2011 UK Sudoku Championship (11-12 June 2011)

Information and discussion on Puzzle Competitions
motris
Posts: 77
Joined: Tue 21 Dec, 2010 5:35 pm
Contact:

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by motris » Tue 14 Jun, 2011 3:24 pm

PuzzleScot wrote: Will you be good for the UKPC the weekend after next? I can guarantee that it is 100% puzzles, with not a sudoku to be seen! ;)
Although some gentle bifurcation may be required...
I'm much more likely to participate in a puzzle competition over a sudoku competition in general, and am definitely looking forward to the UKPC in another week. I can pretend it's a USPC since it will be a similar weekend, even though it looks like our championship will again slip into August.

MichaelC
Posts: 12
Joined: Thu 29 Jul, 2010 8:20 pm

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by MichaelC » Tue 14 Jun, 2011 5:56 pm

Hi all, enjoyed the contest and congrats to Tom and David.

Michael

GarethMoore
Posts: 1204
Joined: Thu 24 Jun, 2010 9:27 pm
Contact:

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by GarethMoore » Wed 15 Jun, 2011 12:48 am

motris wrote:Sure. I guess I'd simplify my comments by saying there are two axes of difficulty with a sudoku. They can have easy or difficult (or non-existent) logical steps. They can have a broad or narrow solving path (meaning fewer places to make the next deduction). These have a low score on the type of step axis and a very high score on the solving path axis. For those who don't track where they have and haven't looked as well, these puzzles can have the feeling of "I'm not making any progress" or "It's coming in fits and starts".
I agree. I've tested these four grids in my own sudoku software, which grades puzzles on all of these factors - in particular it looks at how many different deductions can be made simultaneously and knows that if there are lots then it is a much easier puzzle than if there is one. Furthermore it knows that spotting a "hard" logic move when there are very few options left in the puzzle is pretty trivial compared to spotting it as the very first move in the puzzle.

Based on testing these puzzles I've rated them (for Sudoku 5,6,7,8 in the test) as 3801, 2579, 3336 and 2369 respectively, where lower is easier. For comparison, a very easy sudoku can rate as low as 30 and in this case there will be many easy placements simultaneously available. A very hard sudoku, with a few difficult sets or x-wings that need to be spotted early on in a mess of pencilmarks will score 7000+. The hardest sudoku I've made scores just over 10000. However this still makes the ones in this test pretty tough sudoku, for those of us not blessed with your number-spotting talents. :) Without pasting in a curve of value distribution, if I generate puzzles completely randomly I get puzzles as hard as the first one around 1 in 400 times. So I'd say it was not surprising most people found it tough.

In particular the first one (puzzle 5) requires lots of naked singles fairly early on in the solve, and in each case there is only one spot to make progress. Based on analysing 100,000s of plays on puzzlemix.com I am pretty certain that naked singles for the average solver can be very hard to spot indeed - although ironically they show up trivially if you write in all the pencilmark candidates.

Here's my analysis for each puzzle. The number on the left shows the number of empty squares and the logic required at each stage to progress. If the next number is 1 less then there was only 1 place to apply this logic, so it might have been very hard to spot (certainly you have a lower chance of stumbling on the next 'move' quickly). Obviously this is not a full tree so there may be alternative harder moves at a particular point as an alternative (and you may disagree on my grading of what is harder), but this gives a good guide to overall difficulty:

Puzzle 5: (lots of only-one-place-to-progress naked singles early on - tough)

55=uniqueInRegion 53=uniqueInRegion 52=nakedSingle 51=intersection 51=intersection 51=nakedSingle 50=nakedSingle 49=intersection 49=intersection 49=nakedSingle 48=uniqueInRegion 47=nakedSingle 46=uniqueInRegion 45=nakedSingle 44=nakedSingle 43=nakedSingle 42=nakedSingle 41=nakedSingle 40=uniqueInRegion 38=uniqueInRegion 37=nakedPair 37=uniqueInRegion 35=uniqueInRegion 33=uniqueInRegion 30=uniqueInRegion 28=uniqueInRegion 26=uniqueInRegion 23=uniqueInRegion 21=uniqueInRegion 19=uniqueInRegion 17=uniqueInRegion 15=uniqueInRegion 11=uniqueInRegion 7=uniqueInRegion 3=uniqueInRegion

Puzzle 6: (quite a bit easier, though still a narrow solving path)

56=uniqueInRegion 52=uniqueInRegion 51=nakedPair 51=intersection 51=nakedSingle 50=uniqueInRegion 49=intersection 49=nakedPair 49=uniqueInRegion 48=uniqueInRegion 46=uniqueInRegion 44=uniqueInRegion 43=uniqueInRegion 41=uniqueInRegion 38=uniqueInRegion 35=intersection 35=intersection 35=uniqueInRegion 34=uniqueInRegion 33=uniqueInRegion 32=nakedSingle 31=uniqueInRegion 30=nakedSingle 29=uniqueInRegion 28=uniqueInRegion 24=uniqueInRegion 17=uniqueInRegion 6=uniqueInRegion

Puzzle 7: (more only one-place-to-go naked singles early on make it tough, followed by various intersections and naked pairs needed, also early on)

54=uniqueInRegion 51=nakedSingle 50=nakedSingle 49=nakedSingle 48=intersection 48=intersection 48=nakedSingle 47=nakedSingle 46=intersection 46=intersection 46=intersection 46=nakedPair 46=intersection 46=nakedPair 46=uniqueInRegion 45=uniqueInRegion 44=uniqueInRegion 43=uniqueInRegion 41=uniqueInRegion 38=uniqueInRegion 37=uniqueInRegion 35=uniqueInRegion 33=uniqueInRegion 31=uniqueInRegion 28=uniqueInRegion 21=uniqueInRegion 14=uniqueInRegion 12=uniqueInRegion 9=uniqueInRegion 4=uniqueInRegion

Puzzle 8: (a few only-one-place naked singles, but relatively easy and ironically the only one I couldn't solve...)

56=uniqueInRegion 54=uniqueInRegion 52=uniqueInRegion 51=nakedSingle 50=uniqueInRegion 48=uniqueInRegion 47=nakedSingle 46=intersection 46=intersection 46=intersection 46=intersection 46=intersection 46=nakedPair 46=uniqueInRegion 45=uniqueInRegion 44=uniqueInRegion
43=uniqueInRegion 40=uniqueInRegion 39=intersection 39=uniqueInRegion 38=uniqueInRegion 36=uniqueInRegion 35=uniqueInRegion 32=uniqueInRegion 26=uniqueInRegion 15=uniqueInRegion 5=uniqueInRegion

I'm not really sure why the second two were worth more points. In fact I only just noticed that!

Fred76
Posts: 163
Joined: Mon 09 Aug, 2010 1:53 pm
Contact:

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by Fred76 » Wed 15 Jun, 2011 1:41 pm

I've solved all grids, the latest without time pressureand and chronometer. Here are my impressions
mikecolloby wrote:Well done Tom, great job!

It would be useful to get some more feedback so that we can make adjustments for next year. I've already got the point about tightening the definition of a Sudoku, and that the Classics should be a bit easier. Please don't blame Puzzler for this, because we selected the ones we wanted to use. What I would like is feedback on whether the objectives were appropriate, and whether they were met. The objectives were:
1. Puzzles should all be challenging, but not so hard that they rewarded guessing. Were there any that you guessed? Were any too easy?
2. There should be a number of puzzles at the beginning that new players can have a go at. This should have included the Classics, but it seems they were too hard. What about the other early puzzles, and this objective itself?
3. The end should make the top players sweat, but it should be possible to finish so that the winner has to solve everything and hence there is no benefit in cherry-picking. Clearly it was a bit too long by this definition. We estimated it would take the winner 133 minutes to finish, and this proved to be about right. On the other hand, LMI aim for quite a number of players to finish. What do you think?
About the classics. I don't find them too hard. Perhaps they were worth 2 points more each, but one need not advanced techniques to solve them. I've solved one during tournament, I had the impression that I was not very fast, so I skip the others to concentrate on variants. It's not just a question of time and points: generally, I find variants more interesting to solve and those of the test were just very fun to solve.

1. I was able to solve all grids without guessing. I found futoshiki hard for 14 points (I didn't try it during tournament).
2. I find it unfortunate for beginners that some easy grids were not sudoku. First 2 pages that are printed are skyscrapers and calcudoku... Kid sudoku was easy, but rules are perhaps not easy for beginners, so perhaps they will not try to solve this variant first. Perhaps it should be preferable to have some variants with easy rules for beginners, like diagonal sudoku or irregular (jigsaw) sudoku.
3. I'm agree with uvo and motris. If you are organizing national championships, you should not take into account other players to calibrate the test. We are happy that you give us the possibility to take part, and we can understand if the competition is a little easier (or with less grids) so that many top international players can finish the test. But estimate time and difficulty is always a hard point when organizing tests, even with a lot of good testers. It's important that top players "fight" on the same grids, and that it's not a question of choice, as pointed motris.

Another point: I found very pleasant that they were not just one author for this test. Each author has his own style and his favorite variants. I found the set of grids very varied and interesting. Among grids, I really liked No3 consecutive, frameless, top heavy, palindrome, diagonally non-consecutive, no-donkey step and five pair. They are uncommon and very nice grids.

Fred

david mcneill
Posts: 410
Joined: Mon 21 Jun, 2010 2:02 pm

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by david mcneill » Wed 15 Jun, 2011 10:59 pm

I have also now been able to solve all the puzzles. My favourite was the Frameless, which solved beautifully. I had to guess on all 4 classics during the test and the Futoshiki after the test.

I will try the classics again without the time pressure.

david mcneill
Posts: 410
Joined: Mon 21 Jun, 2010 2:02 pm

Re: 2011 UK Sudoku Championship (11-12 June 2011)

Post by david mcneill » Thu 16 Jun, 2011 3:32 pm

OK. Having solved the classics again, without guessing, I agree they were all fair. Despite what Gareth's software maintained, my solving times were 5:28, 6:32, 5:36 and 12:52 respectively, which seems to correspond with Gareth's own human experience. Not as fast as some others, but hardly justifying my earlier apoplexy.

Apologies to Mike and Liane for my over-reaction. I will try to stay calm in future. Thank you for organising the contest.

Post Reply