Discussion What is the most efficient way to crowdsource the 3D system coordinates

Chromatix · Oct 21, 2014

wolverine2710 said:
I Do like the statistics info. Just checked your site. Your Most isolated 10 systems list gives Hera 11.475 as most isolated. Am I correct to assume that with a ship capable of a 11.475 LY jump range ALL 571 stars can be reached? I ask this because of Snubles post of yesterday which you can find here. Don't know how he measured, manually or using a script but he writes: "The most isolated system that I found in Smackers data is Tring, with 11.325LY to nearest system WREDGUIA DJ-0 B47-1. So inside the pill you need 11.325LY jump capability to visit all "known" systems. Correct? "
Wondering what is the correct jump range to reach the complete SB1 pill?

That's entirely consistent, since I don't have any WREDGUIA DJ systems in my data slightly older data set, and presumably some closer system to Hera has also been found in the meantime.

I've just added code to explicitly probe the "island" structure of the network, rather than only reporting individual isolated stars.

On my older data, the 11.475 ly jump to Hera is indeed the defining point, with the next higher step being the 15+ ly jump to Sol (which would get filled in when the bubble next expands); the newer data will certainly be an improvement on this.

However, in that data Tring is *not* the next smaller step - there is an island of six systems which requires a large jump to reach, just short of that required to reach Hera. It's entirely probable that gap has since been bridged.

I have posted links to my code, data and map sets occasionally in this thread. I'll post them again once I've updated it all to the latest data.

ETA: in B1, the most isolated individual system was Aganippe (11.01 ly), but there were two systems together which could only be reached by a 11.064 ly jump.

TornSoul · Oct 21, 2014

RedWizzard said:
It could also have indicated that there was a problem with our assumption that everything is on a 1/32 Ly grid. However, I've spent a some time tonight looking for a case which disproves that last one thankfully.

Hmm - I'm not so sure anymore.
Would be interested in knowing what you did to "disprove" that theory.

Reason I'm not so sure anymore...

Take these (and some of the others posted)
Tun - Tring shows 83.963 (should be 83.962)
Aeolus - Sigma Bootis shows 55.168 (55.169)
LP 29-188 - LHS shows 42.819 (42.818) <--- LHS what?
Hyperion - Kamchaultultula shows 50.569 (50.570)

Assume that 1/32 is in fact not the max accuracy.
Assume 1/32 corresponds to 8 bit - Now double the bits and you get 1/8192 accuracy (not all that unreasonable really)

Now add or subtract (as the case requires) just *one* 1/8192th from the (1/32) delta between the coords we have for the above systems (and others)

And suddenly - When you do Round(calc dist;3) - You get the same result as displayed in the Galaxy map...
Ie. no complicated mult'ing rounding and multing again and rounding again etc. A simple straight forward Round(x;3) will do it.

In fact I played around - with those particular cases you can go as low as 20 bits (1/131072 accuracy) - and again add/subtract *just one* 1/131072 from the know coordinates - and the Round(calc dist; 3) matches the galaxy map distances....

So.... The sobering thought is that our assumption of the 1/32 accuracy might not be true afterall...
And Michael *has* said that they have much higher precision - At the time we presumed that was in reference to just in system objects and such (as his posted coords didn't back that up)

So... arg.... Dunno what to think really...

With just 3 digits precision on the distances there's really no easy way to get much more precision than what we already do (1/32th) - Unless we require 20-40 reference system per measurement...

wolverine2710 · Oct 21, 2014

Lets say the length of the pill is 130 LY and I have two stars which are at both edges of the pill. Y and Y coordinates are the and only the X coordinates differ. Would I then not be able to jump directly from A to B with a ship capable of a say 300 LY jump range? Or would Dijkstra and A* search shortest path algorithms don't find a direct route.

I've been playing around with trade dangerous (the navigation part) and I'm finding something I can't explain. I don't want to derail this thread so perhaps someone interested in helping me understand could hop over there and tell me what I'm doing wrong or perhaps there is an issue with TD.

TornSoul · Oct 21, 2014

wolverine2710 said:
Or would Dijkstra and A* search shortest path algorithms don't find a direct route.

They should give you the direct route.

Unless you have some option of trade profit included? (The direct route might not be the most profitable)

And then there's fuel.

If I recall (I could be wrong here) - The fuel usage of long jumps is exponentially (or something not linearly) proportional with the jump length.

Ie. 2 jumps of x lightyears (ie. x + x) might be more fuel efficient than one 2x jump in one go.

wolverine2710 · Oct 21, 2014

Chromatix said:
That's entirely consistent, since I don't have any WREDGUIA DJ systems in my data slightly older data set, and presumably some closer system to Hera has also been found in the meantime.

I've just added code to explicitly probe the "island" structure of the network, rather than only reporting individual isolated stars.

On my older data, the 11.475 ly jump to Hera is indeed the defining point, with the next higher step being the 15+ ly jump to Sol (which would get filled in when the bubble next expands); the newer data will certainly be an improvement on this.

However, in that data Tring is *not* the next smaller step - there is an island of six systems which requires a large jump to reach, just short of that required to reach Hera. It's entirely probable that gap has since been bridged.

I have posted links to my code, data and map sets occasionally in this thread. I'll post them again once I've updated it all to the latest data.

ETA: in B1, the most isolated individual system was Aganippe (11.01 ly), but there were two systems together which could only be reached by a 11.064 ly jump.

I now see where I did go wrong. Stupidly I forgot that you were using data which was a bit older. When you post the url I will inlcude it on my OP.

Something I almost wrote today but wanted to keep it a bit short - I never succeed in that. I don't know of any program out there which is able to tell me how big my island is. Lots of viper pilots after upgrading found out the hard way they are now on a 5-6 system island which they can't escape. Is your tool perhaps capable of telling me how big an island is given a certain starting point and a LY jump range?

RedWizzard today added two systems to THE reference in the git repository. Pulled it from there and wanted to post it here in csv format but its to long for a post. The repo is: https://github.com/SteveHodge/ed-systems.git. If you don't have a git client you can also download it as a zipfile.

wolverine2710 · Oct 21, 2014

TornSoul said:
They should give you the direct route.

Unless you have some option of trade profit included? (The direct route might not be the most profitable)

And then there's fuel.

If I recall (I could be wrong here) - The fuel usage of long jumps is exponentially (or something not linearly) proportional with the jump length.

Ie. 2 jumps of x lightyears (ie. x + x) might be more fuel efficient than one 2x jump in one go.

Not 100% sure but I can't recall that TD with the nav sub command takes fuel into account. Yes you are correct about fuel efficiency. I have explained it much better and with output of TD in the TD thread. Its post #362. Perhaps you could visit the thread and see what I'm doing wrong?

wolverine2710 · Oct 21, 2014

Are or aren't there 570-ish star sytems in Sb2.

Send Michael Brookes again a PM. Reason: Yesterday when I asked him about the 570 systems in the newsletter and the findings here, he wrote: "I think there's just over 600 systems in the bubble - 607 if I remember correctly."

Today RW stated that most likely the current count of 572 is correct - unless he's bitten by the nav menu bug.

So I asked Michael Brookes again. This is his response: "I don't have an easy way of checking - if someone else says 570 then I'm happy to go with that."

So we have to decide ourselves if we declare phase 1 of the discovery part done or not. That is aside from the fact 0.001 issue found here.

TornSoul · Oct 21, 2014

wolverine2710 said:
Perhaps you could visit the thread and see what I'm doing wrong?

Sorry can't help - Not familiar with the tool (python is not my thing

)

MadDogMurdock · Oct 21, 2014

Thanks guys, I've used the combined spreadsheet to populate Thrudd's tool which will help enormously with the route planning, etc.

I've suggested that Thrudd adds a 'distance error report' which adds to a list and requests users to add current distance data from where they are to the problem systems.

Build a big enough list and the coordinates can be further refined provided that we have enough reference systems that we're absolutely certain about.

Harbinger · Oct 21, 2014

A new system I happened upon, it doesn't display in the navigation tab.

Wregoe MK-Q b46-1 (-30.65625, 58.28125, -22.03125).

Distances:

Sol: 69.44
Wolf 497: 48.848
Huokang: 29.531
Demeter: 42.313
Clotti: 60.556
Fu Haiting: 53.955
San Guaralaru: 102.69
Haras: 98.417
Arabha: 91.253

TradeDangerous System.csv addition:

Code:

'Wregoe MK-Q b46-1',-30.65625,58.28125,-22.03125,'Beta2-Inferred','2014-10-21 20:03:35'

TornSoul · Oct 21, 2014

Display of ingame coordinates

I guess we'd all just as well get the coordinates of the systems displayed ingame, so we have the canonical values - instead of all this trilateration business.

Consider however how would you display those coordinates?

<systemname>
x 62.09375
y 42.71875
z 37.34375

That's just plain ugly - No UI/UX guy would ever want that.
Which might be why they aren't in there in the first place.

But how about instead doing it like this

<systemname>
x 62:3
y 42:23
z 37:11

*Much* nicer I think.
People will probably quickly learn that 31 is the max after the ":" as well (by simple observation).

Furthermore - Should the 1/32 thing not hold water, and it's instead, say 1/8192

Then

<systemname>
x 62:768
y 42:5888
z 37:2816

isn't completely unwieldy either.

wolverine2710 · Oct 21, 2014

Harbinger said:
A new system I happened upon, it doesn't display in the navigation tab.

Wregoe MK-Q b46-1 (-30.65625, 58.28125, -22.03125).

Distances:

Sol: 69.44

Wolf 497: 48.848

Huokang: 29.531

Demeter: 42.313

Clotti: 60.556

Fu Haiting: 53.955

San Guaralaru: 102.69

Haras: 98.417

Arabha: 91.253

TradeDangerous System.csv addition:

Code:

'Wregoe MK-Q b46-1',-30.65625,58.28125,-22.03125,'Beta2-Inferred','2014-10-21 20:03:35'

Cool. So we are at 573. The nav menu issue was also mentioned by RW. Would it be possible to create a ticket for it and then post the ticket ID here. That way if RW and others ticket it also they can refer to yours so things get "bundled".

Note: The OP is being changed. The urls are working, as far I have them.

JesusFreke · Oct 21, 2014

RedWizzard said:
All data in systems.json has been verified by me.

I crossed-checked the systems.json coordinates as well, at least as of.. 2 evenings ago or so. I came up with the same coordinates in all cases.

Harbinger · Oct 21, 2014

wolverine2710 said:
Cool. So we are at 573. The nav menu issue was also mentioned by RW. Would it be possible to create a ticket for it and then post the ticket ID here. That way if RW and others ticket it also they can refer to yours so things get "bundled".

Note: The OP is being changed. The urls are working, as far I have them.

Ticket #36498 created.

wolverine2710 · Oct 21, 2014

I have redesigned the OP. Its NOT ready. I guess I'm at 80-85%. I hope its an improvement over the old version. Some important parts are still missing.

I really DO appreciate any feedback - positive or negative. You can be frank, I can take it ;-) Tell me what needs to be done to make the OP better. After all its NOT my thread, its OUR project!

MadDogMurdock · Oct 21, 2014

wolverine2710 said:
I have redesigned the OP. Its NOT ready. I guess I'm at 80-85%. I hope its an improvement over the old version. Some important parts are still missing.

I really DO appreciate any feedback - positive or negative. You can be frank, I can take it ;-) Tell me what needs to be done to make the OP better. After all its NOT my thread, its OUR project!

I realize that this is primarily a tech project but its result is not.

Put it up front and central, the list so far!!

In the simple csv format for the less technically literate.

Slopey · Oct 22, 2014

Can someone post the full list (in whatever format).

Does PB3 widen the bubble?

Chromatix · Oct 22, 2014

We're not in *premium* beta any more, Slopey.

I don't know whether B3 will widen the bubble, but it's certainly possible and would be consistent with the B1-B2 transition. Perhaps FD will want to introduce us to the Sol region (where they have to do a lot more manual, rather than procedural, system construction) and/or the Empire (which would be a whole new set of ships and possibly stations).

As a completely wild guess, we might get a sphere centred on Sol.

In other news, I'm definitely missing about 60 systems based on just the distances data I have, compared to the recently quoted system counts. I *can* work with just coordinates - I would just skip the use of the 'resolve' script, which exists solely to turn distances into coordinates - but I also like to keep the primary data around.

JesusFreke · Oct 22, 2014

I was thinking about the issue with distances that we're seeing, and whether stars are actually 1/32 aligned or not. I thought it might be interesting to see how precise we can get in the typical case.

So, here's an interesting visualization of the "candidate region" for a typical well-constrained star (16 Cephei).

It's basically a sliding 2-d slice of the 3d candidate region, using a 1/8192 LY grid. The bounding box is something like 12x15x10 in 1/8192 grids. So it is approximately 5% the width of the 1/32 spacing we usually use.

The candidate region encompass a single valid point until you get down to 1/1024 precision, when there are 2 points, then 9 "correct" points at 1/2048 precision, and 70 points at 1/4096 precision.

Code:

*************
      x      
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
-------------
             
     xx      
     xxx     
     xxxx    
      xxx    
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
-------------
             
    x        
    xxx      
   xxxxxx    
   xxxxxx    
   xxxxxx    
    xxxxxx   
    xxxxxx   
       xxx   
          x  
             
             
             
             
             
             
             
             
             
             
-------------
             
  x          
  xxx        
  xxxxx      
  xxxxxxx    
  xxxxxxx    
  xxxxxxxx   
   xxxxxxx   
   xxxxxxx   
    xxxxxxx  
    xxxxxxx  
      xxxxx  
         xxx 
             
             
             
             
             
             
             
-------------
             
             
 x           
 xxxx        
 xxxxxx      
 xxxxxxxx    
 xxxxxxxxx   
 xxxxxxxxx   
  xxxxxxxx   
  xxxxxxxxx  
   xxxxxxxx  
   xxxxxxxx  
    xxxxxxxx 
    xxxxxxxx 
     xxxxxxx 
       xxxxxx
          xxx
             
             
             
-------------
             
             
             
xx           
xxxx         
 xxxxxx      
 xxxxxxxx    
 xxxxxxxxx   
  xxxxxxxx   
  xxxxxxxxx  
  xxxxxxxxx  
   xxxxxxxx  
   xxxxxxxxx 
   xxxxxxxxx 
    xxxxxxxx 
    xxxxxxxx 
    xxxxxxxx 
      xxxxxx 
        xxxx 
          xx 
-------------
             
             
             
             
xx           
 xxx         
 xxxxx       
 xxxxxxxx    
  xxxxxxxx   
  xxxxxxxxx  
  xxxxxxxxx  
   xxxxxxxx  
   xxxxxxxx  
   xxxxxxxx  
    xxxxxxx  
    xxxxxxx  
    xxxxxxx  
     xxxxxx  
     xxxxxx  
        xx   
-------------
             
             
             
             
             
 x           
 xxx         
  xxxx       
  xxxxxxx    
  xxxxxxxx   
   xxxxxxx   
   xxxxxxx   
   xxxxxxx   
   xxxxxxx   
    xxxxxx   
    xxxxx    
    xxxxx    
     xxxx    
     xxxx    
     xxxx    
-------------
             
             
             
             
             
             
             
             
    xx       
    xxxx     
   xxxxx     
   xxxxx     
   xxxxx     
    xxxx     
    xxxx     
    xxxx     
     xxx     
     xxx     
     xxx     
      xx     
-------------
             
             
             
             
             
             
             
             
             
             
             
             
             
             
             
     x       
     x       
     x       
     x       
             
-------------
*************

JesusFreke · Oct 22, 2014

Slopey said:
Can someone post the full list (in whatever format).

Does PB3 widen the bubble?

I think the agreed canonical source is RedWizzard's data, at https://github.com/SteveHodge/ed-systems. In particular, see the systems.json file.

Discussion What is the most efficient way to crowdsource the 3D system coordinates

Chromatix

TornSoul

wolverine2710

Tutorial & Guide Writer

TornSoul

wolverine2710

Tutorial & Guide Writer

wolverine2710

Tutorial & Guide Writer

wolverine2710

Tutorial & Guide Writer

TornSoul

MadDogMurdock

Harbinger

TornSoul

wolverine2710

Tutorial & Guide Writer

JesusFreke

Harbinger

wolverine2710

Tutorial & Guide Writer

MadDogMurdock

Slopey

Chromatix

JesusFreke

JesusFreke