**FIRST EMAIL**/Updated Alliance Pairing Algorithm

[AcesPease]

Originally Posted by dsmoker

The algorithim was very harsh for us. We frequently found ourselves paired with very inexperienced teams and went up against many of the titans of the CT regional (Uberbots, Gaelhawks). It wasn't until our matches on Saturday that we were paired with any really strong teams (love ya Cyberknights), and there was one team we saw three times (once with us, twice against), while many of the 62 teams we never saw, either with or against us.

I wonder if the time between matches was not entered correctly at CT. Almost all our matches were exactly 1 hour apart. And, I was a little surprised at the lack of variety in the pairings, we were with and then against many of the same teams.

Maybe we should be thankful we weren't against the same teams twice like last year. But, I wouldn't mind going 45 min or 1 hr 30 on occasion, so we could see a wider variety of the competitors on the field.

[Pat Fairbank]

As the scorekeeper for the Waterloo Regional, I was pretty happy with the scheduling algorithm considering what a small event it was (30 teams). The seedings at the end of the qualification round really seemed to reflect team quality well.

One thing that I noticed - not sure if it's a negative or not - is that the algorithm seems to optimize the number of other teams each team is paired with, possibly at the expense of optimizing the number of opponent teams. For example, at Waterloo, where each team played 11 qualification matches, each team had 22 different alliance partners (the maximum possible), while having a number of opponents ranging between 20 and 25. So no team was paired together with the same team twice, while having a repeated opponent anywhere between 8 and 13 times.

[waialua359]

Originally Posted by Dave Flowerday

I think there's still issues with how the individual regionals are setting things up. At Midwest, we felt the scheduling algorithm did a very poor job (especially since we had to play against 1625 three times in a row). However, the schedule at Boilermaker seemed to have excellent diversity (at the expense of occasionally very short cycle times between matches). I also felt that the diversity of the Boilermaker schedule was responsible for a more accurate Top 8 at the end of qualification matches (completely subjective, of course).

I'd say the algorithm is better when used correctly, but honestly I think a lot of the problems in the past were also due to incorrect use (minimum match spacing set too high). I think that option really needs to be taken out of the control of the local regional.

I agree that it should be taken out of the control of the local regional. Why even have a new and improved algorithm if they differ greatly from regional to regional. Our first regional had 64 teams with the second being 61 teams. The VCU one was pretty diverse, however, we played with one team, only to play against them in the next match, repeatedly. The Chesapeake one had us playing same teams over and over again. 4 times in 5 consecutive matches is a bit too much.
I still dont understand why we cant play against and with ALL different teams if there are only 8 matches in a field of 60++ teams. If they have to extend play time in order to do this (due to matches not evenly spaced apart), then I think its the lesser of two evils. Why go to a regional with a large no. of teams only to see the same few ones over and over again.

[ChrisH]

Originally Posted by Pat Fairbank

One thing that I noticed - not sure if it's a negative or not - is that the algorithm seems to optimize the number of other teams each team is paired with, possibly at the expense of optimizing the number of opponent teams. For example, at Waterloo, where each team played 11 qualification matches, each team had 22 different alliance partners (the maximum possible), while having a number of opponents ranging between 20 and 25. So no team was paired together with the same team twice, while having a repeated opponent anywhere between 8 and 13 times.

That's because you need more opponents than partners. In fact with that small a pool, you HAD to have repeat opponents. With 11 matches you needed 22 unique partners and 33 unique opponents to not have a repeat. That is really hard to do with only 29 other robots. Under these circumstances it is understandable that virtually every match had at least one non-unique opponent.

Which would you rather have? Play eleven robots twice or two robots eleven times? Admittedly those are the extremes and the likely result is somewhere in between ie. 6 robots three times and 3 robots twice, but you get the idea...

[mtaman02]

You know whats kinda funny about all this, FTA and the Scoring table can talk live w/ other fields operating that week to ask for help in general. While the FTA don't touch the score table during competition, I'm fairly certain they are trained how to operate the software since they have to sit and make sure the matches generated are correct.

I know our FTA guy sat there w/ a book, laptop and something to pull his hair out. Although unlike my many other FTA friends he actually was running numbers down to the decimal to make sure all was right . They were working real hard on thursday to get er done and even help up a match or 2 in the process. I don't see why they can't just calculate the schedule and have it run in the back round that way it doesn't delay the match.... anyways

LI's pairing algorithm wasn't too bad... may have a couple of repeats but wasn't noticeable at all. Seemed fairly random in LI.

[waialua359]

The algorithm used in the Hawaii one was pretty good.
We saw an opponent twice only twice during qualifying matches, considering there was only 37 teams here, and both of those teams happened in our last match. Considering the small size and 10 matches, I'd say its pretty random and spread out.
Cool!

[ChrisH]

Originally Posted by mtaman02

You know whats kinda funny about all this, FTA and the Scoring table can talk live w/ other fields operating that week to ask for help in general. While the FTA don't touch the score table during competition, I'm fairly certain they are trained how to operate the software since they have to sit and make sure the matches generated are correct.

I know our FTA guy sat there w/ a book, laptop and something to pull his hair out. Although unlike my many other FTA friends he actually was running numbers down to the decimal to make sure all was right . They were working real hard on thursday to get er done and even help up a match or 2 in the process. I don't see why they can't just calculate the schedule and have it run in the back round that way it doesn't delay the match.... anyways

LI's pairing algorithm wasn't too bad... may have a couple of repeats but wasn't noticeable at all. Seemed fairly random in LI.

Running the scoring system is part of FTA training. Partly so we know how it works if something goes wrong and partly so the Scorer can take a break without shutting down the field.

The schedule generating algorithm makes something like 5 milliion schedules and compares them against each other accoring to various criteria. The processor is pretty well maxed out while this is going on. We wouldn't want it to miss scores because it was too busy doing something else now woud we?

[qzrrbz]

Is the software/source for the scheduler available? I had read the white paper that purports to be its basis, but am looking for the real thing now.

Thanks,
rnd

[Joe Ross]

The software is available in the link here: http://www.chiefdelphi.com/forums/sh...1&postcount=34 but the source is not available.

[Jwxie]

Originally Posted by Joe Ross

The software is available in the link here: http://www.chiefdelphi.com/forums/sh...1&postcount=34 but the source is not available.

Thank you man
you save me
hahaha jkjk