Index stat for map sizes

G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

At most hostsites there's a pool of maps to choose from for playing.
Usually they are described with the planet count and the rough size of
the galaxy (given in coordinate cornerpoints).
I would like another number or two added to the map descriptions for a
better overview. Something like planet density. To get an idea how far
I have to travel to get from planet to planet. I believe this to be a
very important statistic (beside size and planet count) to judge the
map on its playability, and if warp or hyp engines will be of advantage
(generally speaking).

It could be something simple like planet density:

p=P/A

p: planet density
P: number of Planets
A: area of galaxy

With area as simple as (length)^2 for squares or pi*(radius)^2 for
circles.

p^(-1) would be something like 'space around each planet', and you
could derive a mean distance d between planets from it:

d=2*sqr((pi*p)^(-1))

I know of course that such a description by numbers would be very
general. Some maps have a varying density, some have clusters, and so
forth. But it would at least give a general idea and a better feeling
for a map you have not seen before. And a look at the map (and its
planet distribution) will provide a quick evaluation of the credibility
of the statistics p and d.

To give a few examples:

Echo Cluster (500 planets, 2000 x 2000 square)
p=1.25*10^(-4)
d=101

Blitzkrieg (220 planets, 400 radius circle)
p=4.38*10^(-4)
d=54

YaleLite (252 planets, 2000 radius circle)
p=0.2*10^(-4)
d=252

I think these examples already give a feeling for each (very different)
map, and if you'd want to play a non-hyp race in it. Something I
don't easily get just from a look at the coordinate points or even a
picture (without scale) of the map.

What I want to know now is, would there be enough of a popular demand
for such stats like 'planet density' and 'mean distance' to
convince the map creators and the hostsite administrators to add these
stats to the map descriptions?
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

I'd love to see this.

Greg Bahr

P.S. This is a result of our discussion isn't it?
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

"Lord Owl" <lord.owl@gmx.de> schrieb im Newsbeitrag news:1112362545.660292.48050@l41g2000cwc.googlegroups.com...

> forth. But it would at least give a general idea and a better feeling
> for a map you have not seen before. And a look at the map (and its

Extend the description and give minimum, maximum, mean and volatility for each map.

GFM GToeroe
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Owl

Could you state your formulae in more exact maths? If the group could
agree a programmatic formula then this could be coded from a cut-down
version of map2ps fairly swiftly. Site admins could then run this over
..MAP files when they install them. New maps would be easily handled.
Indeed maybe I could make map2ps throw off some metrics as a side effect.

Lord Owl wrote:
> At most hostsites there's a pool of maps to choose from for playing.
> Usually they are described with the planet count and the rough size of
> the galaxy (given in coordinate cornerpoints).
> I would like another number or two added to the map descriptions for a
> better overview. Something like planet density. To get an idea how far
> I have to travel to get from planet to planet. I believe this to be a
> very important statistic (beside size and planet count) to judge the
> map on its playability, and if warp or hyp engines will be of advantage
> (generally speaking).
>
> It could be something simple like planet density:
>
> p=P/A
>
> p: planet density
> P: number of Planets
> A: area of galaxy
>
> With area as simple as (length)^2 for squares or pi*(radius)^2 for
> circles.
>
> p^(-1) would be something like 'space around each planet', and you
> could derive a mean distance d between planets from it:
>
> d=2*sqr((pi*p)^(-1))
>
> I know of course that such a description by numbers would be very
> general. Some maps have a varying density, some have clusters, and so
> forth. But it would at least give a general idea and a better feeling
> for a map you have not seen before. And a look at the map (and its
> planet distribution) will provide a quick evaluation of the credibility
> of the statistics p and d.
>
> To give a few examples:
>
> Echo Cluster (500 planets, 2000 x 2000 square)
> p=1.25*10^(-4)
> d=101
>
> Blitzkrieg (220 planets, 400 radius circle)
> p=4.38*10^(-4)
> d=54
>
> YaleLite (252 planets, 2000 radius circle)
> p=0.2*10^(-4)
> d=252
>
> I think these examples already give a feeling for each (very different)
> map, and if you'd want to play a non-hyp race in it. Something I
> don't easily get just from a look at the coordinate points or even a
> picture (without scale) of the map.
>
> What I want to know now is, would there be enough of a popular demand
> for such stats like 'planet density' and 'mean distance' to
> convince the map creators and the hostsite administrators to add these
> stats to the map descriptions?
>
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Lord Owl asks

>What I want to know now is, would there be enough of a popular demand
>for such stats like 'planet density' and 'mean distance' to
>convince the map creators and the hostsite administrators to add these
>stats to the map descriptions?

Mean distance would be interesting to me. Because I would understand it.
Mentioning this is a good idea.

Density would not be meaningful to me. It is not a very intuitive
concept to convey for maps - to me, anyway. It would be a number which I
would ignore.
--
Paul Honigmann
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

The planet density is quite straightforward. Determine the area of the
map by the given coordinates of its cornerpoints, i.e. A=(length)^2.
Or, if it is a circle, take half a side as radius R and determine the
area with pi*R^2. Divide the planet number through the total area, and
you got some kind of planet density.
As Paul mentioned below it is something of an abstract value.

But if you take its reciprocal you get something like average space per
planet. Imagine a circular area of space of that amount, with the
planet in its center. Average distance from planet to planet would be
twice the radius r of such a 'planet-space' circle. To get the
radius, solve a=pi*r^2.

A: Area of map
P: number of planets
a: space per planet
d: average distance between planets

a=A/P, a=pi*r^2

d=2r=2*sqr(A/(pi*P))

I am not sure how close this value approximates the real average
distance between neighbouring planets on a map. Mathematically it
appears to be sound. Depends a lot on the individual map and its
homogenity. At the very least it gives you a rough idea ;).
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Mean distance is more intuitive for me also. One could take a given
planet and calculate the mean distance to all other planets. This could
approximate to something that tells me how a given race would fare in
running a transport network.

Map forms do matter though. A clustered map would behave very
differently than one blob (like Echo). The two cluster extended Yale
would get a weird metric.

I can only think to produce a set of mini values based on local mean
distances using randomly selected planets. Then take the mean and
std-deviation of the set of figures.

KlingonKommand wrote:
> Lord Owl asks
>
>> What I want to know now is, would there be enough of a popular demand
>> for such stats like 'planet density' and 'mean distance' to
>> convince the map creators and the hostsite administrators to add these
>> stats to the map descriptions?
>
>
> Mean distance would be interesting to me. Because I would understand it.
> Mentioning this is a good idea.
>
> Density would not be meaningful to me. It is not a very intuitive
> concept to convey for maps - to me, anyway. It would be a number which I
> would ignore.
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

I'm not playing VGAP much any more, but this is such an interesting
problem to work on that I couldn't resist :)

I came up with two statistics that I think do a pretty good job of
describing:
1) How close planets are on average
2) How big the area is between clusters of planets

I call these measurements the "Median Distance Index" (measured in
lightyears) and the "Boonies Index" (how likely you are to end up "out
in the boonies").

Before explaining how I arrived at these numbers, take a look at my
results (if this doesn't show up formatted correctly, try using a
fixed-width font. Google's newsgroup reader has a link near the top to
switch to Fixed-width):

Map Stars Med Dist Boonies Index
============================================
Yale 1009 67.36 3.24
Echo 500 60.90 0.17
Epsilon 688 44.18 11.30
Stardrift 592 44.15 3.58
dblossom 400 38.21 2.36
Snowflake 40 100.00 3.76
Yalelite 252 139.32 2.55
QuadraStar 2001 40.01 1.44
Starburst 420 55.47 36.25
Phi 500 50.01 3.31

To compute these figures, the map file is loaded and then a Minimum
Spanning Tree is constructed from the starmap. The "Median Distance
Index" is defined as the median edge length of the tree. The "Boonies
Index" is defined as the difference between the mean and the median
edge length.

The "median Distance Index" tells you about how far apart, in
lightyears, you can expect planets to be.

A high "Boonies Index" indicates that there are groups of stars which
are relatively isolated. For example, the Starburst galaxy - which
consists of several clusters of planets separated by a large distance -
has the highest Boonies Index.

Of course, to compute how "warp-friendly" any map is, you'll also need
to take into account the number of stars, and the overall size... but
that information is readily available.
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Wow! That's the difference you get when someone experienced in
statistics and programming takes care of a problem :). Very elegant
approach. I dimly remember something vaguely like it from a statistical
mechanics lecture. Love the pictures :).

I notice that your average distance is only about 60% of the amount I
get with my own rather simple method. Hm. I believe the Minimum
Spanning Tree appears to compute more something like 'average
distance to the closest planet' than 'average distance to all
neighbouring planets'. At least in tendency.

Concerning the Boonies Index, what do you understand under "the
difference between the mean and the median edge length"? It's
probably a language thing. I assume 'mean edge length' is the
average length of a single tree line (make the sum and divide it
through number of lines), while 'median edge length' is computed
by, hm, sorting out the extreme ones until you are left with the last
one right in the middle?
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Roger writes
>Visual aid now available. See pictures of the Minimum Spanning Trees
>for various galaxies:
>
>http://webpages.charter.net/kgnorris/spanningtrees.png

Very pretty!
--
Paul Honigmann
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

Yes, your understanding of mean and median is correct.

I did a little more number crunching based on some other types of
trees, to see if those numbers are helpful in any way. I'm not going
to try to format these.... import it into Excel if you want to view
them properly :)

Map Info,,Minimum Spanning Tree,,,Maximum Spanning Tree,,,Lattice,,,
Map,Stars,Median,Mean,Diff,Median,Mean,Diff,Median,Mean,Diff
dblossom,400,38.21,40.57,2.36,102.84,122.74,19.90,66.24,80.72,14.48
Phi,500,50.01,53.32,3.31,106.94,124.36,17.42,77.32,88.47,11.15
Epsilon,688,44.18,55.48,11.30,102.18,165.30,63.12,70.18,108.63,38.45
Stardrift,592,44.15,47.73,3.58,137.59,170.02,32.43,75.29,107.91,32.62
Echo,500,60.90,61.07,0.17,142.47,165.39,22.92,101.08,112.42,11.34
Yale,1009,67.36,70.60,3.24,160.12,175.96,15.84,111.51,123.34,11.83
Starburst,420,55.47,91.72,36.25,383.44,438.23,54.79,115.26,234.69,119.43
Snowflake,40,100.00,103.76,3.76,202.00,231.39,29.39,158.11,164.58,6.47
Yalelite,252,139.32,141.87,2.55,340.65,357.90,17.25,219.64,248.65,29.01

"Minimum Spanning Tree" - Already described.

"Maximum Spanning Tree" - I don't think this is a real term... but it's
basically the opposite of the minimum spanning tree. Only edges
between neighboring stars are eligible.

"Lattice" - Connects all stars with their neighbors. Stars are
considered to be "neighbors" if the line between them cannot be cut by
a shorter line between other stars.

I may post additional visual aids when I get the chance.
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Roger, well done. This is just what was asking for. Someone with
original ideas and maths skill to write them up cogently. The number
crunching is a bonus and your numbers give good quick appreciation of
map character. Comparing my own Yale and Yalelite we see the increase in
median distance in the thinned out map (making it better for the EE) yet
the Boonies index is of the same order of magnitude, indicating the
underlying map stats. Echo with Boonies 0.17 is obviously a totally
"urban" map.

I think you have the last word on this. Will you post a list of all maps
at Drew's for inclusion on his site?

Everyone else, just stand back in awe for a few minutes.

Roger wrote:
> I'm not playing VGAP much any more, but this is such an interesting
> problem to work on that I couldn't resist :)
>
> I came up with two statistics that I think do a pretty good job of
> describing:
> 1) How close planets are on average
> 2) How big the area is between clusters of planets
>
> I call these measurements the "Median Distance Index" (measured in
> lightyears) and the "Boonies Index" (how likely you are to end up "out
> in the boonies").
>
> Before explaining how I arrived at these numbers, take a look at my
> results (if this doesn't show up formatted correctly, try using a
> fixed-width font. Google's newsgroup reader has a link near the top to
> switch to Fixed-width):
>
> Map Stars Med Dist Boonies Index
> ============================================
> Yale 1009 67.36 3.24
> Echo 500 60.90 0.17
> Epsilon 688 44.18 11.30
> Stardrift 592 44.15 3.58
> dblossom 400 38.21 2.36
> Snowflake 40 100.00 3.76
> Yalelite 252 139.32 2.55
> QuadraStar 2001 40.01 1.44
> Starburst 420 55.47 36.25
> Phi 500 50.01 3.31
>
> To compute these figures, the map file is loaded and then a Minimum
> Spanning Tree is constructed from the starmap. The "Median Distance
> Index" is defined as the median edge length of the tree. The "Boonies
> Index" is defined as the difference between the mean and the median
> edge length.
>
> The "median Distance Index" tells you about how far apart, in
> lightyears, you can expect planets to be.
>
> A high "Boonies Index" indicates that there are groups of stars which
> are relatively isolated. For example, the Starburst galaxy - which
> consists of several clusters of planets separated by a large distance -
> has the highest Boonies Index.
>
> Of course, to compute how "warp-friendly" any map is, you'll also need
> to take into account the number of stars, and the overall size... but
> that information is readily available.
>
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Roger writes

>400,38.21,40.57,2.36,102.84,122.74,19.90,66.24,80.72,14.48
>Phi,500,50.01,53.32,3.31,106.94,124.36,17.42,77.32,88.47,11.15
>Epsilon,688,44.18,55.48,11.30,102.18,165.30,63.12,70.18,108.63,38.45
>Stardrift,592,44.15,47.73,3.58,137.59,170.02,32.43,75.29,107.91,32.62
>Echo,500,60.90,61.07,0.17,142.47,165.39,22.92,101.08,112.42,11.34
>Yale,1009,67.36,70.60,3.24,160.12,175.96,15.84,111.51,123.34,11.83
>Starburst,420,55.47,91.72,36.25,383.44,438.23,54.79,115.26,234.69,119.43
>Snowflake,40,100.00,103.76,3.76,202.00,231.39,29.39,158.11,164.58,6.47
>Yalelite,252,139.32,141.87,2.55,340.65,357.90,17.25,219.64,248.65,29.01

Clear, concise, and self-explanatory. It reminds me of Jon Nunn's
classic "111011" post or Peter Chamber's proof that 2 = 1 on Tuesdays.
Of course in my day we didn't have 1's, we just had zeroes. We need more
postings like this. Perhaps we can change the name of the NG to
alt.games.vgaplanets4.binaries ...
--
Paul Honigmann
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

I'll see what I can do. It'll probably be a few days before I can
scrounge up enough free time to get to all the maps, but I'll post the
link when I finish.
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

The final results are in!

I didn't actually get to *all* of the maps on Drewhead's, but I think I
covered most of them. And after further analysis, I decided that
Minimum Spanning Trees, while visually appealing, really aren't all
that useful :)

Read the full report (follow map links for pictures) here:
http://webpages.charter.net/kgnorris/maps/mapinfo.htm

I decided on a different method for calculating the Boonies Index.
First of all, I'm just using the basic lattice instead of the minimum
spanning tree. (i.e., we're considering all connections between each
planet and its neighbors). The Boonies Index is now defined as the
average deviation divided by the median.

Based on the median and the Boonies Index, I determined an overall
"Warp Unfriendliness" factor. Should be fairly self-explanatory on the
results page.

Hope this info is useful!
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Dear newsgroup.

As L.O. started this thread I was anxious to see the results as it
could mean that the world of p4 maps finally would get some
"clarifying".

I did the nOob test with Rogers table today and guess what happend!

I tested it with two objects. Both beginners to intermediate (1-2 games
played). BOTH of them had absolutley no clue at all what this site was
telling them.

--> idea: great
--> concept: great
--> result: not usefull at all!

Fabian
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

:cool:

Ok - good point!

So the facts are there we just need to find a way to show 'em in a
manner that is easily accessible (even for new players).

Fabian
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

Some simple descriptions for newbies (I'll add this to the site when I
get the chance):

First of all, newbies will probably only care about the first five
columns or so. Everything after that is for the engineer-types who
like to stare at pages full of numbers. So, here are those first five
columns:

1) Map Name
Self explanatory

2) Stars
The number of stars in the map.

3) Overall Warp Unfriendliness
This number attempts to describe how warp-drive races will fare on the
map. It's really not useful for anything accept comparing two maps to
see which one is more warp-friendly.

For example, we can see from the list that Troubadour appears to be the
friendliest map for warp-drive races.

A more technical explanation is that this number takes into account
both the median distance between stars and their neighbors, as well as
whether stars are evenly distributed across the map.

4) Boonies Index
This number describes whether stars are evenly distributed or not. A
high "Boonies Index" indicates that you're more likely to end up "out
in the boonies." More specifically, maps with a high Boonies Index
will tend to have dense clusters of stars separated by large gulfs of
empty space.

A map with all the stars evenly distributed, so that each star is about
the same distance from all of its neighbors, would have a very low
Boonies Index.

In our chart, anything with a Boonies Index larger than about 0.7 is
going to have fairly major "clustering" of stars, or large gulfs of
empty space.

5) Lattice Median (measured in lightyears)
The lattice median is just the median distance beween neighboring
stars. The median, for those who aren't statistics buffs, is just the
middle value. For example, a lattice median of 120 indicates that most
stars are about 120 lightyears away from their neighbors.

Obviously, this is a very important value to consider when choosing a
map, since it tells you how difficult it will be for warp-drive ships
to get to their neighboring stars.
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

Just wanted to say that I appreciate your work very much! Please
proceed!!

Fabian
 

Roger

Distinguished
Dec 31, 2007
743
0
18,980
Archived from groups: alt.games.vgaplanets4 (More info?)

> We see the Boonies index for these two is 0.49 and 0.50. Is > this
what we expect?

Yes. The Boonies Index only takes into account the evenness of the
distribution. Yale and Yalelite have basically the same distribution,
so the number is virtually the same. As you said, even though Yalelite
has a lower planet density, the "clustering" is about the same. So...
perhaps the Boonies Index should be renamed to the Cluster Index or
something like that?

A better measurement of planet density is the Median calculation, which
measures the median distance between neighboring stars.

Only the the "Overall Warp Unfriendliness" number takes into account
both density and clustering. And, if you look at the results, Yale
scores considerably better in that area than Yalelite, due to its
greater planet density.
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

All that is needed is a short explanation.

This great-idea-but-I-am-too-dumb-to-learn-to-understand-it-attitude is
nothing for VGAP players anyway.

Lordfire


"Fabian" <fabs42@yahoo.com> wrote in message
news:1114081577.411415.67770@z14g2000cwz.googlegroups.com...
>
> Dear newsgroup.
>
> As L.O. started this thread I was anxious to see the results as it
> could mean that the world of p4 maps finally would get some
> "clarifying".
>
> I did the nOob test with Rogers table today and guess what happend!
>
> I tested it with two objects. Both beginners to intermediate (1-2 games
> played). BOTH of them had absolutley no clue at all what this site was
> telling them.
>
> --> idea: great
> --> concept: great
> --> result: not usefull at all!
>
> Fabian
>
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

An interesting test of Boonies indexis the pair { Yale, Yalelight }.

Yalelight is a sub-set of 25% of the stars in Yale, randomly chosen.
So the overall stats of the stars distribution should only differ in
terms of density. Not angular distribution or the like.

We see the Boonies index for these two is 0.49 and 0.50. Is this what we
expect? Stars in Yalelite are certainly in a lower density cluster. One
is in sense in a more rural locale. However the 'clustering' is
certainly about the same.

Perhaps the term 'Boonies Index' is merely somewhat mis-named?

Roger wrote:
> The final results are in!
>
> I didn't actually get to *all* of the maps on Drewhead's, but I think I
> covered most of them. And after further analysis, I decided that
> Minimum Spanning Trees, while visually appealing, really aren't all
> that useful :)
>
> Read the full report (follow map links for pictures) here:
> http://webpages.charter.net/kgnorris/maps/mapinfo.htm
>
> I decided on a different method for calculating the Boonies Index.
> First of all, I'm just using the basic lattice instead of the minimum
> spanning tree. (i.e., we're considering all connections between each
> planet and its neighbors). The Boonies Index is now defined as the
> average deviation divided by the median.
>
> Based on the median and the Boonies Index, I determined an overall
> "Warp Unfriendliness" factor. Should be fairly self-explanatory on the
> results page.
>
> Hope this info is useful!
>
 
G

Guest

Guest
Archived from groups: alt.games.vgaplanets4 (More info?)

"Clustering Index" I would support.

Roger wrote:
>>We see the Boonies index for these two is 0.49 and 0.50. Is > this
>
> what we expect?
>
> Yes. The Boonies Index only takes into account the evenness of the
> distribution. Yale and Yalelite have basically the same distribution,
> so the number is virtually the same. As you said, even though Yalelite
> has a lower planet density, the "clustering" is about the same. So...
> perhaps the Boonies Index should be renamed to the Cluster Index or
> something like that?
>
> A better measurement of planet density is the Median calculation, which
> measures the median distance between neighboring stars.
>
> Only the the "Overall Warp Unfriendliness" number takes into account
> both density and clustering. And, if you look at the results, Yale
> scores considerably better in that area than Yalelite, due to its
> greater planet density.
>
 

Latest posts