Logic Masters India announces July 2011 Puzzle Test #1 — Nikoli Selection

Author : TOM “detuned” COLLYER

Dates : 9^th and 10^th July

Length : 90 minutes

IB and Submission Link : here

Just a quick answer key question. For shikaku, does this also have the same rule as fillomino, where you only type the last digit if a multidigit clue is in the row or column? I think this would help make the answer keys consistent.

Yes, I agree that it will be consistent. It is probably not necessary for the actual puzzle in the test.

We are discussing about having consistent answer keys across puzzle tests.
Please check this sub-forum and contribute.

just to clarify,
in the instruction it is mentioned that "It is highly recommended that you do not attempt solving the marathon
puzzles before correctly solving the first part of the test!"

is it like we have to first complete that 16 puzzles then only the points for marathon puzzle is valid.

I assume it means that solving 3 normal puzzles for 50 points will most likely cost you less time than solving one marathon puzzle for 50 points. So it's more point effective to solve all normal puzzles before attempting the marathon ones. Also probably more fun as working on a puzzle for a long time and eventually not finishing in time can be frustrating. You can check the previous nikoli selection, to see the size of the marathon puzzles, they are somewhat intimidating. I'm assuming they will be similar this year.

I have a question about the time bonus as it's mentioned after the first 16 puzzles. Does the time bonus count after the last correct submission from the first 16 or in total? Say someone finishes the first 16 in 65 minutes and another in 70 minutes, but the first solver finishes a marathon puzzle at 89 minutes, while the second solver finishes the marathon at 88 minutes. In this scenario, which player scores most points?

Swaroop,

Para has put it very nicely. You should solve marathon puzzles only after completing all Main section puzzles.

The "point value" of a main puzzle > The "point value" of a marathon puzzle > The "point value" of time bonus
[ "Point Value" is points you get for spending a minute ]
To secure more points, you should be doing the main puzzles first.

Regarding marathon puzzle sizes, they will be almost similar size as last year's.

Bonus points will be given after one submits all 16 main puzzles correctly. So, the first player in your example will get 5 more points than second player.

Essentially this test is about the main puzzles. But again these are puzzle types where solving time difference between top players and average players will be huge. So the marathon puzzles serve as bonus puzzles. Instead of giving significant bonus points for completing earlier, top players are challenged to solve the marathon puzzles to earn those points.

Tom, correct me if I'm wrong, and please share your perspective too.


If top players see this as a problem, they should post here / let us know. It is much much easier to fix a bonus system before the test starts.

Deb is essentially right, I expect to see some very quick times for the main section. I also expect that no-one will get 3 marathon puzzles, although I'll be surprised if there aren't a couple who get 1 and *maybe* even 2 of them. As Deb says, it gives top solvers a challenge to earn some bigger bonuses.

At any rate, there will be a small points-per-minute sub-bonus for anyone who finishes the main section, which will still serve as a tie-breaker to those who can finish but won't have time to get a marathon done. I don't expect this scenario will apply to many people.

I suppose the best way to look at things is this: very few people will get the whole set done in time. The marathon puzzles (OK not quite nikoli giants sized, but I have to compromise with printing and so on) should then be seen as something extra to keep you occupied for a few days after the test, to be enoyed at your own leisure without the pressures of against-the-clock solving.

Except that I expect for the usual suspects that getting that second marathon done will be the entirety of the podium decision since the general time bonus is much smaller. I mostly enjoyed the puzzles last time but had to say the marathon bonus system was not my favorite choice. We'll see how it works out this time.

I'd be highly surprised that even if anyone gets two out, that they'd not be one of the fastest solvers of the main set anyway. I don't think this works out to any plausible scenario as being unfair to any of the top contenders. The time bonus acts as the main tie breaker - but if anyone can get two of these puzzles out then in my eyes they are certainly deserving of finishing ahead of someone who only finishes one (or none), even if there's a small difference between them in the main section.

I agree this system is a little quirky, but I think it's an interesting idea and the best way to see if it really works or not is to give it a go. We've seen some really good innovations come out of past LMI tests.

Edited by detuned 2011-07-05 6:35 PM

There will be 2 small changes to the submission system this time.

1) The timer and password will be shown left to the text boxes.
Serkan had suggested this earlier so that players won't have to scroll to see the timer.
See image below.

2) When only one of the two required field are entered, the other field will be highlighted.
It was suggested by motris, and it had been pending for a long time. See image below for FILLOMINO puzzle.

I agree that we should see what shakes out. But my years of experience say that having three high variance (certainly high length) puzzles, of which solvers must choose which may be easier than others, is a different end selection criterion than who is the fastest solver of everything. Perhaps the Yajilin is actually 5 minutes easier. Perhaps not. I don't know, but the test will probably be determined by squeezing out an unknown large puzzle. It may work for me, or against me, but since I'm often criticized for pointing these things out afterwards if I'm not the winner, I wanted to point out my concern this time well before-hand.

I'm fine with bonus puzzles being used in place of time bonus. But they should almost always be sprint-like puzzles (2-3 minutes max) as you want more, not less, granularity at the end of a test.

can anybody tell how to solve SURAROMU of instruction booklet ?

With most Suraromu puzzles, since you use all gates, one good first step is to draw lines through all the gates with just a single cell width. You'll find doing this and connecting forced paths completes the entire upper-left of the example to give you a line that goes through gates 6 and what must be 7 and 8 (as if they were earlier gates you'd have two gate 4's since the grid has a labeled gate 4). After these initial steps, you have free ends you need to link to future gates and this is where you want to focus your attention:

(1) With the end that just went through 6, how do you get to a gate 5 before passing through gate 4?
(2) With the end that just went through 8, how do you get to gate 9?

To answer (1), you'll see the path through 6 must immediately turn or else it would leave no way to pass through gate 4. So the path goes over two cells (but not three cells) so it can get to another gate and leave a one-cell wide path through gate 4. You'll run into this step a lot. A path "blocks" part of one of the gates (you can X out that cell), but now leaves just one cell to pass through, so it must pass through that cell and the other path has to turn away immediately. With those things marked, the chase of the other end to get gate 9 really only has one option and being "path-efficient" is really key there.

The best thing to do to learn Suraromu puzzles is practice on a dozen or so. The logic is fairly different from other types, but once you learn a few rules the type is almost always very easy to solve. Besides Tom's puzzles, I'd point you to the Circumnavi-Gates on Mathgrant's site as good practice.

Edited by motris 2011-07-06 12:21 AM

thank you
i will try out those.

Re suraromu/slalom, there are some good examples on Melon's Puzzles blog as well. I don;t recall making many suraromu puzzles, so perhaps other sources are a good idea if you want to get some practice in.

Re motris, I'm sort of taking the view from the other side of things. My own testing times for the larger puzzles vary slightly, but then my solving abilities between these 3 types also varies slightly. The fastest possible times are hopefully all about the same, and so then the theory is the participant can choose to sprint on their own terms. This obviously relaxes the whole accuracy of scientifically measuring everyone's performance relative to each other - but on the other hand should hopefully be a little more fun. For example, I've come across comments where yajilin is a bit of a love/hate type for many people. If they happen to love solving yajilin, then great, but if they don't, then they aren't going to be miserable about being forced to solve an arbitrary type as an apparent tie-breaker. The choice adds a little variability, but hopefully not too much.

just as a suggestion.
cant there be another way to answer LITS puzzle its confusing to count number of T's or S's we may miss out one.
answer key similar to NURIKABE or any other.


Edited by rakesh_rai 2011-07-08 7:36 AM

Answer keys can be a fiddly business, and I'm not sure I'm entirely happy with the hashi one either. My tip for LITS is once you've completed the puzzle, to briefly spend some time writing the relevant letter on each of the tetrominoes (this doesn't take very long) and then answers will be relatively easy to read off. Promise!

ok no problem

Puzzle Booklets (one for Main Section, and a separate one for Marathon) uploaded

Main Section Puzzle pdf has 10 pages, including a cover page.
Marathon puzzle pdf has 4 pages, including a cover page. Note that Marathon puzzle pdf file has Landscape orientation.

Both pdf files have same password.

As discussed after Fillomino-Fillia test, a typo will cost 20% of the puzzle value. Of course, one gets 80% only if the author/organizers are certain that the puzzle was solved correctly, but a mistake was made during submission.

Nikoli Selection has started - Good luck to all.

Score page - http://logicmastersindia.com/M201107P1/score.asp

If you've not taken the test yet, you can check who all have taken the test in the score page.
If you've taken the test, you can check details of your submission, and others' scores. If you believe some puzzle is marked incorrectly, please use the score page to 'claim points'.

Hello, I'm not sure if this is the place to ask for this, but can I have a manual override for a typo on the numberlink?

Yes, this is the place to 'claim' typos.

My kingdom for five more minutes! At least I'm error-free this time. Thanks to all for the test.

Check updated score page.

Very nice set of puzzles Tom! Thanks.
The answer key for LITS and Hashi should be made simpler if possible in subsequent tests.

Hashi Suggestions: http://logicmastersindia.com/forum/forums/thread-view.asp?tid=354&p...
For LITS, maybe a 'Tapa-style' answer key for marked rows/columns would be suitable.

Edited by Rohan Rao 2011-07-09 10:26 AM

Nice group of puzzles, more, more!!

I Find the answer key for hashi and lits very simple indeed....can't be simpler...

by the way in shikaku I entered the column next (right) to the column A I know its stupid but I keep doing it.... any chances to have the points awarded?

stefano

All claims made in forum or made using the Score page will be settled. Please allow us few hours.

puzzles were excellent .
I njoyed a lot.
although could have done better.
would be waiting for nikoli 3..!!

Thanks for the puzzles, great test!

If you can reconstruct all my answer entries then you will see that my latest entry was just overwriting one valid entry key (that I entered much earlier) by another. This could save me some seconds for the tiebreaker and one rank in the final table. :-)

Have fun,
Stefan

P.S.: I actually had a reason to overwrite my solution a second time. You can write me an email if you want to have an explanation. Here it would be a spoiler.

@euklid, please see my PM

The answer key for Hashi is simple, I agree. But I'm not convinced its the simplest. The problem is there's a good chance of a counting error. It requires the player to go through the entire grid carefully. And one can easily mistake a double bridge by a single one in counting (especially those with cumbersome handwriting).
There are some interesting suggestions in the answer key topic for Hashi.

I had to change computers part of the way through since mine lost its internet connection. The results page only shows those I entered on the second computer - can you add in those I submitted previously please? Thanks! :)

Yet another enjoyable test... unfortunately my printer ink died (thanks hubby for printing so many things and not telling me) so I had to transcribe most of the puzzles which slowed me down even further than my usual glacial pace.

Great test. Typo correction request for the Hashi: the solution I entered is the total number of horizontal bridges (so 2 * double horizontal bridges + 1 * single horizontal bridges).

Nice puzzles(whatever i attempted) !
Made few really stupid errors while submitting the answers(Maybe im too sleepy at this hour!)
(Please consider them if possible!)
Errors:
1.Shikaku-submitted the column to the right of B arrow.
2.Numberlink-submitted the combination of R2 and R3(i dunno what was i looking at!)
Overall i had a good time solving !
All the best to all the remaining participants !

Very nice puzzles and a lot of fun solves to be found. Thanks Tom!

Unfortunately updated submission form (since IIRC June's sudoku test) does not work for me on Linux (Firefox with Adobe flash player, Chromium with internal flash player).... after login in I get only text "Connecting to server to get status" and nothing else happens; I can't press Start.

Edited by mtronic 2011-07-10 10:00 PM

Really sorry to hear that. It will be too difficult to fix it in short notice.
Please see my message.

I would like to work with you sometime next week to fix the issue since I don't have access to Linux machine.

Thanks for your prompt support.... looks like I've found the issue within the Flash player settings, there wasn't local storage allowed for LMI site. Now the applet shows timings and it looks I'll be able to use it for the test.

I expect, that this same might happen on Windows as well, so even other users might come across this issue if they have disabled or too small storage size setting.

Just like to add my thanks to Tom and all at LMI for this. Very enjoyable - too many silly mistakes as per usual from myself at the time.

Just as a comment - I have a laserjet printer that once it's printing is fine and fairly quick. It seems to take ages formatting some documents before it decides to print them. With this particular contest I had to wait for it to prepare the document prior to print, then it just printed the first page then sat waiting to prepare the next. Unfortunately, because the first page had the list of puzzles I was unable to start solving then and had to wait for the second page. This took 3 minutes in total. If possible, can contest authors inform us whether the first page of the puzzle booklet will have puzzles on or not? - At least then I may be able to set it up for reverse order print. Thanks

You'll find that the site admins have been doing exactly this for awhile, by posting booklet information in the forum when they upload it. Perhaps the info can be on the test page itself, but this was available:

Liane - Next time it could help if you first print ONLY page 1 (if significant) or ONLY page 2 (or any other page you like), then you print the entire document - so the pc/printer will not format the entire document before starting printing.

Thanks Motris. I'd missed the information about the cover sheet - carelessness on my part. I'll be more careful to check next time.
Thanks Stefano for your idea. I'll certainly try that. It's sounds a much better option for me.

Very nice puzzle set. My favorite had to be the Shikaku. Thanks Tom!

With no more participants at this time, Nikoli Selection 2011 is over.

We'll take few more hours to review some of the wrong submissions, but the positions in top 10 will not change.

Congratulation to Hideaki Jo (all 19 puzzles), Thomas Snyder (18 puzzles), NAGATA Yuta (17 puzzles) for topping this test.

Indian results are not very encouraging, Rohan, Amit and Rajesh take top 3 spots.

Total 215 players participated, out of which 168 got non zero scores. This is easily the most that participated in any LMI 2011 puzzle test.

Detailed score with submission timings now available - as usual - at http://logicmastersindia.com/M201107P1/score.asp

Mind boggling performance from Hideaki Jo - he finished 3 marathons in about 42 minutes.
Apart from Thomas, volxa also finished 2 marathons, but with 2 mistakes.

Lot of players finished all Main section puzzles, and many more getting close to finishing them.

Thank you everyone for participating, and making this successful.

Thanks Tom for providing a very entertaining puzzle set, I'm sure everyone had some favourites among them. My personal favourites are LITS, SHIKAKU, NURIKABE and AKARI.

It was not surprising to see that 5 out of top 8 players are from Japan (just like last time). Of course, some players from German were missing, may be because of German Puzzle Championship on Saturday.

Yes, makes sense to put that information in the submission page itself. Will do that for future tests.

One funny quirk I'd like to point on the Kakuro: on the solving route I took, I had most of the lower-left to upper-right diagonal after a short amount of work, basically dividing the rest of the puzzle into two halves with only one blank cell in common. I noticed that the answer entry only looked at the upper half, so I applied the usual "row sums minus column sums" logic to get that blank cell and finished with basically half of the puzzle unsolved. In general I think one should avoid marking rows/columns on the edge of the puzzle if at all possible, since otherwise you're prone to allowing this kind of thing.

MellowMelon: I was fairly confident the solving route had to go something like bottom left -> bottom right (via some work on the lower diagonal) -> top right -> bottom left (via the other diagonal) and then finally end up in the top left. Nothing like this was mentioned in the test solving, although I am always a little wary about concentrating solution keys in one area. There were a couple of other puzzles as well where I thought something like that could plausibly happen, but again I don't think this affected many solvers. It's definitely something I'll think about more in the future.

Thanks all for taking part and for generally leaving such kind comments. I'm very glad that this contest went through without any hitches like last time, and really impressed with some of the scores. Motris actually had the main 16 out faster than deu (and indeed faster than I had managed to test solve my own damn puzzles!), but as I raised the point earlier on in the thread, if player X gets more of the marathons out than player Y then X is indeed the worthy winner in my eyes. I had thought it'd be impossible to get all 3 marathons out and was truly astonished to see that deu had managed it.

I hope there weren't too many people who were in the situation of finishing the main set but unable to get 1 marathon out - actually I think all things considered even getting the main 16 out entails a seriously good performance. Of course, the vast majority of solvers *didn't* finish the set, so I hope there was something in this test for everyone, and that you enjoy the remaining puzzles without any extra competitive stress.

Finally, thanks again to Deb for running the show seemingly effortlessly, and to Rakesh and David for testing.

For Suraromu, the answer key required the number of turns.
For row A, I submitted 5 turns, since I thought the 'turn' at the starting/finishing point is more of a 'make ends meet' as opposed to a turn.

Nice set of puzzles Tom, I liked LITS best.

I'm already as slow as a snail on puzzles, so when I also make mistakes...

I thought I'd add some puzzle specific comments now - specifically the oddities that made me laugh a bit:

For the second straight test, Tom actually wrote a LIT puzzle, not a LITS puzzle. But I was happy that the answer entry this time used the tetromino shape Tom is deathly afraid of (must have eliminated it from memory after bad Tetris experiences?).

I also appreciated the "not 2-9" row of the Kakuro which riffed on my 1-9 complaint from the last Kakuro Marathon. I did sort of expect the change though so had no problem resolving that corner which certainly just feeds as Melon points out one digit to the upper-left. For most Kakuro I like diagonal entries as the USPC does, but I'm not sure with the strong lower-left to upper-right diagonal if this would be an improvement on this puzzle. Probably two long rows and/or two long columns that target both isolated sides as opposed to a mix of rows/columns that only hits one.

After breezing through the actual test, I got caught with a small mistake on a huge puzzle of a type I don't enjoy solving. And tweaking a mistake in a large puzzle is a whole different challenge from just erasing and restarting as in a small one. While the small Yajilin used some nice loop properties, the big one in my opinion was just long albeit with a nice symmetric pattern. I couldn't recover enough to finish it within 90 minutes (100 minutes is a different story). Fortunately I switched to completing the Heyawake in time to not completely botch the second contest. Still, if there are to be marathons next time, I'd prefer more balanced scoring so solvers can go after them at any time. It is odd that deu and I both competed in some other test after the first one ended in 45 minutes. Congrats to H.Jo for his marathon solving ability; I'd obviously used up too much energy expecting the contest to be a sprint and lost in the last miles of some different race.

Edited by motris 2011-07-11 3:04 PM

This obviously wasn't Nikola Selection .

Great performance of top two players, I haven't even got to see a marathon puzzles. Time was up in front of my eyes when I try to enter last "regular" puzzle. However, I will keep this test in my memory because here was one beautiful puzzle - suraromu.

Nikola

Thanks for a great contest!
In Nikoli magazine and Nikoli.com puzzles, some difficult techniques are not used.
So, at least for me, this contest was a good chance to solve Nikoli puzzles which never appears on Nikoli
(Yajilin (small) and Fillomino are typical examples).

As for my performance, I feel very satisfied to achieve what author did not expect.
After entering and checking my 16 answers, I had 40 minutes to solve three marathon puzzles.
When I solved Heyawake in 12 minutes, I became convinced that there was enough possibility of finishing all puzzles if I made no mistakes.

It was a nice contest, and there was something for everyone. And this one turned out much better than the first edition. I hope Nikoli Selection 2012 will be even grander.

Congratulations to everyone who finished all 16 from the main section. They are all winners of this contest. And special congratulations to those who managed to solve even one marathon !!!

From the main section, I liked Nurikabe, Hitori, Shikaku, and, of course, Suraromu.

Very fun test, with some really good puzzles. Nice sudoku, excellent suramoru, ripple effect and LIT(s), amongst others.

I made some stupid tactical choices, like wasting the last 10' on the Hashi, a puzzle I dislike, instead of solving the Numberlink and checking my Fillomino on which I was almost sure of having done something bad - and indeed, I had. But all in all, a satisfying performance for me.

Thanks Tom, and big congrats to Hideaki for such an impressive performance.

I agree, very fun puzzles. During the contest I had to rely a little bit on intuition on some puzzles (the Yajilin which I've solved properly afterwards, the Fillomino where I should probably have summed up the clues at some point, and also the Nurikabe. I still need to revisit that one.

Just now I looked at the marathon Heyawake, and was wondering if anyone else is having trouble with uniqueness? I solved it properly until just a patch in the bottom right and the top right are not filled in. Aside from room 'B' which is still open, my solution agrees with the correct code. It looks like the bottom right should close up so that connectivity forces a path along the right edge, forcing the top right including B=1, but I think I have a valid solution with B=2…

Most likely I'm missing some wrong room count or three-room violation, but I really can't find any at the moment. (A uniqueness argument forces B=1, so there's a very small chance an ambiguity would have been missed in the tournament.)

Edited by rob 2011-07-13 2:47 AM

I shall post a solution...give me a sec to do it in paint :)

Here we go - not quite in Hideaki's time, but I did have to faff around uploading and stuff!

http://www.warwick.ac.uk/~marear/36x20_heyawake_abc_solution.PNG

Thanks, that helped! A 4-room some distance away had just three black cells… I'll give it a fresh try tomorrow.