I'll let Fred put more comments later (and Fred, correct me if I'm wrong), but from our earlier discussion, I figure that the intention was that top players should be able to solve around 100 minutes. At least the test solver's timings suggested so.

Prasanna was the first player to take the test, and looking at his score, we really thought that it is possible for top players should be able to solve better than what the final results suggest. Little did we realize that Prasanna has improved a lot in recent times.

I realize that a statement like "It is expected that top players will be able to solve all sudokus earlier than 120 minutes." look certainly misleading now. [ I had put it originally in Sudoku A/B, where I thought I should let players know that it is more like a Sprint test and not otherwise. ] May be because of that line you started solving from the end, expecting that you are anyway going to solve all sudokus. Obviously, Fred's intention was never to mislead any player.

In my case, it is not because of the toughness of the grids. I somehow did not manage to find a two hour window for the test. And since there were only a few hours left yesterday, even though I was sleepy, I started the (online) test hoping I would be able to stay awake the next couple of hours. But I did not manage it and decided to stop after hardly ten minutes - one grid (anti diagonal) was close to completion but I did not submit as I was anyway going to stop.

Congratulations to all winners and Prasanna for an impressive finish !!

Edited by rakesh_rai 2012-03-13 11:30 AM

Thanks Rakesh.

Coming to the difficulty, I tested all the grids after I completed the test. To be honest, I too expected a lot of people to finish this one, for 3 reasons.
1. I don't really think diagonals are my strength, and I'm never comfortable solving them.
2. I was about a few seconds away from submitting the Diagonal Consecutive, which means there were just 3 grids I didn't solve in the allocated time.
3. Later on, I solved the arrows one quickly enough considering its points, I took long on Argyle and Creasing but I always knew I was gonna get extremely confused on those and thats why avoided them during the test, but since they were low pointers I assumed they were easily solvable for the top players too.

Anyway, all 3 of us Deb, me and Fred expected many to finish this. I'm glad on the personal side as I've moved up, but from a tester's point of view, I apologize that I didn't really give a good evaluation of the grids, but that was not at all intentional.

Edited by prasanna16391 2012-03-13 2:18 PM

All compliments to the author. This was the best sudoku test here, premium puzzles, all of them with very nice solving path. I especially liked new outside and skyscrapers variation. Bravo Fred!

I agree with Thomas, the sentence "top players will be able to solve all sudokus earlier than 120 minutes" should not be here, nor in any other test. This should be the default. Anyway, I don't pay too much attention to it, but someone else could be frustrated. Congrats to Florian!

Nikola

Thanks, Nikola !
I'm happy that you liked the grids !

Congratulations to Florian, Jan and Tiit for taking the 3 first places of this hard test !

Florian also win a beer, as the best German player. I'm a guest of the German sudoku championships in Dusseldorf, so I wanted to do this special price to the best German player. I think he deserves it, winning his 1rst test on LMI. This can perhaps explain the fact that there are 3 German players (it's well-known that German people like beers, ) in the top 10
Tiit played just few hours after becoming Estonian sudoku champion ! Congrats !

Thanks to Deb and LMI team to have hosted this test and the great work they do for the puzzle and sudoku world !

I have to apologize: the test was harder than what I thought. Initially I sincerely thought that 5-10 players can complete the test. Now I can see how the sentence in th IB: "It is expected that top players will be able to solve all sudokus earlier than 120 minutes." was a bit pernicious and could have created some discomfort while playing.

My mistake was to have asked only one tester for the points distribution.
Initially, Bastien tested 13 grids, with the untouch posted on my blog, which is a ~10 minutes grid for a top player. The 1rst version of the pointing evens was harder, also a ~10 minutes grid.
Bastien's sum of individual times was around 120 minutes, So I decided to work on the pointing evens, so that it'll be easier, and not putting the untouch. So the sum of individual times of Bastien was around 105 minutes. I know that the sum of individual times is not the reality of the competition, but he said to me that he hasn't played for 3-4 months (since WSC), so I thought that the global difficulty of the test was now ok. I must say 2 things:

-well, without playing 3-4 months, Bastien seems to be always very competitive
-It's been long time now that Bastien and me work together, testing each other grids for tournaments, for publications, etc... So he knows my work very well, which eventually gives him an advantage...

Well, I want that you understand well my comments: the difficulty of the test was not the fault of the tester, and I want to thanks Bastien here for having tested these grids !
(His time on the argyle was really 4'04. With more testers, perhaps this grid would have more points).

I'll not do a psychoanalysis to understand why I created so difficult grids, haha. But just a few notes:

-Normally I like when there are a few (1 or 2) hard grids, giving really some works for the top players. Easy grids are ok, but when a test contains only easy grids, I find that it's a speed contest, just playing very fast, and this real logical work finally miss. I see now that there were too much of these grids in the test: of course arrows battle, diagonal skyscrapers, diagonal consecutive (I'm not fast on these grids and generally on consecutives, seeing the results of the recent Krtek's round on fed-sudoku shows that the timings of best players are significantly better than mine, so this let me thought that this grid was hard, but ok), diagonal twin, and to a lesser extent Queen and antidiagonal.
-I created the test grids in november, just after the WSC, where my result was not really good. Unconsciously, perhaps I lost confidence on my sudoku solver's skill, and it pushed me to create hard grids ("if it's hard for me, it doesn't mean that it's hard for top players"). But I don't want to go further with these psychological considerations, haha



Prasanna, in my point of view, your role in taking the test few hours before others is not to evaluate the difficulty of grids. Even if you did a poor results, it was not the time to change the grids . I see your role more as the last check that everything works well, and eventually the last occasion to see if there is a problem with one grid (multi solution, or no solution). Thanks for having played this role and congrats for your very good performance.

I promise that the next test I'll author will be much more easier. Grids are already created and tested: it'll be a swiss qualification tournament, which will be hosted by the German LogicMastersDeutschland site. I'm not sure that a swiss player can finish all grids in time, but I'm sure that lot of international players can do it

Fred

Wonderful summary Fred. I'll try my best to make it for the next competition you say you've authored. I have a whole bunch of important exams coming up though so can't really be sure of anything.

As for the role of the tester, that is true, but I suppose if there was a more accurate view of my level globally, we could probably have stopped that sentence in the IB which has caused a bit of a problem. Thats all :) Thanks, I do like testing before time, since I rarely ever get time on weekends for anything

deu posted an interesting statistics in his blog - This is the only the second LMI monthly test where there is no players from Japan in top 10 (the first one being the very first monthly test Master Mind Twins almost 2 years back).

---

flooser won an LMI test first time. His previous best rank was 9 (incidentally another Fred's test). Some other players improved their best rank too (list below has only rank <= 20)

Player	Rank in Diagonal Vision	Previous Best Rank
flooser	1	9 (LMI Spring Sudoku Test)
sinchai4547	4	6 (Crazy Arrows)
tomek_s	6	40 (LMI April'2011 Sudoku Test)
smat	9	18 (Shapes and Sizes)
prasanna16391	15	28 (Best of LMI Mocks)
EKBM	17	41 (Best of LMI Mocks)

My main problem in this test is that the hard parts of the grids were all in the opening steps, even for the easier grids. So if you don't see the opening steps, you can't make any progress at all. There's a difference between a hard step when you have 25 digits in the grid or 45 digits in the grid. I had all these empty grids constantly where I was just looking for an opening which I couldn't find. It feels a bit frustrating to not be able to put any digits in. How much earlier the trickiest step is, the bigger the variation in solving times will be.
I started with the Argyle and couldn't solve it logically, which is why it took me almost 20 minutes to even submit a single grid after I decided to give up on it for now and go solve the diagonal instead.

They were nice puzzles, I just got stuck without making even a decent start too much, which was frustrating. My favourites were the Queen and Pointing Evens as I actually solved those 2 in a good flow, many of the others it took me too much effort to start.

Bram, you're right about the opening of several grids. I think that I like these hard openings. When a grid is already half-solved and need a hard step to be finished, I think it's easier to bifurcate and to finish the grid without taking the hard (but perhaps interesting) logical step. When it's in the beginning of the grid, I think it has better chance to showcase solver's skills. But of course, it creates lot of discomfort if you don't see it !

Abour argyle: I'm sure a player like you is able to solve this grid logically. It really solves without real hard steps. Perhaps you missed a naked single or a pair in a marked diagonal, I think these are the hardest steps needed to solve this one. The opening of this one is not hard, one can complete almost fully boxes 4-5-6 without any effort. After that there are a few things to see, that's right...

I'll perhaps post some explanations in the coming few days about some grids, especially the openings (diagonal, creasing, twins, little killer...).

Fred

I hardly ever bifurcate, especially in sudoku. Bifurcation just leads to errors for me. Showcased in philadelphia when the last round you were basically told to bifurcate and 2 of my 4 solutions had errors. I understand the early harder steps will more avoid bifurcation, but which isn't bad for the higher point puzzles but maybe isn't necessary for all puzzles. It comes over a bit weird to me when my fastest solve is the 85 point puzzle.

For the argyle I missed a hidden single in row 7, after I checked to see what I missed. I was just thinking it would have been nicer for there to be something easier. The easiest puzzle being 4-5 minutes for the faster solvers seems very long for me. But maybe I'm just more used to puzzle competitions which generally have more puzzles than Sudoku competitions and the easier puzzles take less time. It's just a personal thing.

Some little explanations on some grids:

1. Diagonal sudoku

Preliminary remark: The 1rst glance suggests that the opening has something to do with the first diagonal (R1C1 to R9C9), with lot of clues in boxes 1, 5 and 9.

There are a couple of easy placements: 7 in R4C6 and 3 in R6C7.

One can view the clues 134 in boxes 5 and 9, which leads to have a triplet 134 on the diagonal in box 1. Then you can place the 7 and 9 in box 1. Then you can place some other 7's, and then some 9's, which leads to a pair 89 in the central box.

After this point you can solve box 9, and then, etc... finish to solve the grid





Attachments
----------------
diago.png (33KB - 0 downloads)

2. Creasing sudoku

Preliminary remark: This sudoku has a hard opening, because you have to think "globally" about the configuration of marked diagonals, especially 5-cells diagonals !

There are a couple of easy placements: 8 in R2C7 (can't be in C8 because no greater digit can be in this marked diagonal) and 6 in R8C8.

Let's analyse a bit the 5-cells diagonals. If you follow the path of these marked diagonal, it has to alternate between increasing and decreasing diagonal (If you had 2 increasing diagonals following each other, you should have digits 1-9, and then it would be impossible to have 2 decreasing diagonals from 9 to 1). So digits in R1C5 and R9C5 have to be a) both greater than 5 or b) smaller than 5. Ditto for cells R5C1 and R5C9.
We can see that only one digit greater than 5, the 7 can be placed in R1C5 and R9C5. So these cells must contain digits smaller than 5. Thus Cells R5C1 and R5C9 must contain digits greater than 5. We have found the direction of increasing and decreasing diagonals !

Knowing that, you can place the 8 in R5C1, then in R4C8, then the 9 in R5C9 (must be greater than the 8), etc...





Edited by Fred76 2012-03-16 8:51 PM








Attachments
----------------
creasinga.png (24KB - 0 downloads)
creasingb.png (25KB - 0 downloads)

Let's speak a bit about little killer:

Preliminary remark: The grid is constructed on the 6-cells diagonals !

Of course you can place digits on the 4 corners of the grid (I try to delete these clues, but I think it was hard enough with them ).

One can see the 6-cells diagonal with clue 48. That's the maximum possible. Cells have to contain 2 triplet 789. With these triplets, both 6-cells diagonal with clue 16 are then the minimum possible. So you can place pairs 12 and 7 in one box and a triplet 123 in the other box.
And finally, with these triplets placed, you can see that the last 6-cells diagonal, with clue 20 is now the minimum possible, with the sum of two 145 triplet which can be placed.
One fun placement now is the 2 in the middle of the grid (fun because it's rare to be able to place a digit in the center of a little killer in the beginning of the resolution).









Attachments
----------------
littlekillera.png (31KB - 0 downloads)
littlekillerb.png (34KB - 0 downloads)
littlekillerc.png (35KB - 0 downloads)

Finally (then I stop and let you concentrate on the tapa contest ), I want to speak about the diagonal twin sudoku.

There is (at least) 2 openings for these grids. You can begin with the corners. With grid A you can place 8 in R1C9, 4 in R9C9, and Grid B let you know the two other corners.

But there is another opening, which I find more elegant: You can start with the center of the grids: box 5. Grid A let you know that digits in white cells are 3,5,7 and 8. Knowing that individually they can't be placed in the same cell let you place the 5 in R5C6, the 3 in R5C4, then the 8 in R6C5 and 7 in R4C5 in grid B.
Then in both grids, 6 can be placed in R5C5, you have a pair 14 in cells R6C46, then 2 in R4C4 and 9 in R4C6.

Of course, you'll need still lot of work to solve these grids, but box 5 is almost filled





Attachments
----------------
diagotwin2.png (18KB - 0 downloads)