Hi Scude,
Scude wrote:I have recently been learning about the Swordfish and just when I thought I had it mastered I find one that doesn't appear to behave the way I expect it to.
Your puzzle is a good one to practice finding Swordfishes because it has three. Just not in 4s. The one you thought was a Swordfish in 4s is actually an Almost Swordfish, because (like Leren already said) it has an extra candidate in column 5 (r3c5) that is not covered by any of the three rows (r479). Such spoiler candidates are called 'fins' in the fish context, or just spoilers in other almost-patterns. Almost Fishes are also called Finned Fishes, though it's a bit ambiguous term.
The general rule with an Almost Fish is that either the fin is true or the fish is true (but never both), and that knowledge can be used in chaining, or for direct eliminations when the fin sees any potential eliminations of the fish (not available in this case). The latter type with direct eliminations would be a regular Finned Fish. The one here is also one type of a Finned Swordfish, though not the regular kind because it has a 'remote fin' (one that doesn't see any potential eliminations directly) and thus would need a chain to prove any eliminations. With a chain attached to its remote fin it would be called a Kraken Swordfish.
The most effective way to use any Almost Fish would be to eliminate the fin(s), if possible, because then the fish would become alive with all of its potential eliminations turning real. Unfortunately, in this case the fin happens to be a true candidate (you can't know that in advance), so it can't be eliminated and consequently the fish will never be alive. Thus, you should probably forget about this Almost Fish, unless you're ready to build chains. Just for fun, here's one such chaining elimination that I spotted easily:
- Code: Select all
.-----------------.-------------------.---------------------.
| 15 8 4 | 7 259 59 | 126 1256 3 |
| 157 3 257 | 8 6 45 | 124 9 245 |
| 9 6 25 | 23 b235#4 1 | 8 7 245 |
:-----------------+-------------------+---------------------:
| 8 a47* 367 | 16 39 2 | 5 a146* 469 |
| 3456 2 356 | 16 7 359 | 13469 8 469 |
| 356 9 1 | 4 358 358 | 236 26 7 |
:-----------------+-------------------+---------------------:
| 3467 a47* 9 | 5 a2348* 3478 | 26-4 a246* 1 |
| 467 5 678 | 29 1 47 | 2469 3 24689 |
| 2 1 38 | 39 ac34* 6 | 7 ad45* d4589 |
'-----------------'-------------------'---------------------'
Kraken Swordfish (4)c258\r479 + remote fin r3c5 -> r9c89 => -4 r7c7
As an AIC:
(4)c258\r479 = r3c5 - r9c5 = (4)r9c89 => -4 r7c7
In other words, either the fish is true (which would eliminate all other 4s in rows 4,7,9 including r7c7) OR the remote fin 4r3c5 is true, which would force 4r9c5 false, and either 4r9c8 or 4r9c9 to be true. Thus, the chain proves that either the fish or one of 4r9c8 or 4r9c9 must be true, and since they all see 4r7c7 it can be eliminated. See (*) for a simpler way to get that elimination.
That probably seems a bit complicated if you have no chaining experience, but I just wanted to show that what you found was not totally useless.
Like I said, there are three normal Swordfishes in this puzzle, so you should probably try finding them instead of scratching your head too much with the above Kraken Fish. If you find them all, the rest is solvable with basics. Below you'll find some hidden hints.
They're in digits 2, 3, 6
Hint 3 (full spoilers): Show (2)r167\c578 => -2 r3c5,r28c7
(3)r349\c345 => -3 r5c4,r67c5
(6)r167\c178 => -6 r58c17,r4c8
PS. A Hodoku tip: If you click the green square (digit filtering mode switch) it will turn red, and the digit filtering becomes easier to see because it highlights cells instead of individual candidates. I normally use that mode, and it would probably help to avoid these kinds of mistakes too. The green mode (that you have on in the image) is more useful if you want to filter multiple digits simultaneously (press and hold CTRL to choose more than one).
--
(*) Added. Any Kraken Fish that only uses the fish digit for its chain(s), like here, can always be turned into a (possibly complex) finned fish by adding more covers and bases for the chained elements. That way we can get rid of the external chain and embed its logic into the fish itself. Here too. In this case that would give us this weird-looking 4x5 obi-fish:
4x5 Mutant Obi-Jellyfish: (4)r9c258\r479c5b9 => -4 r7c7
It works as is, but it can be simplified because some sectors (r9 and c5) exist as both bases and covers so they cancel each other out. Removing them gives us:
2x3 Obi-X-Wing: (4)c28\r47b9 => -4 r7c7
...which is normally expressed as:
- Code: Select all
.----------------.----------------.---------------------.
| 15 8 4 | 7 259 59 | 126 1256 3 |
| 157 3 257 | 8 6 45 | 124 9 245 |
| 9 6 25 | 23 2345 1 | 8 7 245 |
:----------------+----------------+---------------------:
| 8 *47 367 | 16 39 2 | 5 *146 469 |
| 3456 2 356 | 16 7 359 | 13469 8 469 |
| 356 9 1 | 4 358 358 | 236 26 7 |
:----------------+----------------+---------------------:
| 3467 *47 9 | 5 2348 3478 | 26-4 *246 1 |
| 467 5 678 | 29 1 47 | 2469 3 24689 |
| 2 1 38 | 39 34 6 | 7 #45 4589 |
'----------------'----------------'---------------------'
Finned X-Wing: (4)c28\r47 + fin r9c8 => -4 r7c7
Of course that would have been trivial to find directly had I been looking for Finned X-Wings, but I actually found it using the above process because I was focusing on the already found Kraken Swordfish. It demonstrates that fishes can be converted into other fishes using the rules defined
here. That's pretty advanced stuff, of course, and you probably shouldn't worry about it now. It is, however, the key to truly understanding fishes, so it's good to be aware of.
