Cenoman wrote:Sorry, I find the Senior Exocet presentations a bit terse.
Me too, at first. I think it works, though, if the same extensions are allowed as for JExocets.
The puzzle above doesn't meet these pattern criteria (cover houses > 2 ) So, a pattern extension is used here.
Indeed. It's the same as with the previous "Jellyfish" JExocet, except this one is also a "Mutant" (one S-line is a row). I think such extensions are logical generalizations and should be allowed, though they should probably be mentioned somehow, just like exotic fishes are known by their special prefixes.
Technically the normal fish terms could be used with some of these extended [J/S]Exocets, and I would actually like it. Why invent new terms when we already have descriptive ones that many would understand? Personally I also think the [J/S]Exocet notation should be upgraded to include the fish part (with fish notation), especially in more complicated cases like this. Obviously listing just the base and target cells doesn't fully describe the pattern (not even close). Besides, both are ambiguous terms to begin with, but that's a different problem.
The following statements given by ysfwsf's solver are enigmatic to me...
yzfwsf wrote:1r8 locked in target Cell;
Does the first one mean: "digit one has only one possible target cell => +1r9c56 & -1r89c9 => row 8 void of 1" i.e. the exocet property is demonstrated for digit 1 (no need to count the cover houses) ?
The way I see it is that the SExocet uses R8 as a horizontal S-line (truth) for all digits 1267. Unlike the other digits, 1 doesn't need any other S-lines because all of its candidates in row 8 are restricted to box 8 and the target cell r8c9. Thus, 1R8 is covered by 1b8 and 8n9 directly, i.e. 1r9c56 -> -1r8c46 -> 1r8c9. Obviously it means that there can be no other target for digit 1, because it would be forced into one directly if it were a true base candidate. That is actually true for 6 and 7 also, but not as directly.
yzfwsf wrote:"S" Cells Need Include:2r8,6r8,7r8,
For the second sentence, which cells in row 8 are considered "S" Cells ?
The whole S-cell concept is so confusing in complex cases that I wouldn't know. For digit 1 there are only three candidates in row 8 -- two in box 8 (same as the base cells) and one in r8c9 (target). Conventionally none of them would be S-cells, I think, but this is not a conventional situation so I don't know. For the other digits 267 I guess at least r8c23 are S-cells (of 267R8) covered by 267b7.
The cause of confusion is that traditionally only the cells outside of the JExocet band have been considered S-cells. That's a design flaw in the pattern, because it makes certain generalizations difficult to describe accurately (including the basic SExocet). It would be better to consider all candidates in a given S-house as S-candidates that need covering one way or another. Even better if they were called something else, as the "S-cell" is a loaded and non-descriptive term.
In basic JExocets the S-candidates within the JExocet band are implicitly covered by the base box and base line, and the rest explicitly. Thus only the latter are counted as S-cells (the others are "escape cells"). That approach simplifies that particular situation, but it also complicates other situations where that abstraction doesn't work. In fact, it also makes the simple situation harder to understand for a novice.
And eventually, which sectors are cover houses for digits 2, 6, 7 ?
In row 8 or in general? In row 8 they're covered by boxes 7 and 8 (267b78) and the target cell r8c9 (8n9). In general also by rows 2r2, 67r4, 267r59 and the other target 1n1.
I expect for such hardest puzzles a bit more explanations from experts.
Problem is, these kinds of situations are hard to explain with pure [J/S]Exocet terms. Terms like "S-cells" and "cross-lines" etc. are just confusing in more complicated scenarios, as far as I'm concerned. It's much easier to just view them as normal base\cover problems with the corresponding terms that are non-ambiguous. Here's how I see the truths and links in this case (correctness not guaranteed, as I don't have XSudo):
Exocet (type: Senior, size: Jellyfish, shape: Mutant) :
15x18 {1267R8 267C149 9N56 \ 2r2 67r4 267r59 267b7 1267b8 1n1 8n9} => -35 r1c1 (=> +2 r1c1, +2 r9c56, -2 r8c9); btte
It's a bit simpler than yzfwsf's 17x24 XSudo diagram that provides many more eliminations. The main difference is that mine doesn't use 1C49 as truths -- only 1R8. The obvious problem with the set logic presentation alone is that it's extremely difficult to decipher what it actually proves. It's also difficult to write a single matrix to demonstrate it.
What it should prove is that 1n1 an 8n9 are local Rank 0 regions, which allows eliminating all non-base digits from them, in this case placing 2 in r1c1. I'm not sure if it proves anything else directly, but basic Exocet logic then means that 2 must be a true base candidate, thus true in r9c56 and false in the other target r8c9, which is enough for the solution.
Added. Here's my attempt to demonstrate it with separate sub-matrices, one for each digit assumed in r9c56.
- Code: Select all
1b8
========================================
1r9c56 | 9N56 *
1r8c9 1r8c46 | 1R8
========================================
1r8c9
- Code: Select all
2r9,b8 2r2 2r5 2b7
========================================
2r9c56 | 9N56 *
2r8c9 2r9c9 2r2c9 | 2C9
2r78c4 2r2c4 2r5c4 | 2C4
2r1c1 2r9c1 2r2c1 2r5c1 2r7c1 | 2C1
2r8c9 2r8c46 2r8c23 | 2R8
========================================
|2r1c1
|2r8c9
- Code: Select all
6r9,b8 6r4 6r5 6b7
========================================
6r9c56 | 9N56 *
6r8c9 6r9c9 6r4c9 | 6C9
6r78c4 6r4c4 6r5c4 | 6C4
6r9c1 6r4c1 6r5c1 6r7c1 | 6C1
6r8c9 6r8c46 6r8c2 | 6R8
========================================
6r8c9
- Code: Select all
7r9,b8 7r4 7r5 7b7
========================================
7r9c56 | 9N56 *
7r8c9 7r9c9 7r4c9 | 7C9
7r78c4 7r4c4 7r5c4 | 7C4
7r9c1 7r4c1 7r5c1 7r7c1 | 7C1
7r8c9 7r8c46 7r8c3 | 7R8
========================================
7r8c9
All the other digits except 2 have only one possible target. Since two digits must go to r9c56, one of 167 must go to r8c9 and 2 must go to r1c1.
@mith: it's a good idea to post this kind of puzzles from time to time, provided it is made with some teaching purpose.
I think it's an excellent idea, though a bit late for me. Personally I haven't had any motivation to work on the hardest puzzles on my own, except for the examples in David's compendium and some MSLS/SK-Loop examples. When posted as preselected and public problems like this, it's much more interesting and educational, because you get to actually think about them (instead of knowing in advance what to expect) and can learn from others' solutions.
For a long time now, I haven't bothered to solve any puzzles besides what's posted in the Puzzles here, so this is basically the only way for me to ever gain experience and routine with the exotic techniques. So far I've been satisfied with understanding the theory behind them, but it obviously doesn't fully compensate for the lack of experience.
--
Edit. Removed incorrectly used term "cross-line" (when a cover house was meant).
