The New Sudoku Players' Forum

[HATMAN]

Fairy Samurai

I think you will all find this a bit different - it is for Børge.

It is plain vanilla but with a "slightly" different layout - I've been thinking about a layout like this for over a year.

Nothing beyond pointing pairs and hidden singles.

No single part solves but the solution space is quite restrained, so it all works together.

Improvements on the name welcome.

[HATMAN]

Simon_blow_snow suggested the name "Blade Samaurai" and has posted better pictures and a solution here:

http://www.djape.net/sudoku/forum/viewtopic.php?t=1479

For clarity:
The centre is a plain vanilla sudoku.
The outer Blades have no nonets, horizontals cannot repeat, verticals cannot repeat and the marked diagonals cannot repeat. the other diagonals can (and often do) repeat.

[Pat]

Simon_blow_snow suggested the name "Blade Samaurai"
and has posted better pictures---

but now, djape's forum has vanished

do you perhaps have the better pictures ?

[HATMAN]

Pat

Enxio copied the best one and posted it on the Assassin site:

http://www.rcbroughton.co.uk/sudoku/for ... f=16&t=773

I'll be doing killer version shortly.

By the way I'm asking about the minimum number or cages in a plain killer on that site. Has there been a discussion here that I missed?

Maurice

[enxio27]

For completeness, here is the diagram:

[SudoKai]

blade_candidates.png (107.23 KiB) Viewed 1751 times

[HATMAN]

enxio, SudoKai

Thanks for posting these. You know I've completely forgotten how I created them.

Maurice

[creint]

Is there any solution available? My solver can solve this but it runs into errors on around 35%.

[SudoKai]

Here is the solution:

blade_solution.png (127.13 KiB) Viewed 1680 times

Here is another one to pump through your Solver:

blade_2018-01-21.png (63.3 KiB) Viewed 1680 times

Hidden Text: Show

blade_2018-01-21_solution.png (140.18 KiB) Viewed 1680 times

[creint]

It solved the first one very quickly once i noticed that the purple boxes can contain the same candidate.
The second one it cant make much progress, i doubt if it can be solved with single forcing nets.

Of course i can make trying with logic, but that removes the need to implement any logic.
Its probably too easy to ask which logic can solve this puzzle besides guessing.

[SudoKai]

Maybe it could be the 'Common peer Elimination' solving technique in your solver that needs to be altered a bit. Here are a few examples to look out for.

Basic example of 'Common peer Elimination'

common_peer_elim1.png (61.35 KiB) Viewed 1656 times

Same basic principal as above, but that multiple 'Common peer Eliminations' are found in a single pass.

common_peer_elim2.png (71.81 KiB) Viewed 1656 times

This 'Common peer Elimination' is found on one Sudoku board, but the elimination occurs on another Sudoku board.

common_peer_elim3.png (76.23 KiB) Viewed 1656 times

If it is possible, also check your other basic solving techniques and see if they catch all eliminations based on the last example.

[creint]

Checked those, in the last one it cannot find the 7,8 in the purple box, so its stuck before last elemination example.
Probably because it can't detect fishes, ur and maybe cycles. Now i'm curious what i'm missing.
-UR, but more cells uniqueness is possible too and external exclusions outside ur are known to me but finding every almost truth set is hard.
-Fishes, but a bit harder to build compared to locked sets
I really want to implement BERT but i don't know how to do that.

On your site there was not much information/screenshots to know which features your solver has.

[SudoKai]

These are the Solving Techniques used by SudoKai v4 for now:

Singles
Naked Pairs, Triples, Quads
Hidden Pairs, Triples, Quads
Common Peer Elimination
Locked Candidates

Maybe it could be a Sudoku rule or explanation regarding this variant that is a bit vague. This is what one of the Blades looks like and the purple cells is where these blades overlap.

blade.png (13.85 KiB) Viewed 1636 times

Otherwise if you could post an image of how far your solver gets, maybe I could see what the next elimination should be.

[creint]

I currently have:
Singles
Locked sets of dynamic size: Naked, Hidden
Common Peer Elimination (Locked Candidates is subset)
Single nets without (unit forcing, almost locked sets and ur)
And some more that are not in use

After checking my code it seems that i wanted to do that, after a small change it finds them.
But strangely it still did not solve the sudoku, maybe you can spot the problem.

Blade sudoku before chains (after fix).png (44.67 KiB) Viewed 1626 times

Blade sudoku after chains (after fix).png (46.02 KiB) Viewed 1626 times

[SudoKai]

My solver finds a 'Common peer Elimination' in both cases

common_peer_elim.png (53.92 KiB) Viewed 1622 times