open All Channels
seplocked Market Discussions
blankseplocked System Jump Statistics
 
This thread is older than 90 days and has been locked due to inactivity.


 
Author Topic

Droxlyn
Caldari
Posted - 2010.08.04 14:09:00 - [1]
 

Edited by: Droxlyn on 04/08/2010 16:38:52
I've been running some numbers to learn the answer to one of my long-standing Eve trivia questions. What is the longest shortest route possible?

There are three ties, but if you are in 373Z-7, you get three choices. At 99 jumps, you can go to 3KNA-N, BU-IU4, or SVB-RE.

The most "average" system, which would likely be the closest to everybody, is Kaaputenen with 26.32 average jumps from the rest of the universe.
Amarr is 6th on the list at 26.67.
Hek is 22nd with 26.88.
Jita is 29th at 27.01

Dodixie is at 27.35.
Rens is at 27.41

The most "distant" system is E7-WSY at 62 jumps on average.

Drox

Zenon Mu
Advanced Assemblies and Sciences
Posted - 2010.08.04 14:17:00 - [2]
 

Nice work!

Droxlyn
Caldari
Posted - 2010.08.04 15:00:00 - [3]
 

I just ran another query that limited it to highsec to highsec jump average (shortest routes still) and found that Amarr is first with an average of 11.98 jumps from all other high-sec systems.

Jita is 78th at 13.7 average jumps from all other high-sec systems.

Keba is the worst at 1218th place high-sec system with an average of 35.28 jumps.

Drox

Lucyna
Interstellar Killer Bee Enterprises
Posted - 2010.08.04 17:11:00 - [4]
 

Wow, nice find! Pertinent to MD, but I think this needs to be a thread on every subforum :)

Gabriel Rosencrantz
Gods of Freight
Posted - 2010.08.04 19:00:00 - [5]
 

I'm curious how you're running your numbers. If you're using route generators, not every automatically generated route is the shortest. For example, Jita to Rens will automatically put Lustrevik on your route, even though Pator is shorter. This is because the Lustrevik gate is closer than the Pator gate. However, Lustrevik is larger AU than Pator. There are many other examples of this throughout hisec.

Cherry Picker234
Posted - 2010.08.04 19:01:00 - [6]
 

I believe the longest shortest route is called the network diameter. Have you posted the rest of the data somewhere? It is interesting stuff and I would like to see what else is there.

Charles37
Posted - 2010.08.04 19:19:00 - [7]
 

It sounds like what you did was make a mathematical graph of the EVE universe from the data dump and then performed some graph theory voodoo (which I am not familiar with) on it?

As an additional consideration on what constitutes the 'shortest' route, you might also want to consider the distance between stargates. A route with fewer jumps but many 100+ AU distances might very well end up taking significantly longer then a route with more jumps but less distance between the stargates.

Cherry Picker234
Posted - 2010.08.04 19:41:00 - [8]
 

Edited by: Cherry Picker234 on 04/08/2010 19:41:16
Yes, this is generally done with either the Floyd-Warshall all-pairs shortest path algorithm, or Dijkstra's single-source shortest route algorithm for all pairs individually.

Illectroculus Defined
No Bull Ships
Posted - 2010.08.04 20:36:00 - [9]
 

Any chance you can compute the median distances for trade hubs rather than the average, or even throw up a histogram of number of systems vs number of jumps, since the distribution is not going to be a simple gaussian by any means.

Droxlyn
Caldari
Posted - 2010.08.04 20:48:00 - [10]
 

I did an A* based algorythm (basically intentionally not having a "Goal" system) and stored the results of using each system as a start system.

Calculating warp distances was something I thought about for finding the better shortest path, but then noticed that it would produce a chicken-egg situation as the distance in system y is based on which entry system x to which destination system z and x could change as its distances were figured out differently and that could impact which z and up the chain.

The data is 27 million records in a mysql database of sourcesystemID, Jumps, destinationSystemID. I wish I had taken the moment to add "FromSystemID" which would track the route and with that you could work out the original route.

I'll try a Median query in a little while.

Maybe a data dump once I make sure all of the data for N-Q5PW is intact. Stupid Windows Update automatic update...

Drox

Diomedes Calypso
Aetolian Armada
Posted - 2010.08.04 21:24:00 - [11]
 

Hey thanks.. I was curious about the question..
Not any one approach to answering it is the end-all but its certainly a valid way of considering it.

I'd be curious to see a little more of the distribution . You named the hubs and they seem to come pretty darn close to the best, less tha one jump difference.

If you were to drop to #100, #200, ....I'd like to see those answers...and while you're at it, the number of systems.

It would be interesting to me if you could crunch the numbers again eliminating all 0.0 systems. Jump bridges can change those numbers drastically and well, no need to justify it but perhaps the hub location is set more by empire activity (maybe-maybe not, but still it would be interesting to consider)

If you did do the empire data and this sort of thing didn't take you tons of time, it would be intesting to do the numbers again with "safest" autopilot setting with maybe a 10 or 15 jump threshold. Distance in high sec jumps is probably the most perninent data for traders.


Thoraemond
Minmatar
Far Ranger
Posted - 2010.08.05 00:22:00 - [12]
 

Originally by: Droxlyn
Jita is 29th at 27.01

Some relevant thoughts in this thread from a couple of years ago: Average distance between stargates in high sec? In that thread, Matalino agrees with your 27.01 for Jita.

I was thinking that there had been an even longer discussion at some point, but a couple of minutes with Google didn't lead me to it. Any other links back would be welcomed.



 

This thread is older than 90 days and has been locked due to inactivity.


 


The new forums are live

Please adjust your bookmarks to https://forums.eveonline.com

These forums are archived and read-only