The New Sudoku Players' Forum

[ArkieTech]

Code: Select all

 *-----------*
 |3..|..9|5..|
 |5.8|.37|..6|
 |..7|..1|...|
 |---+---+---|
 |..1|...|.6.|
 |6.4|.2.|8.3|
 |.3.|...|7..|
 |---+---+---|
 |...|7..|3..|
 |7..|56.|9.1|
 |..9|1..|..5|
 *-----------*


Play/Print this puzzle online

[SpAce]

Code: Select all

.-------------.-----------------.-----------------.
|  3   1    6 | 248  48   9     |  5     24   7   |
|  5   24   8 | 24   3    7     |  1     9    6   |
| b24  9    7 | 6    5    1     | c24    3    8   |
:-------------+-----------------+-----------------:
|  89  5    1 | 3    7    8-4   | c2(4)  6    249 |
|  6   7    4 | 9    2    5     |  8     1    3   |
|  89  3    2 | 48   1    6     |  7     5    49  |
:-------------+-----------------+-----------------:
|  1   6    5 | 7    9   a2(4)8 |  3     248  24  |
|  7   48   3 | 5    6   a2(4)8 |  9     248  1   |
| b24  248  9 | 1   a48   3     |  6     7    5   |
'-------------'-----------------'-----------------'

TowER Crane Transport:

(4): b8p[36=8] - r[9=3]c1 - r[3=4]c7 => -4 r4c6; stte

aka:

Mutant Swordfish (3x4):

(4)c17b8\r349c6 => -4 r4c6; stte

[Ngisa]

Code: Select all

+----------------+--------------------+-------------------+
| 3     1      6 |e248     f48    9   | 5     2-4     7   |
| 5     24     8 |e24       3     7   | 1     9       6   |
| 24    9      7 | 6        5     1   | 24    3       8   |
+----------------+--------------------+-------------------+
| 89    5      1 | 3        7     48  | 24    6       249 |
| 6     7      4 | 9        2     5   | 8     1       3   |
| 89    3      2 |d48       1     6   | 7     5      c49  |
+----------------+--------------------+-------------------+
| 1     6      5 | 7        9     248 | 3    a248    b24  |
| 7     248    3 | 5        6     248 | 9    a248     1   |
| 24    248    9 | 1        48    3   | 6     7       5   |
+----------------+--------------------+-------------------+


(4)r78c8 = r7c9 - r6c9 = r6c4 - r12c4 = r1c5 => - 4r1c8; stte

Clement

[SpAce]

(4)r78c8 = r7c9 - r6c9 = r6c4 - r12c4 = r1c5 => - 4r1c8; stte

aka:

Mutant Swordfish (3x4):

(4)r6b29\r1c489 => -4 r1c8; stte

Btw, for those who don't dig my 3D notation... The main reason why I do is not the shortness (though it's nice too) but the clarity it provides for the reasons of the links. The benefit is obvious when you convert X-Chains into fishes and vice versa (when applicable). If Clement's X-Chain is written in 3D, the fish conversion is trivial because you can read the base sets directly from the terms housing the strong links:

b9p[25=3] - r6c[9=4] - b2p[14=2] <-> r6b29\r1c489

The cover sets are just as obviously the the houses containing the weak links, but they're easier to read from an alternative 3D translation:

r78c8 = r[7-6]c9 = r[6-12]c4 = r1c5 <-> r6b29\r1c489 => -4 r1c8

[Cenoman]

Code: Select all

 +------------------+-------------------+-------------------+
 |  3    1     6    | d48-2  48   9     |  5   a24#   7     |
 |  5    24    8    |  24    3    7     |  1    9     6     |
 |  24   9     7    |  6     5    1     |  24   3     8     |
 +------------------+-------------------+-------------------+
 |  89   5     1    |  3     7   b48#   |  24   6     249   |
 |  6    7     4    |  9     2    5     |  8    1     3     |
 |  89   3     2    | c48    1    6     |  7    5     49    |
 +------------------+-------------------+-------------------+
 |  1    6     5    |  7     9    248*  |  3    248*  24    |
 |  7    48    3    |  5     6    248*  |  9    248*  1     |
 |  24   248   9    |  1     48   3     |  6    7     5     |
 +------------------+-------------------+-------------------+

UR(28)r78c68 using externals
(2)r1c8==(8)r4c6 - r6c4 = (8)r1c4 => -2r1c4; ste

[rjamil]

Code: Select all

3....95..5.8.37..6..7..1.....1....6.6.4.2.8.3.3....7.....7..3..7..56.9.1..91....5
 +------------+-------------------+-----------------+
 | 3   1    6 | 2(4)8  [4]8  9    | (5)   2[4]  7   |
 | 5   24   8 | 2[4]   3     7    | 1     9     6   |
 | 24  9    7 | 6      5     1    | 2[4]  3     8   |
 +------------+-------------------+-----------------+
 | 89  5    1 | (3)    7     [4]8 | 2-4   6     249 |
 | 6   7    4 | 9      2    5     | 8     1     3   |
 | 89  3    2 | [4]8   1    6     | 7     5     49  |
 +------------+-------------------+-----------------+
 | 1   6    5 | 7      9    248   | 3     248   24  |
 | 7   48   3 | 5      6    248   | 9     248   1   |
 | 24  248  9 | 1      48   3     | 6     7     5   |
 +------------+-------------------+-----------------+

Grouped Empty Rectangle: ERI 4 @ b3r1c7 Strong Link between ERI 4 @ b2r1c4 and ERI 4 @ b5r4c4 => -4 @ r4c7; stte

R. Jamil

[SpAce]

Grouped Empty Rectangle: ERI 4 @ b3r1c7 Strong Link between ERI 4 @ b2r1c4 and ERI 4 @ b5r4c4 => -4 @ r4c7; stte

Hi rjamil! That's a nice pattern! I like your grid diagram better than last time, too. However, your terminology is a bit confusing. I would suggest using something other than "Grouped" because it has a well-defined meaning in chaining, and you mean something else here. Also, like last time, I don't understand what you mean by "Strong Link" here. It's another term that has a different definition from what you probably mean here. Other than that, it's nice!

Btw, as a fish it's yet another Mutant Swordfish (Rank 1):

(4)b235\r14c47 => -4 r4c7; stte

...easily derived from the 3D chain:

(4): b3p[7=2] - b2p[12=4] - b5p[7=3] => -4 r4c7; stte

[rjamil]

Hi SpAce,

Hi rjamil! That's a nice pattern! I like your grid diagram better than last time, too.

First of all, many thanks for the appreciation.

However, your terminology is a bit confusing. I would suggest using something other than "Grouped" because it has a well-defined meaning in chaining, and you mean something else here. Also, like last time, I don't understand what you mean by "Strong Link" here. It's another term that has a different definition from what you probably mean here. Other than that, it's nice!

Well, I follow the terminology as what is already discussed and confirmed by senior Sudoku experts.

R. Jamil

[SpAce]

However, your terminology is a bit confusing. I would suggest using something other than "Grouped" because it has a well-defined meaning in chaining, and you mean something else here. Also, like last time, I don't understand what you mean by "Strong Link" here. It's another term that has a different definition from what you probably mean here. Other than that, it's nice!

Well, I follow the terminology as what is already discussed and confirmed by senior Sudoku experts.

Who are the other "senior Sudoku experts" in those discussions besides StrmCkr? What parts in those discussions should confirm that you're using those terms correctly here? During my quick scan I saw only you talking about "Grouped Empty Rectangles" there too, so are you in fact the senior expert who confirms your own terminology? I also didn't see anything related to your usage of the term "Strong Link" here.

In normal usage Empty Rectangles are grouped by default, and the minimal (non-grouped) ER is the exception, so in my experience no one but you talks about Grouped Empty Rectangles. In your pattern you have thus one "normal" ER (b2) and two minimal ERs (b3, b5). Yet, even if you really wanted to call the default case "grouped", which is technically correct but redundant, it doesn't explain this:

Grouped Empty Rectangle: ERI 4 @ b3r1c7 Strong Link between ERI 4 @ b2r1c4 and ERI 4 @ b5r4c4 => -4 @ r4c7; stte

To me it seems that you call the whole pattern "Grouped Empty Rectangle", which I interpreted as meaning that it's a combination of three ERs. Only one of them is actually "grouped" but it's not the first one you list after the keyword. Still very confusing.

Also, last time I asked you what you meant by "Strong Link" but you didn't explain it -- you just removed the keyword. Now you're using it again and claiming that it's correct, but still you haven't answered my original question: what exactly do you think is strongly linked here? It seems to me that you're talking about a strong link between two or three ERIs, which is incorrect. The ERIs aren't strongly linked to each other. They do have internal strong links, though, but that's the nature of ERIs and thus not worth mentioning. So, it's very confusing also.

Even if you're using those terms correctly, which isn't yet established, there's really no need to mention either "grouped" or "strong link" here. An ER has always a box-based strong link which causes it to work in the first place. Every ER has that, so mentioning it is redundant. Typically that strong link is grouped as well, so it's the default case -- again no need to mention it specifically. Adding those terms is both unnecessary and confusing, even if you're actually using them correctly. I'm still not quite sure you are, though, because you haven't really explained at all what you mean by them. Your links caused just more confusion, actually, because they don't seem to explain your usage of the terms at all.

[rjamil]

Hi SpAce,

Who are the other "senior Sudoku experts" in those discussions besides StrmCkr?

I assumed StrmCkr and three others confirmations as well (assuming no news is good news from others).

What parts in those discussions should confirm that you're using those terms correctly here?

Look at the last post of the same thread.

Also, last time I asked you what you meant by "Strong Link" but you didn't explain it -- you just removed the keyword. Now you're using it again and claiming that it's correct, but still you haven't answered my original question: what exactly do you think is strongly linked here?

What I interpreted, may be wrong, but is that, one ERI in b3 is strongly linked with two ERIs in b2 and b5.

However, I am not in a position to challenge anything and follows what is simple and easy to understand (to me!).

R. Jamil

[SpAce]

Who are the other "senior Sudoku experts" in those discussions besides StrmCkr?

I assumed StrmCkr and three others confirmations as well (assuming no news is good news from others).

Well, I only saw you two discussing, so I have no idea who those three others might be. That's largely irrelevant, though.

What parts in those discussions should confirm that you're using those terms correctly here?

Looked at the last post of same thread.

Thanks for the clarification; I'd missed that part. I must assume that StrmCkr's confirmation meant that your diagrams were correct (which they seem to be). I wouldn't presume anything about his opinion about the names you had chosen for those patterns. The first two (Empty Rectangle and Dual Empty Rectangle) are standard and match the given patterns, but the other three names are probably your own inventions, aren't they?

I don't like any of them because you're using terms that have different meanings in other contexts (or worst of all, even in the same context -- like "Dual"). Your use of "Dual", "Linked" and "Grouped" are all very confusing, as far as I'm concerned. Don't think of those names as standard just because no one specifically told you before that they're not (no news is not always good news -- it might just mean that no one cared enough to comment). I'm telling you now that they conflict with established meanings and/or don't describe the patterns well (e.g. what you call "Linked" I see as "Looped").

Also, last time I asked you what you meant by "Strong Link" but you didn't explain it -- you just removed the keyword. Now you're using it again and claiming that it's correct, but still you haven't answered my original question: what exactly do you think is strongly linked here?

What I interpreted, may be wrong, but is that, one ERI in b2 is strongly linked with two ERIs in b3 and b5.

That's what I thought you meant, and it's wrong. When multiple ERs are chained the ERIs aren't strongly but weakly linked between each other. The strong links are internal to the ERs, and the weak links between them make the chain propagate. Remember that a chain must have alternating strong and weak links. Since the ERs have internal strong links they must use external weak links.

[rjamil]

Hi SpAce,

I must assume that StrmCkr's confirmation meant that your diagrams were correct (which they seem to be). I wouldn't presume anything about his opinion about the names you had chosen for those patterns. The first two (Empty Rectangle and Dual Empty Rectangle) are standard and match the given patterns, but the other three names are probably your own inventions, aren't they?

Just searched "Grouped Empty Rectangle" and found this post dated Wed Mar 08, 2006 4:45 am by Mike Barker using in point 42.

As far as "Dual Linked Empty Rectangle" wordings is concern, found this post dated Dec 07, 2016 6:15 am by StrmCkr.

Hope it is clear now that the above terms are not invented by myself and/or overlooked by StrmCkr.

That's what I thought you meant, and it's wrong. When multiple ERs are chained the ERIs aren't strongly but weakly linked between each other. The strong links are internal to the ERs, and the weak links between them make the chain propagate. Remember that a chain must have alternating strong and weak links. Since the ERs have internal strong links they must use external weak links.

I have changed at the same time you are quoting my reply. Please reconsider corrected one in my above post.

R. Jamil

[SpAce]

I must assume that StrmCkr's confirmation meant that your diagrams were correct (which they seem to be). I wouldn't presume anything about his opinion about the names you had chosen for those patterns. The first two (Empty Rectangle and Dual Empty Rectangle) are standard and match the given patterns, but the other three names are probably your own inventions, aren't they?

Just searched "Grouped Empty Rectangle" and found this post dated Wed Mar 08, 2006 4:45 am by Mike Barker using in point 42.

Ok, that proves that the term has been used before. However, it doesn't tell anything about what was meant by it. Look at the date, too. Terms have evolved since 2006. In any case, "Grouped" has a specific sudoku meaning in 2019, and any other use is certain to cause confusion and should be avoided. It's irrelevant what it was used for in 2006, but I doubt it was even any different from the current usage.

Unless you have some other source that describes what "Grouped Empty Rectangle" meant back then AND indicates it was the same as yours, I must still presume that you invented your own use for that term. I would be mildly shocked if Mike Barker used it for the same purpose as you, knowing that he used "Grouped" correctly with other patterns. I would rather presume that he used it for indicating a group link (the only correct use for it), and not for a group of ERs.

As far as "Dual Linked Empty Rectangle" wordings is concern, found this post dated Dec 07, 2016 6:15 am by StrmCkr.

Ok, it proves that StrmCkr has used that term, even for the same purpose as you it seems. However, even that doesn't mean it's a good term at all. "Dual Empty Rectangle" has a very specific and well-known meaning, similar to a "Siamese" fish. On the other hand, "Dual Linked Empty Rectangle" apparently means something very different -- a loop actually -- which makes the similarity of the name very confusing.

Even a minor change to "Doubly Linked Empty Rectangle" would be much better as it would imply a loop just like in Doubly Linked ALS XZ, and it wouldn't get as easily confused with the "Dual Empty Rectangle". Yet, personally I'd like something simple and descriptive such as "ER Loop", as I would similarly prefer "ALS Loop" to the unwieldy "Doubly Linked ALS XZ".

Hope it is clear now that the above terms are not invented by myself and/or overlooked by StrmCkr.

Yes, you've demonstrated that those terms have been used before, and the latter even for the same purpose. However, as far as the "Grouped Empty Rectangle" goes (which was the main point of this discussion), it seems to me that you've picked the term from a different context and invented a new use for it. That's actually worse than inventing the term yourself Also, "Dual Linked Empty Rectangle" is a poor term no matter who invented it (but that was a side branch of this discussion anyway).

That's what I thought you meant, and it's wrong. When multiple ERs are chained the ERIs aren't strongly but weakly linked between each other. The strong links are internal to the ERs, and the weak links between them make the chain propagate. Remember that a chain must have alternating strong and weak links. Since the ERs have internal strong links they must use external weak links.

I have changed at the same time you are quoting my reply. Please reconsider corrected one in my above post.

The only "correction" I see is a (bad) change in the order of the elements:

What I interpreted, may be wrong, but is that, one ERI in b2 is strongly linked with two ERIs in b3 and b5.

What I interpreted, may be wrong, but is that, one ERI in b3 is strongly linked with two ERIs in b2 and b5.

I thought the original order was actually more logical, but it doesn't matter because the main problem is still that you insist those ERIs are somehow strongly linked. They're not, no matter which order you put them in.

Look at your pattern in chain form (3D Eureka works nicely for ERs because it shows those box links more clearly):

(4): b3p[7=2] - b2p[12=4] - b5p[7=3] => -4 r4c7; stte

The '=' indicates a strong link, and they're all within the boxes (the ERs). The '-' indicates a weak link, and you find those between the ERs. Thus, your ERIs are weakly linked, if anything (and even that's a wrong interpretation). In fact, it's inaccurate to say that the ERs or ERIs are linked at all -- the individual candidates or groups of candidates within them are. It's probably better to say that the ERIs are connected, which doesn't incorrectly imply either strong or weak links between them.

[rjamil]

Hi SpAce,

Regret to inform that I am unable to conclude your above mentioned post.

Just what I grasped and accepted is to call "Doubly Linked Empty Rectangle" instead of "Dual Linked Empty Rectangle".

What should I use instead of:
1) Strong Link between two ERIs;
2) Grouped Empty Rectangle; and
3) Dual/Doubly Linked Grouped Empty Rectangle?

R. Jamil

[SpAce]

Regret to inform that I am unable to conclude your above mentioned post.

Just what I grasped and accepted is to call "Doubly Linked Empty Rectangle" instead of "Dual Linked Empty Rectangle".

What should I use instead of:
1) Strong Link between two ERIs;
2) Grouped Empty Rectangle; and
3) Dual/Doubly Linked Grouped Empty Rectangle?

Short answer: something that doesn't conflict with established terms. First you should probably strive to understand what those established terms are and really mean. Can you explain what "strong link" and "weak link" mean? Can you explain what "grouped" normally means? What about "loop"? Unless you have a clear idea of what they are, you shouldn't use them. Unless I can be sure you understand their common definitions, I think this discussion is fruitless. Also, I don't think this is the right place to have a detailed discussion about ER pattern naming in general, and I do think such a discussion should include StrmCkr as the ER expert.

Personally I have little interest in getting in very deep with ER naming anyway, because any of the ER patterns can be understood and expressed unambiguously using either Grouped X-Chains/Loops or equivalent fishes, and both of them are better generic tools. ERs are kind of cool patterns but they're also a niche, and it's mostly a wasted effort to try to catalogue and name all possible combinations of them. It's much more valuable to understand generic chaining, including grouped nodes and loops, because that covers all ER patterns and much more.

It seems to me that you've focused so much on a few specific pattern families that you've missed some very important general concepts, which makes communication difficult and also restricts your solving abilities. If you really want to improve your solving skills, I would suggest you learned more about generic chaining concepts because then you could understand and use almost any patterns instead of just the ones that you've specifically memorized. That includes ERs.

Like my martial arts instructor once told me: forget about learning individual techniques and tricks which only work in very specific situations, and concentrate on the fundamentals. The value of understanding the fundamentals is that you can then derive and apply every possible trick without specifically naming and memorizing them. The same principle works very well for sudoku too. There are very few fundamentals you need to learn to be able to understand and use almost all known solving techniques, and possibly come up with some new ones too (though unlikely these days). However, if you approach the problem by trying to learn every individual trick and pattern that has ever been published, you're mostly wasting your time.