Unsolvables Quote:
In August 2007 the original Unsolvables thread was posted on Ruud's site
Glyn: Perhaps we need to collect all these unsolvables together, as we have a much smaller sample set of Killers to work with than vanilla Sudokus.
Para, in Unsolvable thread lead-in: On Glyn's suggestion, i browsed the forum for all puzzles (at that point) considered unsolvable....Here is the first list of unsolvables
This was later expanded from 9 to 14 puzzles; 10 of these have since had human walkthroughs posted, either by individuals or as "tag" solutions, and another was solved by JSudoku. The original Unsolvables thread is in Archive B
here. If you manage to solve any of the remaining ones, please send your walkthrough to
Ed for inclusion in the archive.
I am now posting an Unsolvables list for puzzles in this Killer forum which haven't (yet) had any walkthroughs posted. Each entry in the list has a link to the original puzzle.
If you manage to solve any of these puzzles, please post your walkthrough
in the original puzzle thread. I will then add a link in this post.
1) Nasenbaer's Assassin 114 V3 was posted
here.
In the introduction to his Twin Killer, Nasenbaer wrote:
While trying to create a decent Assassin with the now familiar design I stumbled upon this one. If you add the diagonals constrain to the left Killer you actually have a Killer-X with a unique solution. But SSolver(3.2.1) rating is 4.13, with extended T&E (several Bowman's Bingo). So I wouldn't advice you to try it, unless you are a complete masochist.
Thanks to
Børge for providing the diagrams and code string for this variant.
2) Børge's Assassin 178 V3 Short Brunette was posted
here.
Børge posted a series of related puzzles for Assassin 178 with the collective title
Blind Dates. This puzzle, and the one below, are the two hardest ones. Having solved the earlier ones, some with difficulty, I decided not to try these two.
3) Børge's Assassin 178 V4 Tall Brunette was posted
here.
Probably even harder; two pairs of cages in R789 have been combined.
4) Børge's Assassin 179 V13 was posted
here.
5) Dan(h3lix)'s FNC Killer V3 was posted
here.
This one is cheating a bit. I've got it in my Unfinished folder.
Dan posted one of his puzzles in the Killer forum before posting other Killer variants in the Other Variants forum. If anyone solves this puzzle and posts a walkthrough in the Other Variants forum, please let me know by PM.
This Killer is Ferz Non Consecutive (Diagonally Adjacent Cells Cannot Be Consecutive)
Removing two more cages (from the V1.5) makes a massive difference. As well as removing two cage totals, it also removes two useful 45s.
Børge’s diagrams for all three versions of the FNC Killer are posted
here.
6) Børge’s variant of HATMAN’s Human Solvable 7X Elevens was posted
here.
I only discovered this one while compiling the archive entry for Human Solvable 7X Elevens.
7) Pinata Killer #2 was posted
here.
Puzzle 2 gets a SudokuSolver score of
10.95 and "Uses Extended Trial and Error Steps". JSudoku has a big struggle too.
8) Assassin 119 V3 was posted
here.
I discovered this one, in a hidden part of the Assassin 119 V2 post, when I was making archive copies of original diagrams; sorry it got missed when this Unsolvables post was first created.
Afmob wrote:
By the way, if you got too much time on your hands then try to tackle this one.
The only hint I can give you is that there really is only one solution but no software is going to help you solve it.
SudokuSolver couldn't finish this puzzle.
9) Tarek's Killer Jonin was posted
here.
I am in the long process of developing a killer solver/generator ...
In my experiments, the following puzzle came out. It is very difficult as confirmed by some independent computer solvers. I thought that it would be a good idea to post here for future reference. No need to distract youselves from the good solvable puzzles posted here which would be my ultimate generating target
There are also three variants with similar levels of difficulty.
Afmob wrote:
Just as a word of warning: All these Killers have a SudokuSolver score above 4.8 and need T&E!
10) Tarek's Unsolvable Killer Sudoku List was posted
here.
After downloading SS321, I managed to score several of my generated puzzles. I have compiled a Solver-defeating list which can be (for the time being) part of the hardest, impossible or unsolvable sudokus. This list is likely to grow. The 1st 30+ puzzles are either too long to solve or unsolvable by SS321/JSudoku .... So approach with care
I will post them in the 81-chracter format which JSudoku has implented (Thanks again JC). Some of them are paterrn based, so you may find several puzzle with the same cage patterns.
I'm not sure how to compare these puzzles to decide which is hardest. If solving time is a factor then the following should be near the top (For the moment at least)
Unsolvable #41 Thanks
ixsetf!