Log in

No account? Create an account
06 February 2011 @ 07:43 pm
Puzzle Competition Update  
There was apparently even a third puzzle competition this weekend (Puzzle Fountain by Alberto Fabris which I've yet to start trying as there simply is no time). But the two I played in went pretty well.

I got the top result in the Tapa Variations V. By 22.5 minutes. My best puzzle types are probably Sudoku and Nurikabe. Tapa is a lot like Nurikabe - make a lot of logical deductions and then visualize what might be required, try it out, and advance with logic and intuition. Tapa is very similar in style and is certainly one of my better types. I really enjoyed these puzzles, and look forward to the next one in two weeks where maybe I can break the three way tie for first place (although I'd gladly take .1 points per minute early to have a small 2.25 point lead!). I hope "adding more puzzles", as Serkan promises, doesn't mean too many more as the test was well timed for everyone else and upping the incomplete percentage for most solvers could be discouraging.

I also got the top result in SudokuCup V, with a close second by Jan Novotny of the Czech Republic. I didn't intend to take this test, but found a little extra time this morning after I was kindly asked by Karel to try it. There were a few really good puzzles, particularly the Surprise (a really elegant design that worked perfectly for an instructionless type) and the last Killer and Arrow variants. However, far too many puzzles seemed to use K1Z as if that is a meaningful theme. Yes, it's the authors' group, but overuse that theme in the Instructions and not in the puzzles itself where more novel things would shine. My two big stumbling blocks on the test were the Coded Sudoku (I wish I'd seen that it was making letter-forms to resolve the middle quicker - had to work from a bifurcation) and the Roman Sudoku, which wasn't hard, but where having to write over the letters made a messy paper that I mis-copied the first time. Fortunately, I fixed my entry during my check but I lost 5 minutes from when I was actually done. Still, my time held up and I was first there as well.

For those that think I need some humble pie, simply request more Sternenhimmel and Laser puzzles on Crocopuzzle. That's the only thing that's been tripping me up lately.
TH: pencilrpipuzzleguy on February 7th, 2011 03:55 am (UTC)
Great work. I spaced on entering as per usual; will print them out tomorrow.
TH: shakerpipuzzleguy on February 7th, 2011 04:12 am (UTC)
What's the Sudokucup password? Follow-up: Why would they take it off the site?
motrismotris on February 7th, 2011 04:13 am (UTC)
I didn't copy it down this time. I was hoping there would be a password-less test posted soon. But maybe the internet will provide an answer to your question soon.
willwc on February 7th, 2011 04:20 am (UTC)
The password for the file is: UA8JNNW9!WPN5

Can't help with the second question :)
THrpipuzzleguy on February 7th, 2011 05:09 am (UTC)
Thanks, but I'm afraid that didn't work. I care substantially less about the second question. :-D
willwc on February 7th, 2011 05:29 am (UTC)
Whoops! I somehow lost a question mark along the way... the actual password is UA8JNNW9!?WPN5

This time I really mean it!
TH: pencilrpipuzzleguy on February 7th, 2011 05:32 am (UTC)
Brilliant; thanks!
kodyn on February 7th, 2011 08:23 am (UTC)
The password is free now...

There was a block to let people know, that the SC5 competition is over...
zundevilzundevil on February 7th, 2011 04:38 am (UTC)
I copied the Roman Sudoku onto a separate sheet -- with given letters replaced with candidates -- and eventually stared at it for about five minutes wondering where the next "regular" logical step was. There was a Forced Chains thing or whatever it's called that finished it up, but I still fear I was missing something.

Good showing on SC5, phenomenal showing at the Tapa thing. That's a terrific puzzletype with a surprising number of variants; I anticipate something better than 39th next time. Tho probably not 22.5 minutes ahead of the field.
motrismotris on February 7th, 2011 04:40 am (UTC)
I didn't need anything unusual(ly hard) on the Roman, so my guess is you incompletely copied the candidates or something.
zundevilzundevil on February 7th, 2011 05:10 am (UTC)
That's good to know -- I'll give it another hit sometime. I think that's the first of that type I've ever actually done correctly. Something about those (and the S for Sudoku style) just eats me up.
motrismotris on February 7th, 2011 05:12 am (UTC)
Well, I greatly prefer the clues segmented into a corner, as when we formatted both of the types for WSC5, so you can write the normal numbers in the center of cells without problems. Here and on Pencilmarks I simply wrote dark over everything (I didn't care enough to make separate grids).
(Anonymous) on February 7th, 2011 05:12 am (UTC)

Fantastic scores by our USA contingent!

For the Roman, I usually solve on the original grid and write my numbers above the letters. I did a basic "find the 9s", then 4s, then 5s, then 1s --> as this helps identify the numbers under the most constraints. III is 3 or 8, etc. etc. etc. Eventually you get to the solution. No fancy techniques like forced chains or "alternating inference chains (oy!)". A relatively straightforward solve.

Of course, that was after the time pressure, when I broke the darn thing about 4 times... :(

kodyn on February 7th, 2011 08:37 am (UTC)
I hope that there wasn't necessary to use any of higher techniques in the whole SC5.
I was revising almost all puzzles in the set, to make them easier...
kodyn on February 7th, 2011 08:42 am (UTC)
Thank you for your participation :)
And congratulations to your win!

I have seen your sending of answers and I was scared, you woun't find the mistake (I have hoped you'll make a verification of course)...
Jan (Krtek) has a mistake too and he corrected it too. I'm happy when I can see the result list with great names of sudoku scene :)

Thank you a GL to your next cometitions :)
(Anonymous) on February 7th, 2011 12:39 pm (UTC)
Managed to solve 13/15.

I question whether Kenken deserve merit for inclusion.
To me, it's not very sudoku-y. It's too mathematical as opposed to simple arithmetic in some variants - which kinda ruins the whole sudoku concept.

Not enough time for Arrow, since I tackled the test in order.
I just noticed after the test that Coded Sudoku was themed, :D

MellowMelonMellowMelon [wordpress.com] on February 7th, 2011 08:26 pm (UTC)
This is the first I've heard of you doing Nurikabe by some amount of intuition. Even for the ones that appear in WPCs (which are harder than most nikoli.com specimens) I tend to keep my solving very logical. But I do take much more of a fiddly approach to Tapa. I'm still trying to decide if that means I'm not as good at it or it's just easier to do it that way.
motrismotris on February 7th, 2011 08:42 pm (UTC)
Maybe I mean "pre-visualization", in the same way I sometimes intend the meaning of intuition on a Numberlink. Seeing both the islands and the ocean before writing them is something I do a fair bit when I don't immediately see the next step (and oftentimes doing so shows me where that step was).
MellowMelonMellowMelon [wordpress.com] on February 8th, 2011 12:39 am (UTC)
That does make a bit more sense. We still differ though. I only see my way through for puzzles in the style of that Botsu Bako one with the four rectangles in the middle, and that kind of black cell configuration is fairly rare.

Of course, like any nikoli.com puzzle, I plow through in try mode if I'm stuck and close to the end.
motrismotris on February 8th, 2011 12:42 am (UTC)
I almost never use Nikoli.com's try mode, except on sudoku time trials or unfriendly 17x17 Hitori. Of course on Croco-puzzle I take advantage of the 9-furcation capability all the time. I suppose I trust the Nikoli designers a lot more to have logical steps in mind.
(Anonymous) on February 8th, 2011 05:28 pm (UTC)
What will be the English definition of the Sternenhimmel puzzle?
motrismotris on February 8th, 2011 05:31 pm (UTC)
Re: Sternenhimmel
It translates to Starry Sky. The rules would be "place stars into blank cells so that the indicated number of stars appear in each row/column. The arrows in the grid must point to at least one star, and each star in the grid must be pointed to by at least one arrow".

The puzzle is a bit tedious on the crocopuzzle applet as the first step on paper is often drawing all the lines from the arrows to mark the possible fields. Once that step is done, the puzzle is pretty much bookkeeping work. I've been trying for awhile to not do that online, but I think marking cells first is really going to be necessary to get good at the puzzle so I'll go back to it the next time it appears.