The New Sudoku Players' Forum

[Luke]

When folks ask for help with really tough puzzles in this forum, the replies fairly often point out Kraken Fish eliminations. What's the program folks are using that include this technique? None of mine do, alas.

Does anyone know of a Kraken Fish primer? I've done my woodshedding on the topic, and most web sites (like Sodopedia) repeatedly point to the same diagram that appeared in this forum a coupla yrs ago by Mike Barker. (I think Mike should get royalties for the number of web sites that have borrowed this information!) This thread and most others in SPF quickly advance way beyond the base concepts, leaving the uninitiated in the dust. The web sites that use the diagram just leave it at that, with not so much as one actual example of the technique in action.

If anyone has the time to do a Kraken primer they might end up more celebrated than Mike Barker .

[Myth Jellies]

Try this to start

[daj95376]

daj95376's Kraken Fish Primer:

Perform all lesser elimination techniques prior to Fish().

Call Fish() and perform traditional unfinned/finned/Sashimi/Franken/mutant fish eliminations.

Call Fish() and note when a single base candidate cell within a pattern falls outside of what would otherwise be a proper fish pattern.

Clone the grid and set the cell to the candidate value.

Call Solver() to perform whichever techniques you desire on the cloned grid.
(Note: error checking of the cloned grid will be necessary as eliminations progress.)

Compare the cloned grid eliminations to the eliminations that would have been performed by the original proper fish pattern.

Any common eliminations are from a Kraken Fish.

Note: This approach can be extended to include the case where more than a single base candidate cell falls outside of what would otherwise be a proper fish pattern.

Now, before you try this approach, wait for ronk and others to tell us why I'm all wrong.

[ronk]

before you try this approach, wait for ronk and others to tell us why I'm all wrong.

For me, non-kraken fish are comprised solely of single-digit constraint sets with possible local and/or remote fins. Kraken fish add chain(s) using other digit(s). I've never really graduated to kraken fish.

[tarek]

Call Fish() and note when a single base candidate cell within a pattern falls outside of what would otherwise be a proper fish pattern.

Is this easier said than done ???!!!

tarek

[daj95376]

Call Fish() and note when a single base candidate cell within a pattern falls outside of what would otherwise be a proper fish pattern.

Is this easier said than done ???!!!

It depends upon how you implement Fish(). Besides (old) dedicated routines to find unfinned/finned/Sashimi fish, I have a generic fish finder (GFF) routine that uses a brute-force approach to find Franken/mutant fish. Inside GFF, fin cells drop out as a bitmap from testing a particular constraint set. I reject a constraint set if one or more fin cells don't see any cells in the elimination set. Although I haven't done so, it's easy enough to have it save off the constraint set and a bitmap of these errant cells (to test for Kraken status).

Like ronk, I tend to bypass Kraken Fish -- unless I run across a single-digit variant while manually examining a specific grid. I'm always surprised at how this variant will jump out at me.

[Mike Barker]

Thanks for the kind comments. I think if you look at the Kraken Fish Guide you'll find that Jeff, Carcul, Anne Morelot, and Myth deserve most of the credit.

[Bud]

Does anyone know of a Kraken Fish primer?

The Kraken X-wing Pattern occurs rarely in puzzles. One form of a new more general technique called the Extended 2-String Kite closely resembles the Kraken X-wing. The pattern for this has one fewer candidate cells but it gives the same cell eliminations. This pattern occurs more frequently.

The technique is described in the 2-string kite page of Sudopedia and an example is shown on the Kraken page.