The New Sudoku Players' Forum

[Jeff]

Hi Mike

Please add this link to the list under the heading of nice loops.

The notation used in Nice Loops and SINs
http://forum.enjoysudoku.com/viewtopic.php?t=3628

[Mike Barker]

Actually I already have. Its up in the first section which talks about nomenclature.

[Skern]

I am new to sudoku, as in it has been in my local newspaper for about a month so I tried it. I enjoyed it, so I bought a $20, handheld game with 24 levels. I've tried all of the levels and though some take me up to 25 mins. I wonder Why anyone would need those techniques that you all have posted. If you could come up with an example of why I would appriciate it. Thanks.

[Mike Barker]

First welcome to Sudoku. I know I've found it lots of fun and was surpised when there were puzzles that could not be solved with basic methods. I started to keep the collection list as I discovered more approaches which others had developed and am gradually solving more puzzles. Non-advanced techniques (up through nice loops) appear to solve all but about 5 randomly generated puzzles in a 1000. Even with 24 levels there is no guarantee that your handheld contains the most challenging puzzles. Many collections limit difficulty to locked sets, simple fishy cycles and simple almost locked sets. More challenging puzzles are believed to require the more sophisticated techniques described by contributors to this site (or a great memory and the ability to do trial and error or other approach in your head). Try the puzzle that Carcul has solved here. If you can solve it as easily as you say, I'd love to learn of your approach. For more moderate to challenging puzzles try the Top1465. Also visit the sites in the "collection" they have some diabolical puzzles!

[jeffloeb]

shareware program Sudokoach

from KMR Consulting (http://www.kmrconsulting.com)

Sudokoach generates the candidates so that you can concentrate on solving the puzzle

hints - small, medium, and big

[jeanjaz]

I used to visit this website before it changed and downloaded the Sudoku program. I LOVED IT. I still play it all the time. I decided I wanted to learn some more advanced techniques and came back to check out the forum and after wandering around for awhile, found this topic with its list of links to tutorial sites. I immediately took off on a link and spent any spare time in the next couple of weeks working my way thru a tutorial. THEN I wanted to find the list again and I have spent the last HOUR trying to find it. As you can see, Here I Am!

I am SO glad you made this list. Anyone can see it was pretty labor intensive and I appreciate that more than I can express.

Is it possible to make it easier to find or publish it separate somewhere else on the site?

I participate in a Sudoku group on Gather.com. We are all addicted, but I don't notice that anyone is a super-duper solver like I see here. At Gather.com, we write "articles" to share with the rest of the group. The website makes it possible to comment and rate the articles. It is possible for people outside of our group to find our articles if they search.

Based on this, I would like to ask 2 things. (1) Is it okay if I paraphrase some of the things here and write it up in an article to share with my group on Gather.com? (2) Would anyone here like to visit our humble sudoku Group and maybe join and contribute?

It seems that the group isn't familiar with the basic vocabulary and the different solving techniques and their names. I have already written some into comments, but I would like to write a few articles on these subjects because learning the terminology really helped me with my personal solving techniques - gave me useful handles.

I would, and have, given credit when I've quoted people around here and even supplied clickable links at times. Even though I would be paraphrasing, I wouldn't want the great people around here to feel I was plagarizing their hard work.

THANKS again for providing a spot for us Sudoku-addicts that don't have anyone close to talk to and no one to help us learn more. (smile)

- Jeanjaz

[Mike Barker]

Jean, thanks for the encouragement. As far as moving the post that is not up to me, but know that it is a sticky and you can always find it here ( http://forum.enjoysudoku.com/viewtopic.php?p=21804#p21804 ). As far as referencing the ideas you find here, there is no higher praise than someone citing your work. The whole idea is to share ideas, so share away. Also feel free to ask questions. There are many here who love to lend a helping hand.

I stopped by gather.com and it seems like a good group. I use a very different strategy than Robert for solving (I use dots when I find candidates which only occur in two cells in a unit), but consider mine still to be too slow (at least in comparison to other posted solving times I've seen). I'll stop by from time to time to look for other ideas. Thanks for sharing the site.

[sirdave]

Mike- would appreciate it if you could add the following to your list under Almost Locked Sets:

http://www.sudoku.org.uk/discus/messages/29/4448.html?1185670798

Thanks

[denis_berthier]

Mike
You did a very good and useful job in gathering all these links.

I have nevertheless a few remarks about your classification:
- xyt, xyzt, nrct and nrczt chains can hardly be seen as Nice Loops, even if downgraded with "multiple inference"; they don't satisfy the major property of Nice Loops: reversibility; the best way to see them all is, as I've always done, as various extensions of xy-chains;

- putting GEM (a tagging algorithm) in the same category as xyt, xyzt, nrct and nrczt chains (pattern based rules) is very misleading; I think it would better fit with the colouring algorithms.

[ronk]

- xyt, xyzt, nrct and nrczt chains can hardly be seen as Nice Loops, even if downgraded with "multiple inference";

Carcul "downgraded" NLs with "multiple inference" a long time ago.

they don't satisfy the major property of Nice Loops: reversibility

I'd rather we leave the "reversibility" debate on the Eureka! forum.

[denis_berthier]

- xyt, xyzt, nrct and nrczt chains can hardly be seen as Nice Loops, even if downgraded with "multiple inference";

Carcul "downgraded" NLs with "multiple inference" a long time ago.

"multiple inference" is often used for such a downgrading purpose. I think it should be avoided (as "bifurcation" and all other undefined terms …)
Anyway, it corresponds to a confusion between the way a chain rule is proven and the way a chain can be found.

they don't satisfy the major property of Nice Loops: reversibility

I'd rather we leave the "reversibility" debate on the Eureka! forum.

I meant reversibility in the precise, factual sense I defined in the "concept of a resolution rule" thread, together with a lot of other criteria for chains.
What happens on Eureka should not prevent rational people from debating of real questions.

[ronk]

"multiple inference" is often used for such a downgrading purpose. I think it should be avoided (as "bifurcation" and all other undefined terms …)

Multiple Inference is defined in Jeff's Forcing chains: Terminology and Definition. Even if it weren't, frequent usage by many on this forum (and others) makes the meaning quite clear.

Anyway, it corresponds to a confusion between the way a chain rule is proven and the way a chain can be found.

Exactly, but you delude yourself by thinking that manual solvers can find nrc(z)(t) chains without following a stream of inferences. By following that stream of inferences, solvers are effectively "proving the rule" each and every time.

[denis_berthier]

you delude yourself by thinking that manual solvers can find nrc(z)(t) chains without following a stream of inferences. By following that stream of inferences, solvers are effectively "proving the rule" each and every time.

Manual solvers who follow my approach do not have to "follow a stream of inferences" but to find precisely defined patterns. This is exactly the same thing as finding a Swordfish: if you know the pattern and what eliminations it allows, you don't have to re-prove their validity. Of course, if they like, they CAN re-prove the rule every time they use it, which is always a very simple matter, but they don't HAVE TO.
The main problem with the current vision of chains as chains of inferences is that it discredits them as patterns and it leads to classifying them according to the way the associated chain rule is proven (whence terms such as "multiple inference") instead of relying on factual criteria based on the pattern itself.

[wintder]

I was looking for "hidden unique rectangle" and couldn't find it listed here.

http://www.scanraid.com/Hidden_Unique_Rectangles

This link shows the technique, is it here under a different descriptive name?

Thanks.

[Mike Barker]

I haven't heard the term "Hidden URs" much. What we've done here is to develop an entire taxonomy of different types of URs using strong links, bivalues, and ALS with UR's containing 1 to 4 non-bivalued cells. They are described as Strong Link and Nice Loop Types and ALS Types. To keep track of things each has a designation, but the names are generally not used. The specific one described in your link is a UR+3C/2SL (a UR with 3 non-bivalued cells and 2 equal strong links attached to a common cell) which is one of the most common. To allow traceability I'll add "hidden" to the colletion.