from a real star catalogue.    The closest star would be Proxima, and then Alpha, Barnard, and so on.    The resulting map would have the right stars at plausible distances (for the nearest ones, at least), and would not suffer from “projection clutter”.   

Francois
(Be kind on me please, first post….   )

[Steve Walmsley (OP)]

Hi,

But wouldn't the clustering problem remain the same? Real stars at distance R belong to a sphere with surface 4 PI R^2.  On a 2D map, they get drawn over a circle 2 PI R long.  If you double the distance, the surface of the sphere is multiplied by four, but all systems at this distance will be drawn over a circle only twice as long.  Unless the density of stars falls drastically as one gets away from the "central system", you will have clustering.

This has two consequences.    If star density is constant, the further you move away from earth, the more clustered the stars should appear on the 2D map (note this might be attenuated when using real stars, because star catalogues are segregated by magnitude: as you move away from Earth, smaller systems don’t get listed).    Also, 2D projection greatly reduces the distance between two systems chosen at random.  In my opinion, this is the main flaw in your “string method”.  You are keeping the correct distances to Sol, but shrinking all the rest.   

There is no simple solution: 3D space is just much larger than 2D space.   

How are you modeling the actual positions of stars, in 2D or 3D (meaning you have bearing and elevation, or just bearing)? If your actual data model is 2D, I would suggest the following approach for real stars generation:

Generate random stars in 2D, from a density map (something that measures the average number of systems in every area of the map).    Then pick one as Sol, sort all others by distance to it, and name them from a real star catalogue.    The closest star would be Proxima, and then Alpha, Barnard, and so on.    The resulting map would have the right stars at plausible distances (for the nearest ones, at least), and would not suffer from “projection clutter”.   

Francois
(Be kind on me please, first post….   )

Its a good point about the surface of the sphere vs the circumference of the circle. Also a good point that as you move away from Earth, smaller systems don’t get listed. Looking at the map I have generated for the nearest 3000 stars for my test campaign, those two factors seem to balance out fairly well, leaving a much denser distribution of systems than a random map but with no greater density at greater distances. Besides, 'real stars' systems tend to produce less habitable worlds than random systems so the actual desnsity of useful systems may not vary much between the two options.

I wouldn't be keen on random systems with the only element of 'real stars' being the idea of naming them based on order of proximity. That is almost the same as random stars. The whole point of real stars is using the actual mass and spectral class of each star and having them in the correct xy direction.

I am pretty happy with what I have ended up with at the moment, at least enough to use it for playtesting. As I get into the test campaign, I'll be able to get a better idea of the playability of the mapping system.

Steve

[jseah]

I wouldn't be keen on random systems with the only element of 'real stars' being the idea of naming them based on order of proximity. That is almost the same as random stars. The whole point of real stars is using the actual mass and spectral class of each star and having them in the correct xy direction.

Can we have this as a random map option?  Please?  =)

[Steve Walmsley (OP)]

Can we have this as a random map option?  Please?  =)

You mean an option to take the list of system names from the real stars table in order of distance from Sol and apply them to the systems in a random map games in their order of distance from Sol?

Steve

[fcharton]

I wouldn't be keen on random systems with the only element of 'real stars' being the idea of naming them based on order of proximity.    That is almost the same as random stars.    The whole point of real stars is using the actual mass and spectral class of each star and having them in the correct xy direction.   

You can do better than that: just randomize positions, and assign real names and properties according to distance.    This way, you get plausible distances, correct class, mass and names.   

I believe you could even change the bearings to their correct value (in your 2D projection system, that is).  Such random corrections would not change overall density.   

To summarize, the procedure would be :
1- generate a random set of nameless, massless and classless stars, according to some prior density (controlled by game parameters)
2- sort the systems according to their distance to the sun, and associate each one with a real star, thus determining name, class and mass
3- change the bearing of those stars, to match those of their real namesakes (therefore maintaining the constellations in order, a good thing since many stars are named from their constellation)

I haven't done the maths, but I believe that so long you don't change the distances from step one, you remain in line with the original density.   

Francois

[jseah]

You mean an option to take the list of system names from the real stars table in order of distance from Sol and apply them to the systems in a random map games in their order of distance from Sol?

Steve

Yeah, that. 

A bit like a random map, but without crazy names. 

[Ominous]

You can do better than that: just randomize positions, and assign real names and properties according to distance.     This way, you get plausible distances, correct class, mass and names.    

I believe you could even change the bearings to their correct value (in your 2D projection system, that is).   Such random corrections would not change overall density.    

To summarize, the procedure would be :
1- generate a random set of nameless, massless and classless stars, according to some prior density (controlled by game parameters)
2- sort the systems according to their distance to the sun, and associate each one with a real star, thus determining name, class and mass
3- change the bearing of those stars, to match those of their real namesakes (therefore maintaining the constellations in order, a good thing since many stars are named from their constellation)

I haven't done the maths, but I believe that so long you don't change the distances from step one, you remain in line with the original density.    

Francois

I like this idea.   While players will be able to tell whether a system has the name of a real star, the correct number of stars, and so forth, few will be able to tell whether the stars are placed accurately on a grid.   I don't see it taxing the suspension of disbelief for the vast majority of players.

[Steve Walmsley (OP)]

You can do better than that: just randomize positions, and assign real names and properties according to distance.    This way, you get plausible distances, correct class, mass and names.   

I believe you could even change the bearings to their correct value (in your 2D projection system, that is).  Such random corrections would not change overall density.   

To summarize, the procedure would be :
1- generate a random set of nameless, massless and classless stars, according to some prior density (controlled by game parameters)
2- sort the systems according to their distance to the sun, and associate each one with a real star, thus determining name, class and mass
3- change the bearing of those stars, to match those of their real namesakes (therefore maintaining the constellations in order, a good thing since many stars are named from their constellation)

I haven't done the maths, but I believe that so long you don't change the distances from step one, you remain in line with the original density.   

Francois

OK, I am probably missing something here but if I generate a random map and then assign names/star types based on distance and change the bearing to match the real one - isn't that a more complex method of achieving what I already have?

Steve

[UnLimiTeD]

No, because the range from Sol will be slightly random, and some stars might be missing if the random density doesn't equal the actual density.

[fcharton]

if I generate a random map and then assign names/star types based on distance and change the bearing to match the real one - isn't that a more complex method of achieving what I already have?

Not quite.   Your method will keep correct bearing, type, and distance, while projecting a 3D sphere onto a 2D disk.   This will increase star density by a factor proportional to the distance of the sun.  It will also make systems (save Sol) closer to each other (due to the absent third dimension). 

Mine keeps the type and bearing, but allows the distance to vary from the actual value (only keeping the systems sorted with respect to their distance to Sol).   This way, the clustering caused by the move from 3D to 2D, and the resulting asymetry of star density (ie its increase as one moves away from Sol), are corrected.   You can also adjust the density to keep distances between stars more realistic (although I don't think you can completely eliminate the "shrinking effect" due to the projection). 

Note, though, that the clustering will probably be reduced if some the less bright stars are absent from the catalogue you are using.  But this means that such systems should be added, or the star type distribution will be wrong (ie biased in favour of the larger/more luminous systems).

Francois

[Steve Walmsley (OP)]

I think given the option between:

a) Accurate distances from Sol
b) Accurate direction from Sol
c) Higher stellar density than reality
d) Incorrect distances between stars

or

a) Order of distance correct but not absolute distance
b) Accurate direction
c) Realistic stellar density
d) Incorrect distances between stars, although a more realistic average distance.

My preference is the former, although of course this is based on my personal bias of which factors are more important. I tend to think in terms of what the average player would accept in terms of the suspension of disbelief and also how I would mention star systems in fiction. The main one for me is accurate distance followed by accurate direction, so this factor drives the method I use to generate the map. The vast majority of players wouldn't know if stellar density or distances between stars (other than Sol) is correct.

However, it may be possible to generate the alternative as well. I'll have to look at how easily it would fit into the galaxy generation.

Steve

[fcharton]

I see your point.   The one thing I'm wondering about is the effect the change in stellar density may have on game play.   My feeling is that the 3D to 2D projection, where you 'fold' all stars over and below onto the horizon, will reduce average distances between systems*, except distances to Sol, which are kept at correct 3D values.  Also, this reduction will not take place in a homogeneous way.   The further you are from Sol, the more "shrinkage" you get.   

In other words, the problem is less that density or distances change, but that the universe becomes "centred" around Sol, which would be further away from other systems (because it is the only star with "unshrunk" distances).   In contrast, your typical NPR race, which begins the game far from Sol would almost always live in a denser area of the galaxy.  This might create some imbalance, no?

But then, this certainly can be tested, by generating a large "real star" universe, and calculating some indicator like 'average distance to the nearest N stars', for a significant number of stars, and plotting it against distance to Sol. 

Francois

* this is on average, of course, for two systems very close to each other right at the "top" of the sky, distance can actually be vastly increased

[Steve Walmsley (OP)]

I see your point.   The one thing I'm wondering about is the effect the change in stellar density may have on game play.   My feeling is that the 3D to 2D projection, where you 'fold' all stars over and below onto the horizon, will reduce average distances between systems*, except distances to Sol, which are kept at correct 3D values.  Also, this reduction will not take place in a homogeneous way.   The further you are from Sol, the more "shrinkage" you get.   

In other words, the problem is less that density or distances change, but that the universe becomes "centred" around Sol, which would be further away from other systems (because it is the only star with "unshrunk" distances).   In contrast, your typical NPR race, which begins the game far from Sol would almost always live in a denser area of the galaxy.  This might create some imbalance, no?

But then, this certainly can be tested, by generating a large "real star" universe, and calculating some indicator like 'average distance to the nearest N stars', for a significant number of stars, and plotting it against distance to Sol. 

* this is on average, of course, for two systems very close to each other right at the "top" of the sky, distance can actually be vastly increased

I am playing a 3000 system real stars game at the moment for test purposes. Because we can't detect smaller stars at greater distances, the density remains fairly constant across the 100 LY radius. The issue becomes the decreasing number of red dwarf stars as you move outwards, although I still don't really know how significant an issue that will be long term

Steve