Richard -

My compliments on the well-thought out and symmetrically pleasing puzzles; thanks for the test.

Dave

Copy Paste is over. Congratulations to TiiT, jaku111 and Kota for taking the top 3 positions.

Richard will share his thoughts soon.

First of all: TiiT, jaku111 and Kota: congratulations with the top spots!
Amazing and impressive results!
All other 258 players: thanks for participating!

I am very delighted with the ratings for the puzzles. Especially for ‘Equal’, although I think it is amazing that it is rewarded 9 times with a score of 10!

Let’s go back to Shapes and Sizes. Based on the results of the test-solvers, I really had the idea back than that a few of the world-top-players would be able to finish all grids. Although it was close, nobody was able to do so. So after the test was over, I did a fair bit of evaluating and came to the conclusion that:
- none of the puzzles on itself was too hard;
- all in all it was too much;
- it lacked a few really easy puzzles, since the list of players who solved only one or two grids was pretty long;
- working with irregular grids was probably the reason for errors by top players and reason for not finishing the complete test. But that’s intrinsic to irregulars.

These conclusions were the ingredients for my second test considering that LMI expects authors to write tests which should be finished by few top players..

About the theme
Since almost five years I make a set of sudoku variants every four weeks for a Dutch/French sudoku magazine for a hobby. The set always contains 4 consecutives. After a while I started to make two different puzzles with the same given digits every now and then. Since the puzzles are published two on a page, I thought it would be nice from an aesthetic point of view to make such a combo once in a while. When I started thinking about a second theme for a LMI-test, this combo-idea crossed my mind pretty soon. Shortly after SaS was over, I suggested this theme to Deb, who was enthusiastic straight away.

I expected it not to be easy, since some of the constraints in the variants are very ‘demanding’. Therefore I chose to use combinations as much as possible where one grid was very ‘demanding’ and the other was much easier to construct. Per page it’s mostly very easy to see which was of the first and which of the latter category.

With the conclusions from SaS in mind I started making the sets of puzzles.
I wanted to compose a set of puzzles that I was able to solve myself in about 2/3 of my own times for SaS. By doing so, I hoped that about 10 players would be able to solve all grids in the test, and lots of players would be able to solve between 5 and 10 grids.

After I finished all 14 sets I decided what puzzles had to be in the IB and, ergo, what had to be in the real test. I wanted to have at least five relatively easy puzzles, a wide range of difficulties and two or three hard puzzles. All test solvers reported that they needed 25% to 35% less total time than for SaS, so that confirmed my feeling this test would be better solved.

I have chosen mainly types that I am familiar with, but wanted to include also a few types that I only made a few times earlier (Minimax and Low) and a completely new type (Equal).

Here are a few words about some of the puzzles:
Chaos Diagonal and Scattered
Unfortunately the puzzles that I liked better from an aesthetic point of view turned out to be too hard for the real test. So they were abandoned to the IB. I wanted one of them to be in the easy category and one in the middle category.

Consecutive
I am really fond of this type. It’s so much fun to work with 5 odd and 4 even digits in a puzzle and having a symbol that says: on either side of me you’ll find one odd and one even digit. For the real test I wanted to make a puzzle with very few circles.

Minimax
For earlier puzzles of this type I didn’t place givens in the grid. It’s not necessary anyway. But with givens and not too many clues outside the grid it was an nice low-pointer.

Ascending
Longer lines provide a fair number of digits to place so that the puzzle solves pretty smoothly after that.

Ascending nonconsecutive
I preferred to use longer lines here too, but the nonconsecutive constraint doesn’t really cooperate. The way to get into this puzzle is of course the grey line in row 5. You can fill in the digits 97531 on that line immediately. After that the two series of 4 digits in rows 3/4 and 6/7 are helpful. Since 1 and 2 cannot be placed in R6C7 and R7C9 you know instantly that the asked order of digits is 3579.

Maximin
For this type you can either work with big differences or with a lot of givens to provide a good start and a smooth solve. One trick that might have been helpful is the series of 3 2’s outside the box top right. This implies that you have to work with the trios 123, 456 and 789. This can be a helpful shortcut to come to sooner placements.

Low
Since I like to work with symmetries this puzzle type gave me a bit of a hard time. I thought it would be nice to use a ‘dice’pattern for one of the puzzles (IB), and connect all grey cells in the other puzzle to give it a nice look.

Position
The corner blocks provide the most info for this type.

Renban
What is very helpful for this puzzle is to notice what digits have to be in a renban group, and what digits cannot be in a group. You can thus place a 3 in R6C9 straight away. Also helpful is that both renban groups starting in R7C2 and in R9C1 need a 6 as well as a 7. In nonet 9 the 6 and 7 are placed somewhere in R8.
Last but not least: two groups that are in a ‘hooked’ pattern cannot contain the same series of digits. There has to be at least one different digit. This rule was very useful in determining the series that started in R1C6 and in R6C9.

Equal
For the test-solvers, Equal was just a bit harder than Renban. But Equal gave players a much harder time than I expected. At least, from what I could see ‘looking behind the screens’.
Very important in this puzzle is to realize that in every cage an even number of odd digits has to be placed. Also in the cages with five cells. In the cage in box 7, a five is included. So the 3 stays out of the cage, giving a sum of 14 (5,9) for three even digits: 2,4,8. The 3 and the 6 go in R9C23.
The cage of four cells in block 1 needs a 4 and a 6.
The cage of three cells in R123C6 needs a 1 since 3,5,8 is not possible.
The most important step is now to realize that the cage of 3 in R6C67 + R7C6 needs a 1. The combination 3,5,8 is not possible, since 8 would go in R6C7 and 3,5 go in R67C6. This blocks both possibilities in the cage of 3 cells in R123C6.
So 1 goes in R6C7.
From here it should run smoothly.

Once again: thanks for participating, and I would really appreciate to read you comments to the test!

Awesome, thats my first one! :)

My favorite was in fact the Equal. I didn't find it hugely difficult, it had a nice solving path. Then come the Renban Groups(although that was one of my mistakes during the contest duration). I really liked the use of row 3/4,5,6/7 as you mentioned in Ascending Nonconsecutive. The Consecutive and Scattered sudoku were among the favorites too.

Thanks again for a fine set of puzzles. :)

So how many times did you refresh the score page? :P

There was a lot of refreshing - It's been a loong weekend for me. That's the result of doing the test on Friday.
In addition, I want to say thanks for the practising material - I solved all of them and It was really useful !!! :)

TiiT

It is always interesting to see new winners. Congratulations!

Thanks for the wonderful test ! Every sudoku was very nice and fun to solve !

I was not very comfortable with consecutive and with renban. Equal was beautiful, hard but it can be solved smoothly, I really liked it !

When I saw the renban, I immediately recognized the configuration of the renban of DSM, which was much easier than this one. During solving, I had all the time this little voice in my head saying: "Why are you stuck? you know this grid, it's an easy one, you should crack it without problem !" creating a little discomfort and I really had the impression I was not good on this one (which was perhaps not the case).
After the long grid "equal", I just wanted to pass the sound barrier on the easy "low", but as often is that case, I did a stupid mistake at the end... not very important, but it cost me 2 ranks.

Fred

I liked the puzzles. Equal was my favourite too. I messed up on the easier puzzles. got stuck on the all odd/even for far too long and made a mistake in the Low. I ended up not solving it as I switched to try to solve the Equal instead of solving the Low again. It seemed to have been the right decision as I managed to finish Equal it just in time.

Maybe it's good to write something about the practising material.
A lot of the regular participants in LMI sudoku tests are regular visitors of the puzzle portal of LM Germany and/or do also participate in the German Sudoku championship. So they might be more used to my 'style of puzzle making' (if I have one...) than others. To give all players the opportunity to practise some more of my puzzles, I think it's fair to provide the links to this practise material.
And meanwhile I hope to attract some players to the puzzle portal Germany too (of course). There's a lot of very nice puzzles out there, from various authors. (Not only sudokus BTW.)

It's good to read that it's helpful in at least one case!

Nice set of puzzles.
I was able to solve only 4 during the competition, but could not submit the fourth because I ran out of time.
Now I am taking my time to solve other puzzles.
Renban was very good although I have started from the R1 and R3 Renban groups. As you said "What is very helpful for this puzzle is to notice what digits have to be in a renban group, and what digits cannot be in a group", I did the same thing and puzzle was solved very smoothly.

Still more to solve.

Thanks Richard and LMI for wonderful test.

Regards,
Ravi

Let me first thank the author for providing the sudokus.
I'm quite new to these competitions, that might also be why I missed something, as I cannot see why there is a correct solution to one of the simpler sudokus: all odd/even
Let me explain, and hopefully one of you can prove me wrong :)

The grey fields either contain all even or all odd numbers. There are 5 odd numbers (1,3,5,7,9) and 4 even numbers (2,4,6,8) possible. Look at the rightmost column, there are 3 empty grey fields and even numbers 2 and 6 already present. That means that the 3 grey fields in this column cannot be filled with all even numbers (as there are only two left: 4 and 8). Hence, all grey fields should be odd-numbered.
Now look at the second row, there are only two non-grey empty fields, which are located in the top-middle 3x3 square. Number 4 is still missing in this row, hence it should be filled in somewhere in the middle 3, however there is already a 4 present in the top-middle 3x3 square. Hence, the grey fields cannot all be odd-numbered. A contradiction!

edit:
oops... I now see that the restriction on odd/even should be interpreted per 3x3 square... sorry

Edited by gjdv 2012-07-12 2:01 AM