Stearns algorithm

Debates and discussions on the various race scheduling methods that can be used and their fairness and accuracy in determining the winners.
Post Reply
heatsink
Apprentice
Apprentice
Posts: 5
Joined: Tue Sep 16, 2014 2:00 am
Location: Christchurch, New Zealand

Stearns algorithm

Post by heatsink » Sat Sep 20, 2014 4:14 am

Is the algorithm for creating a Stearns rotation published anywhere? I've come across many links to a .exe that apparently generates a Stearns chart but all of them point to sites that don't exist anymore. Thanks.



User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6888
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Stearns algorithm

Post by Stan Pope » Sat Sep 20, 2014 9:40 am

heatsink wrote:Is the algorithm for creating a Stearns rotation published anywhere? I've come across many links to a .exe that apparently generates a Stearns chart but all of them point to sites that don't exist anymore. Thanks.
The source for the Stearns .exe was not available 20 years ago (or thereabouts) when I was developing accuracy comparisons of various methods. My request for source and/or more detailed info to the author of the program (not Stearns, but a lady who was distributing the .exe) was not well received.

Someone should have a copy of the .exe lying about on an old disk drive ... I might even have one as I tend to keep the old HDD from my now obsolete computers! If your request hasn't been satisfied by someone in a few days, perhaps I will be able to plug in those old HDD's and find it for you. Too many pending obligations at the moment to just rush off and do it now.


Stan
"If it's not for the boys, it's for the birds!"

heatsink
Apprentice
Apprentice
Posts: 5
Joined: Tue Sep 16, 2014 2:00 am
Location: Christchurch, New Zealand

Re: Stearns algorithm

Post by heatsink » Sat Sep 20, 2014 4:27 pm

After some further searching today I found the Stearns executable here http://www.wtrfrd.com/pack339/pinewood.exe. However it appears to be a 16bit application (to be expected I suppose given its age) and so won't run on my 64bit machine.

I found an algorithm for creating chaotic rotation schedules here http://www.rahul.net/mcgrew/derby/methods.html#chaotic. I gather that a program written using this algorithm would produce a similar result to the Stearns program.

However the motivation behind my post was really just to understand the scheduling options, and to work out which option is best given various scenarios. I have just read the excellent information on scheduling contained in the help files for the GPRM software. This proved very helpful, especially the charts comparing scheduling method features and the advantages and disadvantages of each. My key takeaway from this is that Perfect-N type charts tend to be the best option in most cases, especially if you are using point based scoring. The exceptions would be if you want to run a no-chart or or no-schedule based derby.



User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6888
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Stearns algorithm

Post by Stan Pope » Sat Sep 20, 2014 7:46 pm

Yes, for large groups usually found in district or council races I like the 15th Burlington cuz lots of packs can bring tracks. Gives each participant lots of runs in a modest duration. My next preference for large groups in NoChart Elimination, typically 4 or 5-loss elims! This can also be helped by having an extra track or two.

Since you have the .exe in hand, you can always "reverse engineer" (decompile) it! :)

The method in Stearns is that in each "round" each racer runs exactly once. So if it is scheduling 20 racers on a 4 lane track, it will schedule in rounds of 5 heats, depending on the time allotted. (The formula seemed to me that it assumed that track staff would stage the cars... they are about twice as fast as youngsters.) The program claims to schedule each racer against as many different opponents as possible and, if I recall correctly, to assign lanes reasonably equally. (My experiments with it indicated that once it had scheduled against all other opponents, it seemed to no longer try to equalize the number of times each opponent was matched.)


Stan
"If it's not for the boys, it's for the birds!"

User avatar
Darin McGrew
Pine Head Legend
Pine Head Legend
Posts: 1735
Joined: Fri Jul 18, 2003 1:23 pm
Location: Mountain View, CA
Contact:

Re: Stearns algorithm

Post by Darin McGrew » Sat Sep 20, 2014 10:33 pm

It's been ages since I looked at it, but IIRC, the results produced by my Perl script (based on the algorithm on my web page, linked above) were better than the results from the Stearns EXE. And since then, Stan has published information about PN and CPN and PPN scheduling.



User avatar
Vitamin K
Master Pine Head
Master Pine Head
Posts: 981
Joined: Sat Apr 20, 2013 7:26 pm
Location: Montgomery County, MD
Contact:

Re: Stearns algorithm

Post by Vitamin K » Mon Sep 22, 2014 8:18 am

So is the Stearns algorithm actually published, or is all of the logic contained in an opaque DLL?

I'd be a little wary of trusting my races to a 'black box' generator like that. :)



Post Reply