dailysudoku.com

[keith]

This puzzle is quite interesting:

[Asellus]

Warning: I may lose some folks in this message. It is a somewhat more formal way of restating what Keith has said above and deals with some slightly arcane logic. However, it may be helpful for those who are ready to absorb its content.

Whether or not a new "W-Link" term is needed depends upon whose terminology is being used, it seems. While I'm no authority, I have been reading up a bit on implication chain logic. In implication chaining, the term "strong link" is often defined differently from how it is used in common parlance on this Board. This is confusing at first (and certainly confused me), but makes a great deal of sense as one explores the logic further.

Given 2 cells that have a common candidate digit, there are 4 possible cases:

A: F F
B: T F
C: F T
D: T T

"T(rue)" means the puzzle solution has that digit in that cell. "F(alse)" means the puzzle solution has some other digit in that cell.

For the purposes of implication chaining:

A weak link means that at most one of the cells is True. So, Cases A, B and C comprise possible weak links.

A strong link means that at least one of the cells is True. So, Cases B, C and D comprise possible strong links.

Note that this definition of a "strong link" differs from the common usage on this Board. Commonly, we think of a "strong link" as Cases B and C: i.e. an exclusive either/or relationship. [This is probably because we normally limit ourselves to "buddie" cells when discussing links, in which case an implication strong link can only be Cases B and C.]

So, the "W-Link" is not a "strong link" in common parlance on this Board. But, it is a "strong link" in implication chain parlance. [Note that Case D can only occur between "remote" cells. In fact, a W-Wing is the only way of which I am aware to establish such a relationship between two remote cells.]

Finally, standard implication chains require that the links alternate strong and weak. So, the "W-Link" is acceptable in a chain (such as coloring) provided that it occurs at a point where a strong link is necessary (since it is actually an implication "strong link").

[Note: The common parlance either/or "strong link" is comprised of Cases B and C. This makes it a sort of "universal link" for implication chain purposes since it complies with both the "strong" and "weak" definitions and can thus occur anywhere in the chain.]

Using Eureka notation, one can write the <1> elimination from R6C7 as follows:

(1)R6C7-(1=1)R1C74-(1=7)R9C4-(7=7)R3C45-(7=1)R6C5-(1)R6C7; R6C7<>1

The red portion of the chain is the "external link activation of the W-Wing." If the red portion is omitted (and the cell references tidied up!), the chain is still valid since the resulting "1=1" "W-Link" is, in fact, an implication chain strong link (though reference to the W-Wing is necessary), as required by the alternating notation .

In conclusion, I'd say that a "W-Link" works in coloring provided that it occurs at a "strong link" point in the underlying AIC logic. Since basic/simple coloring relies only on strong links (as understood in the "universal" common parlance sense), this shouldn't pose any problems and Keith's approach works fine. However, beware of using a "W-Link" in a "Color Wing" (or "Multi-Coloring") or in some more general AIC unless you pay careful attention to the sequence of alternating links!

[keith]

Asellus,

I have asked this question elsewhere:

Strong Link: One of two cells is A.


Weak link: One of two cells is not A.


OK so far. But, what is this, what I called the W-link? One or both of two cells is A.

The other place where this occurs is the two tips of a skyscraper (turbot fish). (And, I agree, I need to learn and to use, Nice Loop notation.)

The other cases are: Both of two cells are A.

Neither of two cells are A.

I agree with what I think you are saying: You cannot just use a W-link as a strong link in any established logic. But, yes, I really want to go there!
Best wishes,
Keith

[keith]

Asellus,

I have been contemplating your reply.

[Asellus]

Keith,

The idea is actually much simpler than it seems from all the explanation. Ruud's strong link definition does not contradict the "at least one is true" definition. Ruud's definition is: If A is false, B is true AND if B is false, A is true. But, that does not logically preclude both being true.

When we normally consider a link between two cells, we are concerned only with "buddie" cells. In that instance, Case D (both are True) can never occur so the "strong link" becomes the "either/or" type as we usually think of it. But, if we extend the notion to remote cells, things are different.

For implication chaining we have the following requirement:

An "if A is False, B is True" link must be followed by an "If A is True, B is False" link, and so on in strict alternating fashion. The first is "Strong" and the second is "Weak".

The conventional buddie cell "either/or" notion of a strong link can satisfy either type of implication requirement and so can occur at any point in a chain.

The W-Wing cells are remote and so are capable of falling into Case D (both True). This means that they can satisfy the "Strong" implication requirement, but not the "Weak" requirement. Hence the need to be careful of their sequence in a complex chain. (As I mentioned before, in the case of simple or basic coloring, we can ignore this detail because we are otherwise using only the conventional "either/or" buddie cell strong links. However, if you got creative and tried to weave two separate W-Links into a chain, you would have to be careful of their sequence placement. Or, if your chain included a weak link, you would similarly need to attend to the sequence placement.)

This more generalized strong link notion is not something I dreamed up. I have read it elsewhere but didn't save any links to the reference. At the time, it didn't make much sense to me. But, it stuck in the back of my mind. Your W-Link made the light bulb turn on!

While confusion can easily result, there isn't really a terminology conflict. The "either/or" strong link is a special case in which remote cell links are excluded. The "at least one is true" strong link belongs to the more general case that does not have such a restriction.

Or, at least, that's how I see it.

[keith]

Asellus is correct. You cannot blindly use the W-link as a regular strong link.

If you write the W-link as
W=+=W

and a strong link as
W==w

then the coloring chain is something like
... W==w==W=+=W==w==W==w ...
and you can eliminate the candidate from any cell that sees two W's, one from each side of the +.

Or, if you write it as
... W1==w1==W1=+=W2==w2==W2==w2 ...
you can make the elimination from any cell that sees both a W1 and a W2.

w1 and w2 are not useful to make eliminations.

Keith

[Asellus]

[Steve R]

A comment on the terms used may be helpful.

To illustrate them A, B, .... stand for sets of cells (treat them as individual cells if you like) and x, y, … stand for the symbols to be entered in the puzzle.

In terms of chains a strong link, written A =x= B, means x must be an entry in a cell of A or a cell of B. Thus it implies that x is a candidate for both A and B. Also it permits x to be entered in both A and B. This seems to be the meaning that Keith desires.

A weak link, written A -x- B, means that x is a candidate for both A and B and can be entered into at most one of them. It permits neither A nor B to have x as an entry. As an aside the definition is exactly that of “restricted common candidate,” a term introduced by bennys in relation to almost locked sets but applicable more widely.

As “strong” is used in other contexts to mean that A and B are conjugate with respect to x, the term “strongly inferential” is sometimes used for the link described as strong above. In chain terms conjugates A and B with respect to x give rise to two different links: A =x= B and A -x- B. There is no link meaning “x is an entry for precisely one of A and B,” that is for conjugacy, because so far no one has come up with a scheme incorporating it which can transport useful information along the chain.

There is one other useful link (mixed), written A =x- B. The strong and weak links read identically from left to right and right to left. The mixed does not:
- (from left to right) x is a candidate for both A and B and, if x is not an entry for A, it is not an entry for B
- (from right to left) x is a candidate for both A and B and, if x is an entry for B, it is an entry for A.
However, if you think about it, these are two ways of saying the same thing. This type of link is most commonly associated with near fish. Take an x-wing based on rows for example. If B is a cell in one of the transverse columns which has x as a candidate, x may be eliminated from B. Now go one step further and suppose the bottom row has a third cell, A, open to x. There is no x-wing but, if x is excluded from A, the x-wing reappears and the elimination from B is sound. That is, a link A =x- B becomes available for constructing a chain.

I have used nice loop notation. The principles underlying AICs are identical though I think those unfamiliar with AIC notation find it less intuitive.

Steve

[Asellus]

Steve,

Interesting and helpful post.

I have notice the "mixed" sort of link, but never tried to express it explictly or notate it. (I generally don't follow implications to that level of depth... yet.)

"Strongly inferential" does capture the essential nature of the "remote" strong link of this thread.

I would reiterate that the "conjugacy" link (the common sort of either/or "strong" link) is certainly useful in implications, since it can satisfy both strong and weak inferences. It is a sort of inferential "wild card." What doesn't seem to be useful is having a special symbol for it in chain notations. I assume this was your meaning.

[Steve R]

Thank you for your kind remark.

As to the question, it was indeed, though I think it goes a little deeper than that.

Let me bore you with the parable of the intelligent baby – please forgive the oxymoron. Its parents are teaching it simple colouring. These cells are conjugate so, if the first does not contain x, the second must. The second cell is conjugate to a third so, if the second contains x, the third cannot. Thus it continues with +/-, red/green or whatever until a conclusion is reached. Being intelligent, the baby thinks “Aha. My parents are unduly restrictive, like all parents the whole world over. It is not necessary for all these pairs to be conjugate and to be coloured. Every other time they need only be associates and eliminations can still be made.”

The baby has split the conjugacy into its two meanings and proceeds to solve more puzzles than its parents with exactly the same effort.

Steve

[TKiel]

[Asellus]

Tracy,

It seems you understood more than you gave yourself credit for. I have posted a correction on the other thread.

This "strong inferential" link exists on the ends of XY Wings/Chains as well as on W-Wings. The same sort of coloring results. I suppose we could call it an "inverse color wing" or "inverse multi-coloring" since the polarities used for elimination are the opposite of those in the weak link based "color wing"/"multi-coloring".

[TKiel]