The New Sudoku Players' Forum

[tso]

Some *hard* Sudoku variations that have appeared in Internet Puzzle Competitions:

From PQRST 11, puzzle #5:




From IPST 2005, 2nd round, #6
Every figure contains digits from 1 to 9. Place these figures into the grid so that all digits in every row and column will different.



From IPST 2004, 1st round, #4
Put numbers from 0 to 9 into the cells, so that every row, every column and every marked region have not repeating number.







PQRST stands for Puzzles Quarterly Rate Solve Ten. As it's name implies, it is Quarterly, competitors must rate the puzzles as well as solve them -- and there are ten puzzles each time. Users have 7 days to submit answers. There is no cost, no prizes other than getting your name on the list of high scorers. All past puzzles and answers are online.


PQRST 14 starts on August 6th, Saturday at 20:00 (GMT+02).


IPST stands for Internet Puzzle Solving Test. It is also quarterly, has no entrance fee or prizes other than bragging rights. Again, you have 7 days to solve. This contest is run by Diogen, a Russian puzzle club, so the translations into English can add an additional layer of compexity to solving. All past puzzles and answers are online.

[dukuso]

the first example is a 6*6 sudoku-variant, but instead of giving
preplaced clues, you have a list of (digit,cell)-pairs where the
digit can't go into that cell.
Solving it should be similar to solving sudokus.

the 2nd is quite different from sudokus. You have to rearrange
some geometric shapes such that the given digits won't conflict.
You are not to fill in new digits, as I understood.

the 3rd is a sudoku-variation but not each of the constraint-area
has to keep all digits. So this is not an exact cover problem,
or only one with "some primary and some secondary columns".
Solving it should still be similar to solving sudokus, I assume.

I haven't tried to solve the puzzles yet, though.


Guenter.

[Nick70]

the first example is a 6*6 sudoku-variant, but instead of giving preplaced clues, you have a list of (digit,cell)-pairs where the digit can't go into that.

Another way to see it is that you are given pencilmarks instead of hints.
Clever idea.

[tso]

the 3rd is a sudoku-variation but not each of the constraint-area
has to keep all digits. So this is not an exact cover problem,
or only one with "some primary and some secondary columns".
Solving it should still be similar to solving sudokus, I assume.

Guenter.

I should have mentioned -- the diagonal string of shaded cells is ignored during solving. It's only used to submit an answer. In the solution, each row, column and marked 10 cell area will have the digits 0-9. This is actually a common type of Sudoku variation. One advantage of using irregularly shaped areas instead of square or rectangular regions, is that it allows puzzles of any size such as 5x5, 7x7, etc.

[tso]

the first example is a 6*6 sudoku-variant, but instead of giving preplaced clues, you have a list of (digit,cell)-pairs where the digit can't go into that.

Another way to see it is that you are given pencilmarks instead of hints.
Clever idea.

Funny I hadn't thought of it that way, but you're right:

Code: Select all

(.23456)(...456)(123456)|(.23.56)(1.3456)(...456)
(.23456)(123456)(123456)|(1234..)(123456)(123456)
------------------------+------------------------
(.23.56)(123456)(.23456)|(...456)(.234..)(...456)
(.2...6)(.23.56)(.23.56)|(123456)(.2...6)(123456)
------------------------+------------------------
(123456)(.2...6)(.2...6)|(123456)(.234..)(.234..)
(.23456)(...456)(...456)|(123456)(...456)(.2...6)


(I think I transcribed that right -- but I'd double check for typos before solving.)

I'd like to see 9x9 puzzles with no hints, just candidate lists. Something like each cell has just two or three candidates, or each of 24-30 cells has two candidates and the others are empty. You might have start right off looking for forcing chains.