The New Sudoku Players' Forum

[DonM]

If I correctly interpret Don's comments over the last week or two, he recommends:

(3=9)r2c4-(9=157)r153c5-(7=3)r2c6=>r2c7<>3

Are we foregoing labels all together now? I'm fine win ANS, but what has happened to ALS? Same-same?

Perhaps he wishes to avoid the ALS vs ANS "debate."

BTW I don't recall bennys, the originator of the ALS-xz and the ALS-xy-wing, ever using the ANS term.

To clarify my position: I strongly support the use of labels. Those who think they are redundant or unnecessary are IMO ignoring the fact that some of those reading the notation are not as experienced as they are and don't recognized a non-labeled ALS (or whatever) onsight. However, I'm not going to go to the mat on the use of labels- one has to know when to hold'em and when to fold'em.

I prefer ALS over ANS. I really don't know where ANS came from and find its use perplexing seeing as how virtually all the threads on the subject going back to 2005 used ALS. In addition Locked Sets has been the term used frequently in tutorials; I don't recall seeing them referred to as Naked Sets. I used ANS in my post above (since that's what the poster used) only because I wanted to address the one issue I am most adamant about- that of using the (9=157)r153c5 format and not muddy the waters by changing ANS to ALS.

[Luke]

ALS! ALS! ALS!

(Sorry. That's the soccer hooligan in me coming out...)

[daj95376]

My concern is that (IMO) the notation used for the ANS does not reflect what's happening in an ALS/ANS:

<snip>

On the other hand, if it is written (as has been IMO the standard for years, except for the discretionary label), the actual logic flow as notated not only is accurate, but makes it easier for the reader to understand:

(3=9)r2c4-ANS(9=157)r135c5-(7=3)r2c6 loop

Hmmm!!! Okay, now please fill in the missing part of this version of the loop:

(3=7)r2c6 - ANS( ??? )r135c5 - (9=3)r2c4 - loop

Are you running into a lack of consistency in the ALS/ANS expression???

[DonM]

My concern is that (IMO) the notation used for the ANS does not reflect what's happening in an ALS/ANS:

<snip>

On the other hand, if it is written (as has been IMO the standard for years, except for the discretionary label), the actual logic flow as notated not only is accurate, but makes it easier for the reader to understand:

(3=9)r2c4-ANS(9=157)r135c5-(7=3)r2c6 loop

Hmmm!!! Okay, now please fill in the missing part of this version of the loop:

(3=7)r2c6 - ANS( ??? )r135c5 - (9=3)r2c4 - loop

Are you running into a lack of consistency in the ALS/ANS expression???

I only answer fill-in-the-blank questions from my grade school teacher who, unfortunately, passed away 2 decades ago, but I'll humor you this one time.

(3=7)r2c6 - ANS(7=159 )r135c5 - (9=3)r2c4 - loop

I'm wondering why you didn't ask this question years ago when the format I'm supporting was first used and when it was used in countless solutions of puzzles you've seen & posted on before. But I'll answer it this way: I read and write chains from left to right. My suggested format works logically forward and backward, but in order to make the notation mirror the logic flow as clearly as possible, an ALS has to be notated differently if you choose to start the AIC from the other end.

[daj95376]

I'm wondering why you didn't ask this question years ago when the format I'm supporting was first used and when it was used in countless solutions of puzzles you've seen & posted on before. But I'll answer it this way: I read and write chains from left to right. My suggested format works logically forward and backward, but in order to make the notation mirror the logic flow as clearly as possible, an ALS has to be notated differently if you choose to start the AIC from the other end.

Hello Don,

Good question. The answer comes in three parts. First, I tried to stay away from ALS logic and notation for a very long time. When I was later forced to deal with its existence, I did what I could to understand what was happening, and why there was an inconsistency in the notation compared to how I viewed what was happening. It wasn't until after Luke451 posted an ALS solution containing a loop, with eliminations that I hadn't expected, that I became critical of ALS notation. Finally, recent discussions on the "correct" choice for ALS notation drove me to express my notational perspective on ALS logic.

Normally, I accept most ALS expressions as simply a way of presenting the "players" and their "locations", as opposed to a formal representation of the underlying relationships. I even use solutions following this premise because they are compact and understandable to informed readers. It's only when people start getting bitchy about the notation that I get critical.

Regards, Danny

[DonM]

I'm wondering why you didn't ask this question years ago when the format I'm supporting was first used and when it was used in countless solutions of puzzles you've seen & posted on before. But I'll answer it this way: I read and write chains from left to right. My suggested format works logically forward and backward, but in order to make the notation mirror the logic flow as clearly as possible, an ALS has to be notated differently if you choose to start the AIC from the other end.

Hello Don,

Good question. The answer comes in three parts. First, I tried to stay away from ALS logic and notation for a very long time. When I was later forced to deal with its existence, I did what I could to understand what was happening, and why there was an inconsistency in the notation compared to how I viewed what was happening. It wasn't until after Luke451 posted an ALS solution containing a loop, with eliminations that I hadn't expected, that I became critical of ALS notation. Finally, recent discussions on the "correct" choice for ALS notation drove me to express my notational perspective on ALS logic.

It appears that there is a marked disparity in the time we've been considering ALS notation. I've been entering manual solutions using the Eureka notation since latter 2006 and it seemed of particular importance that I adhere close to a standard that was set by some pretty smart folks especially considering that they were presenting their own solutions as well as reading and considering mine. And following a standard wasn't important just because people like Myth Jellies & Ruud were more knowledgeable than me, but also because their ALS format intuitively presented the actual ALS logic flow better than anything anyone else came up with.

Normally, I accept most ALS expressions as simply a way of presenting the "players" and their "locations", as opposed to a formal representation of the underlying relationships. I even use solutions following this premise because they are compact and understandable to informed readers. It's only when people start getting bitchy about the notation that I get critical.
Regards, Danny

I guess we have different views of 'bitchy'. I don't call it bitchy when someone is simply a) calling for some consistency and adherence to a few standards that were set long ago in the use of a notation that has been around for 6 years and b) suggesting that the the notation be as clear as possible (such as the use of labels) for those less solving-sophisticated. Besides, isn't there a more constructive reason to be critical than just because someone is (albeit incorrectly) perceived as being 'bitchy'.

Anyway, I may have missed it, but have you come up with an (allegedly ) better ALS notation that shows the actual logic flow and is reasonably easily interpreted by even newer solvers?

[eleven]

BTW I don't recall bennys, the originator of the ALS-xz and the ALS-xy-wing, ever using the ANS term.

A pity for me, that he expressed it that way, maybe to make it a pattern rather than a chain. Then all the solvers integrated it and left me with a hard-to-spot pattern - unless when i am looking for single ANS's instead as starting points.

[tlanglet]

Miss a day and miss all the action.

3r2c46=3r1c6-(3=7)r2c6-ALS(1579)r135c5[7r3c5=9r1c5]-(9=3)r2c4 => r2c4<>3

Now I need to review all the action.....

Ted

[daj95376]

Anyway, I may have missed it, but have you come up with an (allegedly ) better ALS notation that shows the actual logic flow and is reasonably easily interpreted by even newer solvers?

I also admit that there are smarter people than myself in this forum -- probably everyone. But, I don't see any of them using Myth Jellies cumbersome boolean notation for an ALS.

First, I'm perfectly happy with the following notation for an ALS: (4=69)r12c9. L-to-R says that the absence of <4> in these two cells forces the Locked Set <69>, and R-to-L says that the absence of the Locked Set <69> in these two cells forces one of the cells to contain <4>.

Second, I'm also perfectly happy with the following notation for an ALS: (9=7)r135c5. No matter which way you read it, it says that the absence of one candidate forces the presence of the other. However, using this notation doesn't fully describe the underlying activity in the ALS. I resolved this in my earlier post by breaking the compact notation into two smaller components.

(3=9)r2c4 - (9=15)r15c5 - (15=7)r53c5 - (7=3)r2c6 - loop -> r1c6,r2c37<>3, r9c5<>15

Here, the weak <15> inference in cells r135c5 justifies the eliminations in r9c5.

So, sometimes, you have to decide between convenience or a fully qualified relationship. Most choose convenience, and I accept that. However, I do admit that some of the expressions being passed off as ALS are stretching the "convenience" option to its limit.

Now, think back to the good-ole-days when a locked set -- pair, triple, quad -- only performed eliminations in the house/unit containing the locked set. The (9=7)r135c5 ALS is confined to unit [c5], yet the subsequent elimination -7r2c6 is not in [c5]. You don't want to know how I adjust the notation to justify this property. SuDoKu has changed a lot since 2006!

Regards, Danny

[DonM]

Anyway, I may have missed it, but have you come up with an (allegedly ) better ALS notation that shows the actual logic flow and is reasonably easily interpreted by even newer solvers?

But, I don't see any of them using Myth Jellies cumbersome boolean notation for an ALS.

Let's keep it real: Myth Jellies notation was simply the (7=159 )r135c5 format. He often added ampersands to signify AND to constructs involving ALS and otherwise, but never suggested that everyone should. Myth was an amazing combination of theorist, innovator, teacher and solver so he was always trying to make Boolean relationships apparent in his posts.

First, I'm perfectly happy with the following notation for an ALS: (4=69)r12c9. L-to-R says that the absence of <4> in these two cells forces the Locked Set <69>, and R-to-L says that the absence of the Locked Set <69> in these two cells forces one of the cells to contain <4>.

Second, I'm also perfectly happy with the following notation for an ALS: (9=7)r135c5. No matter which way you read it, it says that the absence of one candidate forces the presence of the other. However, using this notation doesn't fully describe the underlying activity in the ALS. I resolved this in my earlier post by breaking the compact notation into two smaller components.

(3=9)r2c4 - (9=15)r15c5 - (15=7)r53c5 - (7=3)r2c6 - loop -> r1c6,r2c37<>3, r9c5<>15

Here, the weak <15> inference in cells r135c5 justifies the eliminations in r9c5.

So, sometimes, you have to decide between convenience or a fully qualified relationship. Most choose convenience, and I accept that. However, I do admit that some of the expressions being passed off as ALS are stretching the "convenience" option to its limit. Now, think back to the good-ole-days when a locked set -- pair, triple, quad -- only performed eliminations in the house/unit containing the locked set. The (9=7)r135c5 ALS is confined to unit [c5], yet the subsequent elimination -7r2c6 is not in [c5]. You don't want to know how I adjust the notation to justify this property.

I thought you were going to show me something that was an improvement over (7=159 )r135c5 instead of an attempt to sacrifice clarity in the name of making the notation totally reversible and/or compatible with 'extended' ALSs. I have seen no sign that the solvers in this forum are feeling the need to do anything but read chains from left to right.

SuDoKu has changed a lot since 2006!

Exactly! Which is why I don't understand why something that was put to bed long ago is being resurrected as if this is 2006-7 all over again. The fact is- and you should know this having posted there- that advanced manual solving of puzzles far more difficult than those here continued on the Eureka forum until recently. That included UK Extreme puzzles often posted by GreenLantern and the ER>=9 and above 'Puzzles From Others' presented by TTT, Steve K et al until around 2010-2011.

That type of solving tested the Eureka notation to its limits and, particularly when it came to the UK Extremes, virtually everyone was adhering fairly closely to the same Eureka notation standards. There isn't anything going on in this forum that hasn't already been seen on that forum so the premise that some allegedly new concepts or discoveries require fiddling with standards already set is IMO a non-starter.

[storm_norm22]

There isn't anything going on in this forum that hasn't already been seen on that forum so the premise that some allegedly new concepts or discoveries require fiddling with standards already set is IMO a non-starter.

well said, agreed.

[daj95376]

I thought you were going to show me something that was an improvement over (7=159 )r135c5 instead of an attempt to sacrifice clarity in the name of making the notation totally reversible and/or compatible with 'extended' ALSs. I have seen no sign that the solvers in this forum are feeling the need to do anything but read chains from left to right.

And I thought that you wanted a notation that newbies to ALS could follow what was happening. However, in my zealous to achieve that goal, I missed the forest for the trees on a usable notation.

Code: Select all

Without the loop, I'd use: (9  =  7)r135c5
With    the loop, I'd use: (915=157)r135c5


The presence of <15> on each side of the strong inference should be sufficient to justify the eliminations in r9c5.