The New Sudoku Players' Forum

[DonM]

However, as a manual solving method, a forcing chain is quite different from an AIC.

Excuse me, but this is just rubbish. Logically they are exactly the same (when you restrict forcing chains to 2 alternative starting points), as well as for a manual solver. What is the difference, when you say "one of x and y must be true, if x then .. and if y then ..." or (x=y) with an AIC left and right?

When it comes to more difficult puzzles with few, if any, bivalue cells, the weakness in the forcing chain method for manual solvers becomes more apparent

Do you really know, what a forcing chain is ?

SINs were interesting from a theoretical point-of-view, but in practice the concept, not the least of which when it came to notation, was unwieldy for manual solvers and was never picked up as a practical manual method

If you ever have tried to solve hard puzzles manually, you will know that when trying to expand one direction of an AIC, it will happen, that this number runs into a contradiction. Embarrassing ?

Gee Eleven, when you talk like that I get all sorts of warm fuzzies.

That's going to constitute my only response because I've come to learn that you're a sniper. You fire these shots then disappear or come up with an obscure puzzle challenge that you already know the answer to, the reason for which is obvious, though irrelevent to the original subject.

[ronk]

As a generic term, discontinuous AICs are forcing chains and ...

What's a discontinuous AIC?

[DonM]

As a generic term, discontinuous AICs are forcing chains and ...

What's a discontinuous AIC?

The opposite of a continuous AIC. Or select from many choices from a Google search- it's a pretty common term. Better yet, ask Myth Jellies (good reading material for Eleven in particular):

http://www.dailysudoku.com/sudoku/forums/viewtopic.php?t=3816

Note Myth's last post: 'there is always a shorter and usually simpler discontinuous AIC that eliminates the intersection...'. I have a feeling you'll be able to guess what he's referring to.

[eleven]

You fire these shots then disappear or come up with an obscure puzzle challenge that you already know the answer to, the reason for which is obvious, though irrelevent to the original subject.

What are you talking about ?

You know that i appreciate MJ very much as one of the most innovate sudoku solvers. The only weakness i can see with him is, that for whatever reason he was ideologically blinded on one eye in thinking that AIC's are somewhat better than forcing chains, while his other eye could see:

The reality is you may have found it like I do, and made no such assumption

Unfortunately you are boosting this weakness into the near of fundamentalism.

[DonM]

You know that i appreciate MJ very much as one of the most innovate sudoku solvers. The only weakness i can see with him is, that for whatever reason he was ideologically blinded on one eye in thinking that AIC's are somewhat better than forcing chains, while his other eye could see:

The reality is you may have found it like I do, and made no such assumption

Hmm, seems you also don't like AIC notation that Myth was highly supportive of.

Unfortunately you are boosting this weakness into the near of fundamentalism.

Myth looked on AICs as emanating from patterns and from everything both he and Steve K have said, they saw patterns from which they constructed AICs. As a pattern-solver I see it the same way though I don't put myself in their category. And I find the process quite different than randomly plugging values into a cell and seeing what affect it has on a digit in another cell.

You apparently disagree with this concept and in fact, your form of describing your solutions (without notation) reminds me of the way I described my forcing chains before I knew anything about nice loops or AICs.

[eleven]

Hmm, seems you also don't like AIC notation that Myth was highly supportive of.

I neither dislike AIC's nor nice loops (apart from some by Carcul, where he intentionally forgot to mention a UR he used). I only had some trouble with them, because i had to learn to read them without seeing any necessity to invent special notations.
But i dislike a claim, that using this or that notation would lead to better solutions or solving strategies.

A guitar song can be notated with notes or tabs. But most of them were found by playing without having notes or tabs in mind.
People who are used to notes, will prefer them, that's all (and i don't mind both, because i don't play guitar).

And I find the process quite different than randomly plugging values into a cell and seeing what affect it has on a digit in another cell.

Apart from patterns (we all like to use), which are included in a solution, the notation does not say so much about, how the solution was found.
Not only, that i can use the same method (e.g. spotting an almost pattern and working on it) as an AIC player, but formulate it with implication chains. Also the other way round someone can use a monkey's (or program) method to find an elimination and afterwards formulate it as an AIC.

PS: My favourite player here is Marty, who does it on paper and knows, how hard basics puzzles can be also for an excellent solver, as well as that often a net is easier to spot than a nice AIC. And i love his heroic attempt to learn to write AIC's.

[aran]

Myth looked on AICs as emanating from patterns and from everything both he and Steve K have said, they saw patterns from which they constructed AICs. As a pattern-solver I see it the same way though I don't put myself in their category. And I find the process quite different than randomly plugging values into a cell and seeing what affect it has on a digit in another cell.

Don
To me a pattern should have some visible quality eg XY-wing or visible logic eg dual-linked ALS.
Or put another way, the pattern must be "seen" in advance of the AIC, for any resulting AIC to qualify as emanating from the pattern.
If the pattern is "seen" by first performing mentally the steps of the AIC, then I would argue that the AIC came first, and that there is no pattern.
I'm interested to know what types of pattern you have in mind which precede their AIC translation.

A second issue is this, and it was raised long ago by Steve K :
amongst the most obvious patterns are fish.
Yet a fish expressed as an AIC produces - as one rises up the fish ladder - an increasingly complex and most unattractive supernet.
Equally any attempt to AIC-ify a dual-linked ALS structure would most often result in spaghetti.
Allan Barker's remarkable contribution notably demonstrating the simplicity of fish (base sets, cover sets, rank, with fish, and dl-als being rank 0 structures) showed that there was another fundamental concept out there.

PS that was rather harsh on eleven a few posts back...I think he was just defending a point of view.

[David P Bird]

It strikes me that what's behind this debate on forcing chains is whether or not start points are selected intelligently or not.

If two forcing chains are started from strongly linked candidates then effectively an AIC approach is being used with the chains being linked together. On the other hand if forcing chains are started from nodes that have no strong links a deduction will only be possible if a contradiction is found between different chain legs. I'd like to be convinced there were good reasons for choosing such start points otherwise I'd be tempted to believe that a blind "suck it and see approach" eventually came up with a hit.

There are terminology issues here too, for example, people describing contradictions arising from more than two legs as "forcing chain" rather than "forcing net" deductions.

Returning to the original debate with Myth Jellies, I made the point that Sudokus are supposed to be solved intelligently. Consequently players shouldn't blindly apply AIC linking rules in a parrot fashion, but should consider what they hope to achieve. I therefore don't rubbish those that think in terms of truths and falsehoods as it's a good way to appreciate the possibilities.

Finally, it's a pity that some very intelligent people here still choose to notate their deductions as implication streams. This makes it difficult for the less experienced players to follow their logic. If they're dyslexic that's understandable, but otherwise it shows that after all the hours they've spent studying the logic, they aren't prepared to spend a little time mastering something quite basic, which just seems arrogant.

[daj95376]

It appears that there are two different definitions for forcing chain in use. I've been using the definition presented by Jeff and supported by SadMan, HoDoKu, and some web sites. David P Bird is using the definition in Sudopedia and supported by some other web sites.

Because of the contrasting definitions, we either need a consensus on which definition to use, or we need to stop criticizing each other on the definition someone else chooses to use.

[David P Bird]

The suggestion that we should stop criticising each others choice of terms reminds me of two old ladies playing bridge in a tournament.

When one of then bid 1 No Trump one of their opponents asked her partner 'What is the range of the high card points should you have to make that bid? "12 to 14" was the reply. With that the opponents contested the bidding and were soundly trounced when the hand was played. In the post mortem one of them turned again to the lady and accused her of misinforming them as her partner turned out to have a hand worth 16 points. "No, I told you what points I would bid one no trump with, but my partner likes to make it stronger" she responded.

Jeff's definitions aren't carved in stone – read his definition of a strong link and you'll see. Intuitively people picture chains as being un-branched as would result if only two legs were of interest. If more legs than that are involved then some other term should be used IMO.

[DonM]

Myth looked on AICs as emanating from patterns and from everything both he and Steve K have said, they saw patterns from which they constructed AICs. As a pattern-solver I see it the same way though I don't put myself in their category. And I find the process quite different than randomly plugging values into a cell and seeing what affect it has on a digit in another cell.

Don
To me a pattern should have some visible quality eg XY-wing or visible logic eg dual-linked ALS.
Or put another way, the pattern must be "seen" in advance of the AIC, for any resulting AIC to qualify as emanating from the pattern. If the pattern is "seen" by first performing mentally the steps of the AIC, then I would argue that the AIC came first, and that there is no pattern. I'm interested to know what types of pattern you have in mind which precede their AIC translation.

Hi Aran,

I find this at the same time a simple and complex subject. Trying to describe to someone what goes on in the mind when solving any kind of puzzle or puzzle-like game is difficult. I'm a chess player and I think there's a broad similarity in what goes on when one solves a sudoku puzzle or makes moves in a chess game. Interestingly enough, the chess wunderkind kid, by most accounts top player in the world, Magnus Carlsen, doesn't seem to be able to explain why he sees what he sees.

Some of the patterns I see are ones that are like kernels of larger patterns such as you mentioned an xy-wing which with transport can lead to a more extensive chain. Steve K did much the same thing with his y-wing => y-wing style (of course, his y-wing = xy-wing). Then there are larger patterns, often ALS related, that lead to more complex chains.

But still there are parts of the process I can't explain. When I read Steve K's blog over a period of a couple of years I was amazed at the patterns/constructs he came up with. He told me once something to the effect that he could look at a puzzle and just see relationships that would lead to finding these patterns.

All that said, I'm sure that there are times when part of my process is assumptive.

A second issue is this, and it was raised long ago by Steve K amongst the most obvious patterns are fish.
Yet a fish expressed as an AIC produces - as one rises up the fish ladder - an increasingly complex and most unattractive supernet. Equally any attempt to AIC-ify a dual-linked ALS structure would most often result in spaghetti. Allan Barker's remarkable contribution notably demonstrating the simplicity of fish (base sets, cover sets, rank, with fish, and dl-als being rank 0 structures) showed that there was another fundamental concept out there

Perhaps in the same category: Some relatively obvious patterns used in an AIC, such as an ER, are almost as easy to find as things like x-wings, but result in a net when written as an AIC. To me, finding nets such as almost-AICS is a more assumptive process than finding an ER-based net because the latter is a pure visual pattern.

You're an excellent manual solver. How do you describe the process you use?

[eleven]

Above i mentioned the embarrassing situation, that you could run into a contradiction, when following one side of an AIC.
But i forgot to say, that it also could happen, that one side of the AIC solves the puzzle completely (when you hit a backdoor). Don't panic, in this case you can postulate a uniqueness solution (you just have proved, that all other candidates cannot be part of a unique solution).

[aran]

Hi Don
Just continuing on patterns and all this preceded with imo :
Major features of patterns
- visibility
- eliminations obvious (eg fish, dl-als)
- explanation if ever required : by way of simple contradiction (imagine explaining a fish, dl-als or even xy-wing elimination in any other way)
- never usefully expressed as AICs.

An AIC by the very fact that it requires expression as an AIC is not a pattern.

Some short AICs have name tags (skycscraper, m-wings) but names are not enough for patterns.

Incorporating patterns and uniqueness within an AIC is most satisfying (even I might add if it produces nothing) (within a chain I call patterns...patterns...and not almost patterns, since the chain removes the almost element).
AICs with names, but which are not patterns as I describe them, are rarely (I would say) incorporated into AICs precisely because their eliminations are not obvious...because they are not patterns.