Country : Switzerland
Fred76 posted @ 2011-11-12 8:31 PM
I'd like to open a discussion on the manipulative sudoku of part 2 of WSC 2011.
Here are the rules:
"This is a manipulative Sudoku puzzle
where small pieces with pictures are to
be used instead of numbers. Standard
Sudoku rules apply. In addition, some
pieces are already placed into the puzzle
grid. None of these pieces is at the right
place. However, for any piece placed into
the puzzle, at least one of its (at most)
four edge adjacent neighbour cells must
contain the same piece in the solution as
the cell itself has in the puzzle grid.
E.g. in the sample puzzle, row 2, column 6 of the puzzle grid has a shape similar to a forward slash. This
implies that in the solution grid, at least one of the neighbours of the same cell must contain the similar shape.
Partial scoring is available in this round, depending on the number of correctly placed pieces: up to 21:
0 points, between 22-80: (number of correctly placed pieces - 21) * 3, 81: full score (180 points).
Any piece that is placed onto the grid but is incorrect, results in a deduction of 3 points (in other words, the
number of correctly placed pieces will be reduced by the number of incorrectly placed pieces for the scoring
purposes). If the total score is less than zero, then zero points will be awarded (i.e. nobody will finish the round
with a negative score).
No pencil, eraser or other marker is allowed in this round. Individual name stickers will be provided prior to
this round, competitors are kindly requested to bring it along and use them to label their solution sheets just
before this round starts."
and here is the starting picture, on which we had to put the pieces:
There are many starting points to begin the puzzle. I'd like to show that the logical path you choose could affect the difficulty of the puzzle. The point is: if you put some pieces on crucial clues, then it'll be hard to continue, because you had to raise some pieces to see important clues.
We begin by studying this piece:
The presence of such clues in R6C34 and R7C34 looks interesting. There are 2 possibilities in regions 3, 4, 7 and 8. Not very easy to see, but the same clue in R4C4 can solve the situation. These pieces must therefore be placed in R5C4, R6C2, R7C5 and R8C3.
A easier way to solve this is to look at these pieces:
in column 2.
The only possibility to have a neighbor for the clue in R6C2 is to put one such in R7C2 (otherwise the clue in R5C2 wouldn't have any such neighbour). Here we are:
With the piece placed in R7C2, we can directly solved the first piece I talked about:
in R8C3, then regions 3, 4, 7 and 8 look like this:
Let's forget this for a while, return to starting point and concentrate on other pieces:
1rst, this one:
We can easily see that the clue in R5C1 must have a such neighbour in the solution in R6C1.
Now look at the three clues
in column 4.
One can easily place one such on R4C4 (neighbour for R3C4, otherwise, one of the two others will not have such neighbour).
The situation becomes:
What is interesting with this last picture is that we hide exactly both clues that were needed to solve the clues in row 6 and 7. You have now to raise pieces you've put to see these crucial clues.
So if you begin with pieces and , it'll will be more difficult to see the other starting point with and . The reverse is not true !
I think many such situations could happen while solving. If you're lucky, pieces you put doesn't hide important clues, and if you are unlucky, you get stuck and has to raise some pieces to see how to continue (but which pieces?).
Considering that, I asked myself if some players just studied the puzzle for a while before starting to place pieces on the puzzle and take the risk to hide some important clues.
@ 2011-11-13 2:10 AM (#5953 - in reply to #5951) (#5953) Top
Country : United States
cnarrikkattu posted @ 2011-11-13 2:10 AM
I didn't have the pleasure of solving with pieces (solved the A-I version in 21 minutes), but I think there are some ways to get around the covering issue. You could put pieces in different rotated orientations to mean different things (such as "the covered clue is still important"). You could also offset your placement so that part of the underlying picture is still visible. When I was solving on paper, I crossed out the small clues as I satisfied them, and placed my letters such that those clues remained visible otherwise. Either-or notation on the boundary adjacent cells would work (until you ran out of a piece). There were a lot of either-or in cells two cells apart, though, so I'm not sure this strategy would have worked well live with physical pieces.
@ 2011-11-13 2:19 AM (#5954 - in reply to #5951) (#5954) Top
Country : United States
motris posted @ 2011-11-13 2:19 AM
I'll say that my approach was to position the pieces off-center in the grid, to keep the right borders of all pictures underneath visible. This let me work through both sets of clues for as long as I needed before just going to standard sudoku placement.
I've commented elsewhere that the round would have benefited from solving on a separate grid, both for the reasons you bring up, and to have made the grading much easier (you'll know every place a solver has placed a photo much more readily!). But there were certainly approaches that wouldn't have fully removed the clues you needed.
@ 2011-11-13 2:32 AM (#5955 - in reply to #5954) (#5955) Top
Country : Switzerland
Fred76 posted @ 2011-11-13 2:32 AM
motris - 2011-11-13 2:19 AM
I'll say that my approach was to position the pieces off-center in the grid, to keep the right borders of all pictures underneath visible.
I simply didn't think about this approach, which is quite good
Edited by Fred76 2011-11-13 2:32 AM
@ 2011-11-13 3:24 AM (#5956 - in reply to #5954) (#5956) Top
Country : France
Ours brun posted @ 2011-11-13 3:24 AM
motris - 2011-11-12 10:19 PM
I've commented elsewhere that the round would have benefited from solving on a separate grid, both for the reasons you bring up, and to have made the grading much easier (you'll know every place a solver has placed a photo much more readily!).
I wish so much that it would have been the case. I repeatedly made mistake on mistake by confusing the pieces and the given clues before trying an approach very similar to yours, but at this point I was just so much frustrated and stressed by the perspective of messing up the second round after having been pretty slow on the first one that I was never able to enter in the puzzle during the last minutes. The puzzle itself was not a too bad idea, there was definitely some originality in this, but the exploitation of this idea could have been way better.
Anyway your own solving time was quite impressive - and added much to my stress, by the way. Thanks for that !
(Just kidding of course, but true nevertheless)
@ 2011-11-14 12:46 PM (#5960 - in reply to #5951) (#5960) Top
Country : Hungary
wicktroll posted @ 2011-11-14 12:46 PM
I used a similar approach, I placed pieces so that they covered the bottom half of the cell where the piece should be placed and the top half of the cell below (e.g. if the piece went to R3C4 then I covered the bottom half of cell R3C4 and the top half of R4C4 in the grid). It worked fine until 20- pieces, but it became quite confusing when I placed more pieces in the grid (and I forgot 2 or 3 times that pieces belonged to the upper cell in the grid...).
One of my teammates mentioned after this round that it could have been a good strategy to place 30-35 pieces in the grid then turn all of them so that the numbers become visible and solve it as a regular Sudoku puzzle. That sounds a good idea but quite risky because it might happen that having 30-35 numbers in the grid isn't enough to solve the regular Sudoku quickly and in this case you still have to raise the pieces... I didn't check this technique yet but I'll give it a try later.
Edited by wicktroll 2011-11-14 12:48 PM