The New Sudoku Players' Forum

[Lengen0538]

I'm new to sudoku and until yesterday I was under the impression that Nishio techniques in sudoku are just picking a candidate at random, then trying to solve the puzzle. Then yesterday I saw at sudokuwiki.org that Nishio is actually a chaining technique and realized it similarity to weak linked discontinuous NICE loops. In my opinion, finding NICE loops and all other similar alternating inference chains (x cycles, xy chains, forcing chains etc.) with exception to the Nishio chain are valid sudoku solving techniques because they use logic to find , and that these techniques do not assume a candidate to be true.

My question is this. In your opinion, do you consider Alternating Inference Techniques to be valid sudoku solving techniques? If not, what techniques do you consider valid, considering that many of the other patterned based techniques are chains themselves?

[JasonLion]

Yes, I consider AIC to be a valid (though advanced) technique. Nishio is invalid because it is based on guessing and thus is equivalent to brute force, rather than proceeding logically through proofs. It has nothing to do with chains, though it is possible to look at it as a chaining technique, that isn't where the problem lies.

[David P Bird]

Welcome Lengen0538

To elaborate a bit on Jason's response.

Nishio techniques use forcing chains to check alternative cells that may hold a digit assuming each one to be true in turn looking for conflicts so that one of them can be eliminated.
Nice Loops and Alternating Inference Chains are equivalent techniques but use different terms and notation methods. They use bidirectional chains starting from an observation that one of two conditions must be true looking for candidates that can be eliminated because they can't be true either case.

This post <An AIC Primer >, contains a brief description of the differences between forcing and bidirectional chains.

DPB
.

[champagne]

Yes, I consider AIC to be a valid (though advanced) technique. Nishio is invalid because it is based on guessing and thus is equivalent to brute force, rather than proceeding logically through proofs. It has nothing to do with chains, though it is possible to look at it as a chaining technique, that isn't where the problem lies.

Hi JasonLion,

Nothing against your wording if the technique is applied with "no limit"

However, in Sudoku Explainer, the so-called "Nishio" starts on a bi-value.
The expansion process used in Sudoku Explainer is very similar to complex chains (AIC including objects as ALS, locked sets ...)

Rejecting "SE Nishio" would reject as well all ratings qualified as "SE dynamic" and surely higher levels (SE nested chains)

[SpAce]

I don't know if there's a standard definition of Nishio (seems to be a common complication with sudoku terms) so it's kind of hard to make strong judgments one way or the other. I also think it's a matter of perspective what is considered brute force or a "valid" sudoku technique. The only thing that is surely invalid is something that uses invalid logic.

I don't personally think the SudokuWiki definitions of Nishio and other forcing chains are necessarily brute force much more than AICs, Nice Loops or coloring techniques. In all of those you're using chains or nets to look for contradictions or verities, and they're all logically sound (and can often -- but not always -- be converted into each other). Nishios and forcing chains/nets are just more complicated, and I'd only use them if AICs didn't work.

Building any chain (including AICs) from scratch is always a somewhat trial-and-error approach, because you can't know beforehand if and what kind of results it will produce (unless you're a genius whose brain just "sees" it -- but then you don't need the chain anyway except to explain your logic to dumber people). A non-genius needs to make educated guesses to try the most potential chaining possibilities first (the number of available strong links is a good indicator). A less guess-based way is to use something like GEM to reveal contradictions and then figure out the chains to prove them, but even then you need to make an educated guess for the seed point to get the best results. Experience helps to make better guesses -- but they're still guesses.

A pure brute force approach is when you're randomly picking candidates and seeing what happens if you turn them on, and whether Nishio is in that category depends more on how you use it and not so much on its intrinsic properties. I think almost any technique can be used in a sort of brute force way, if you're just randomly scanning for patterns without any strategy. Nothing is black and white.

[coloin]

Well .... i agree with SpAce - and Champagne
There were many vicious arguments on the old "Eureka" site !

However a guess is a logical technique - its just not immediately obvious if its a right guess or a wrong guess - and if Sudoku Explainer uses it as a technique well - who are we to argue.
The first world championship was won on a guess in the final puzzle - mush to the disgust of some American
A friend of mine on here [stormCkr] was chucked off some competition sites by solving too quickly [I think by using double guess techniques]

I would say uniqueness methods are not logical - you can use the method on a multisolution puzzle - and get a not completely correct answer ..... so that makes it ugly in my book

C

[StrmCkr]

Nisho is a blind guess with an implication subnet, most of the time you cannot prove the chain implies the starting cell on Nisho, it is however also capable of identifying real bi directional chains such as discontinues nice loops as you can back and forwards to the starting cell from the end points.

Logical constructs use a node based construct to prove forwards and reverse the construct to an identical outcome. Without actually trying the candidate in said square.

yes it's used by se to rate. Depth of chains. Moreover that time frames solving technique list was limited and Se badly needs updating to include today's technique sets. Like these
AIC, loops, als, ahs, muti fish, nxn fish, excots. DDs, adds

[StrmCkr]

The first world championship was won on a guess in the final puzzle - mush to the disgust of some American
A friend of mine on here [stormCkr] was chucked off some competition sites by solving too quickly [I think by using double guess techniques]

If I recall correctly the first world's puzzle at the finals are mutisolutions, as no one verified them befor hand.
I was kicked from mutiple sites, as I would solve large sections of their puzzles by full sectors instead of using a singles chain and following it around like everyone else. As their anti cheat methods flagged everything except the singles chain path
Most competition puzzles are nothing but singles naked hidden and the occasional xy wing.

I explained a few of them how I did it quickly ND even strongly recommended an update to their anticheat systems, but still was banned. No skin off my back so to speak.

But that was befor a car accident and head injury from it.

I did however beat the world's fastest mutiple times in practice on many different sites, which didn't aid my case for not being banned.

[SpAce]

I was kicked from mutiple sites, as I would solve large sections of their puzzles by full sectors instead of using a singles chain and following it around like everyone else. As their anti cheat methods flagged everything except the singles chain path

So how did you do it? Sounds to me like you're one of those geniuses I mentioned who just see solutions! (I'm definitely not, so I'm intrigued.)

I did however beat the world's fastest mutiple times in practice on many different sites, which didn't aid my case for not being banned.

Nice!! Except for the banning part which sucks. Being too good for your own good -- what can be more unfair?

[StrmCkr]

- what can be more unfair
Head injury.. Some memory loss, nerve damage to hands shoulders neck. Reduced ability for in head Mathmatic.calculations I've had to relearn alot of sudoku stuff.
. All from some a hole that fell asleep.at the wheel and rear ended my car at a stop light.

[SpAce]

- what can be more unfair
Head injury.. Some memory loss, nerve damage to hands shoulders neck. Reduced ability for in head Mathmatic.calculations I've had to relearn alot of sudoku stuff.
. All from some a hole that fell asleep.at the wheel and rear ended my car at a stop light.

That really sucks. I'm really sorry to hear that happened to you. Some traffic accidents one could avoid but not those. Yep, it's much more unfair.

[coloin]

Yes sorry to hear Chris.

motris/ Thomas Snyder here
He lead all the way through the competition but lost on the last puzzle - he was so miffed he won it a year later !

I recon you could have done it !

[eleven]

Nishio is invalid because it is based on guessing and thus is equivalent to brute force, rather than proceeding logically through proofs.

I don't think so. Guessing for me is to to use a number or elimination, which is not proven to be correct. Is that your definition of Nishio ???
At least it is not Sudoku Explainer's definition. There Nishio is a single digit elimination, which always can be proven.

[JasonLion]

There are several definitions of Nisho, I don't know which one SE is using. I am assuming the Sudopedia definition, which explicitly includes guessing.

[eleven]

You are mixing trial&error and guessing, see http://sudopedia.enjoysudoku.com/Trial.html. Like bifurcation AIC's (which cover bifurcation) are a trial&error method too.