Logic Masters India announces August 2011 Puzzle Test — Japanese Puzzle Land

Dates : 20^th and 21^st August

Length : 100 minutes

IB and Submission Link : http://logicmastersindia.com/M201108P/

Authors : Yosuke Imai, Atsumi Hirose, Takeya Saikachi, Daisuke Takei

Test Solver & Translator - Hideaki Jo

The quality of the IB is impressive. 4 practice puzzles for Shakashaka, Mochikoro and Yajilin+, plus the example puzzles. Just great !

I am just slightly afraid about the timing of the test, seeing the name of the test-solver...

For Yajilin+ it says "Numbers indicate the amount of black cells in at least one of the two possible directions". Should that not just be in one of the two possible directions, without the 'at least'? All of the examples seem to match this... Otherwise it means we can't assume that it doesn't apply to both directions right?

ETA: Forgot to say, the test looks great!! Really looking forward to trying out the new types... I just hope I can make it home that weekend in time to compete!!

Edited by DreamRose311 2011-08-11 8:24 PM

All of the examples satisfy 'exactly one direction', but rules do not require that. So, you have to keep in mind that both directions can be correct.

Title of the topic: "Japanese Puzzle Land — LMI August Puzzle Test — 20th and 21st July"

Thanks deu. I was also unsure because the instructions also said 'left or right' and 'up or down' instead of and/or. I meant to add that earlier... Anyway, thanks for the clarification :o)

100 minutes to do over 4500 grid cells worth... yikes! Better not be anything seriously difficult if anyone is going to finish. I look forward to trying to dash all of these out next weekend.

The other way to look at it is "one has to figure out the direction of the arrow, which is missing. it doesn't matter what happens at the other direction." The practice puzzles are very helpful.

As others noted, the concept of practice puzzles for the new type and not so well-known types is nice. Many thanks to the authors.

Interesting measure - 4500 grid cells.
Just computed that July Nikoli Selection had 4100+ grid cells, although more than half of them belonged to the marathons.

Guessing with enough Nikoli practise it is probably possible. Sadly I don't really have any. I'm mostly worried about the general size of the grids. I don't mind big puzzles, just not to solve as fast as possible. Too afraid to make one dumb mistake I won't be able to locate.

Any chance I can get an extra 20 minutes as a birthday present?

Edited by Para 2011-08-13 10:31 PM

Is it too late to make this test more Paint friendly? Grey or black-and-offwhite cell borders (see puzzles formatted by, for example, MellowMelon or mathgrant) make solving with a fill tool possible. Also, faint diagonal lines in the Shakashaka puzzles (see http://mathgrant.blogspot.com/2010/12/rules-proof-of-quilt.html) are helpful.

If you want to practice, nikoli.com can be a good material. There are 10 free sample puzzles for each type (Sorry, there are no Fillomino, Shakashaka and Mochikoro). For Mochikoro, there are 30 puzzles in janko.at.

Considering the results of other tests (especially Nikoli Selection), authors' intention and my test-solving time, "100 minutes" seems not too short. Some puzzles are big but will not need too much time to solve. For instance, in large Akari, ONE lightbulb illuminates 37 cells (out of 336, including itself) if there are no black cells in that row and column!

For top solvers:
There will be an announcement about partial bonus points later.

Shakashaka online at http://www.janko.at/Raetsel/Shakashaka/index.htm

Are the puzzles in the test going to be the same size as the ones in the instruction booklet? I mean the cell sizes not the dimensions of the grid. I was having a bit of trouble with them, especially with sudoku and kakuro, since with this size they are not very pencil-marks-friendly...

Other than that I am really looking forward to this test. The quality of the puzzles in the instruction booklet is superb, and we got training material too :)

Marcin

Thanks for your suggestion and sorry for the late reply. We have added very faint solid lines to some grids in Puzzle Booklet. We think you can solve with a fill tool.
As for Shakashaka, we decided to include two grids for each puzzle (Left: without diagonals, Right: with diagonals) in Puzzle Booklet. Each competitor can choose whichever he/she likes.

In most puzzles, a little bigger than Instruction Booklet. More precisely, size of cells (length of an edge) is
7mm (Sudoku, IB), 6.3mm (Kakuro, IB)
10mm (Sudoku, PB), 7.5mm (Kakuro, PB)
if printed on A4 paper.

Here is a sample page from the puzzle booklet (with Shakashaka puzzle). See image below. Click here for pdf version.

Just polished off the instructions - a tantalising appetiser to be sure. I'm definitely looking forward to the main course!

100 minutes for 23 puzzles - and at roughly 5 points per solving minute available - these timings and gradings look more than reasonable to me. Even without deu competing I'd be surprised if no-one finished the set in time!

Tom, I counted again, and there are 25 puzzles.
Your last few tests have been going neatly, and I hope you had your last counting mistake / typo in this test

Bonus points and partial bonus points

Bonus points will be computed from the time "Claim Bonus" button is clicked.
5pts/min -- solved 25 correctly
3pts/min -- solved 24 correctly and a 5-20 points puzzle is wrong
2pts/min -- solved 24 correctly and a 21-30 points puzzle is wrong
1pt/min -- solved 24 correctly and a 31-40 points puzzle is wrong.

Note : Partial bonus will be given after authors/organizers judge the wrong answer not to be dummy.

Puzzle booklet is uploaded. It has 13 pages. There is no cover page. Each page has one puzzle type.

Couple of notes :
1) After the password is displayed, the password can be selected using mouse and copied using right-click. Typing the password manually takes time and is error-prone.
2) All point claims should be made using the score page. This test has simplest of answer keys. So, we aren't sure yet if we can accept many typos.
3) While the test is running, it is best to avoid puzzle specific comments (e.g. puzzle ABC is too hard or puzzle XYZ is too easy). Please discuss them after the test.
However, feel free to leave your feedback about the test in general after you complete the test. Its everyone's feedback that most authors look forward to.

Wow is all I have to say. The quality of some of those puzzles was excellent, and I could see someone solving all of them in 100 minutes. I was able to do 14 of them, and every one I solved was good. The bottom Fillomino and Top Hitori stood out to me, but there were a lot of beautiful designs and solving paths. Thanks to the puzzle designers for such a great competition.

Edited by gpagano 2011-08-20 8:08 AM

Chinese translation available here - http://www.sudokufans.org.cn/forums/index.php?showtopic=288

Nice puzzles.Second Fillomino is the pick of them.

Edited by anurag 2011-08-20 2:02 PM

puzzles are of nice quality..
every puzzle has different ways of solving and a trick to do it.
although i could have done better..
thanks to the authors for organizing such a beautiful test.. :)

A quick note about claims : No claims will be accepted for first 7 puzzles in case of typos. There answer keys for them can not be simpler.

great test.....................awesome puzzles.......thanks to all the authors and ofcourse LMI.

Amazing test! It's pretty easy to say that this is the best competitive round of Japanese puzzles that I've seen on LMI. But I also feel like saying that this is one of the best tests we've seen as well. Thanks for the work to put this together, and I hope it can inspire my next test to be just as good.

That was an excellently put together set of puzzles. Thanks for introducing me to Mochikoro as I'd never heard of it before this test and now it's one of my favourite puzzle types. Yajilin+ is a clever variant that complements the original puzzle type nicely and I hope to see more made. Great job all.

A good set of puzzles, arigatoo gozaimashita, unfortunately I wasted too much time on big puzzles that never revealed their secrets to me... if only I was better at puzzles! :)

But at least I'm the first person to (claim to) have solved the password! Maybe that's worth some bonus points?

My compliments to the chefs!

This is not the first time I encounter problems with the answer keys for loop-based puzzles, I triple checked my yajilins and masyus and it doesn't seem to work.
Both my Masyus still turned out "incorrect" (while the puzzles were solved correctly) - I'm wondering if one-off typos can be claimed since both masyus total up to a whopping 41 points.

I definitely enjoyed this, although I must say I was frustrated somewhat with the physical printed size of the larger slitherlink and heyawake puzzles. Both were lovely puzzles but during the time I restarted and reprinted both of these several times and kept making silly errors which (aside from the usual dose of my own carelessness) I can only put down to overextending lines or dots due to the small sizes of the grid cells.

Aside from this one grievance, hats off to the authors - I much prefer puzzle tests where authors don't indulge in their own variants too much, after all the classics are classics for a reason :) Although that said, the yajilin variant here was definitely good fun!

Adding to everyone else's thanks to the authors. Unfortunately for me, I spent far to long with the Slitherlinks and made mistakes in them - wasted effort on my part.

General question: How soon after the test ends (and where) can we get the password? I rarely get the block of time to do a test, so I'm almost always just waiting for the end of the official time so I can start it.

It is fixed at

5:30:01 AM IST + Length of the test
(i.e. 00:00:01 AM GMT + Length of the test)

You can get the password from the test page itself. (You have to login though)

OR, couple of hours (not fixed) after the test is over, pdf without password is uploaded at http://logicmastersindia.com/lmitests/downloads.asp

Thanks for the quick response! I am logged in, but the start button is disabled, so I can't 'start' the test and get the password. I can wait.

The test gets over at 5:30AM. That means no one can start the test after 5:30AM.
The test gets *really* over at 7:10AM. So, you can see the password after 7:10AM.

Technically, we could display the password after 5:30AM. But we don't do that because may be some players are little late, and may ask to push the "end time" by some time.

Japanese Puzzle Land is over.

Congratulations to Thomas Snyder (all 25 puzzles, 550 pts), Kota Morinishi (21 puzzles, 434 pts), Taro Arimatsu (20 puzzles, 426 pts) for topping this test. And I must mention that Ken Endo got the same score as Taro.

Out of top 10 players, 7 are from Japan, 2 are from USA, and 1 is from Poland. From the results, we can easily see that Nikoli readers and/or Nikoli.com members had much advantage in this test.

258 players participated, out of which 203 got non-zero scores. In terms of the number of participants, this was the most successful puzzle test in LMI! Thank you for participating!

I will write something about the test later. I am exhausted after a 48 hours test now.

I totally agree that this test had the simplest of answer keys. And, most of the solvers had to spend only a few seconds to input the answer keys.

But, at the same time, the answer key also became much more prone to guessing - not for all puzzles, but for specific ones. If you had solved a little of these puzzles, you could guess the answer key with a fair chance of success. According to me, the puzzles which fell into this category were the ones which involved counting in certain rows/columns (eg hitori, shikaku, akari, etc).

Because of the answer key, in the end, I was encouraged to make guesses for broken or even almost blank puzzles. I made 4 guesses and got 2 right - for a total of 48 points. Thats almost 10% of points.

So, question to ponder: should we make the answer key just a bit harder - for counting based answer keys?

Overall, a very enjoyable set of puzzles. I specially liked the new types - Shakashaka and Mochikoro - as never saw them before. Thanks to the puzzle team for an excellent test. And, Congratulations for the most successful test yet on LMI !!! (in terms of participation)

I also think you deserve some bonus points. But did you click on "Claim bonus"?

Can you also share the secret here, as I am not able to make anything out of it? (I initially thought something to do with Roman numbers, but there is no W in Roman numbers.)


Edited by rakesh_rai 2011-08-22 1:31 PM

A lot of giants was here. I always remember my perfect zero score on giant part in Minsk 2008.

I hope this is not a future of the puzzles. Try to imagine such sudoku contest with every second 12x12, 16x16 puzzle or larger. I think it wouldn't be interesting. Anyway, thanks to the authors. Also, I learned something about the history of the puzzles.

Nikola

Solution Booklet - http://logicmastersindia.com/lmitests/dl.asp?attachmentid=172&v1

It also has who created which puzzle and some interesting links.

Didn't we have similar answer keys in other tests, especially in Nikoli Selection 2011?

Well, I'm not sure about other competitions, but we unlikely to have 12X12 / 16X16 Sudokus in a Sudoku contest. (Our flash doesn't support it :-)

Regarding puzzles : We generally don't have large puzzles in LMI tests. But 'larger' puzzles are part of Nikoli's tradition. So I thought having slightly larger puzzles will appropriately represent how Nikoli's puzzles are.

I mentioned the same (as a test solver) - e.g. hashi in NS2011 was a very guessable answer key.

Thanks for pointing out. We adopted simple answer keys because we would like competitors to concentrate on solving, not counting. But we must admit that some of them were too simple and easily guessable. We will consider their balance more carefully next time.

P.S. At first, answer keys for Slitherlink, Masyu, Shikaku and Yajilin+ were much simpler. I asked authors to change them.

I see a difference between the large puzzles here and many of the Giants part in Minsk 2008. Basically, there are some puzzle types that, for the most part, are locally solvable meaning that the difficulty mostly scales with the number of cells. There are other puzzle types that use more global constraints and get much harder as the size increases. Sudoku is certainly in this category, as are the types Tents, Star Battle, and Pills from Giants in 2008 that get much much harder as the size increases.

But puzzles like Kakuro, Slitherlink, Masyu, Heyawake and others that went "big" here are not types that require you to think about the whole grid at one time very often if at all. The size makes it more of a challenge of not making mistakes (as errors will eventually propagate and cause trouble), but it still scales with the overall size of the puzzle. On Nikoli.com, I call these "larger but not harder" puzzles. And this test matched the Nikoli style perfectly well where most of the hards that are 14x24 are as challenging as a 10x10 hard puzzle, but take 336/100 times the time.

Edited by motris 2011-08-22 8:48 PM

I read all comments. Thanks for posting how you felt about the test.

Puzzles: In making Nikoli test, we had two options about difficulty and grid size:
(A) only small grids (8x8 -- 10x10), from easy to extra hard [WPC standard]
(B) from small to big, from easy to hard [Nikoli standard]
Authors adopted (B), because they wanted to think highly of Nikoli style. We are confident that we presented one aspect of good puzzles/competitions.

One of the purposes of this test was to popularize a new puzzle (Shakashaka) and an old puzzle (Mochikoro). That is why we included some practice puzzles for these types. We are very pleased to know 89 solved Mochikoro (Top) and 59 solved Shakashaka (Top) correctly.

Timing: My test-solving time was 80:14. This includes checking time for most puzzles, but without competition pressure and answer entry. Competition time was decided considering authors' intention and my time. This was a little short especially for those who are not familiar with Nikoli puzzles, but we were relieved when motris solved all puzzles correctly. Apart from motris, MellowMelon solved 25 and xevs solved 24, but both of them made mistakes in high pointers.

Password: This is the only puzzle I made for this test. It is a verbal arithmetic problem with a unique solution (no leading zeros). M and W represent specific values with respect to the date when this test was held.

Also motris and xevs started the test almost at same time (may be 5 minutes apart). It was really interesting watch their submissions (both of them submitted each puzzle after solving). After 70 minutes into their starts, motris had 391 points and xevs had 390 points.

I'm starting to really like the shakashaka puzzles. I think the triangled grid works much better for solving purposes for me. The normal grid I am not really sure how to keep notes. I didn't like them the first time I saw them as I always messed up notationwise (somewhat like corral, which had the same problem at first). Don't think I'm the only one who feels this way, but I think it's a better way to represent them, if you want to try to make it more popular.
Mochikoro I was a litle less a fan of as it seems very intuitive a genre. While solving the big one afterwards I kept feeling intuitively where rectangles had to go, because I couldn't see how it would turn out unique otherwise. I did it logically eventually, but those thoughts somewhat clash with an enjoyable solve as I want to do it logically.



Think the timing was good. I know, I'm not a quick big puzzle solver, so knew I wouldn't be in the top. Just missed out of the top 20 as the timer ran out when I was about to click submit on the larger heyawake. I used the test a bit to practise puzzle I never really solve: namely akari, heyawake and the 3 new ones. Still learning the tricks and patterns on akari and heyawake. Love to get some pointers on them at one point. Feel like I'm missing a lot of things still.



I understand why it was done. Have no problem with it. Just know that without a regular Nikoli practise on puzzles this size, it's harder to do them fast. the designs all were okay for a fast solve. Still got stumped by puzzles I normally wouldn't have a problem with in smaller size (masyu/slitherlink).
I'm still not too sure about the kakuro points though, as I was faster in my kakuro solve for the small one than the big one. But that might be because I make a lot of these hard small puzzles and quickly recognise the breaking points in hard kakuros.