The New Sudoku Players' Forum

[StrmCkr]

After some though the word eventual for canabaliatic eliminations refers to a subset smaller fish eliminating the base candidate in the larger fish... So current fish model won't hit these eliminations directly. As per previous post.

[sultan vinegar 2]

Firstly, I am 'sultan vinegar', but I have lost my old account details (moderators, I'm guessing we can't merge these accounts given I appear to have no way to prove I really am sultan vinegar?)

Secondly, I have not looked at sudoku for a couple of years, but this recent interest in fish reminds me of my post a couple of pages back which I never followed up:

One thing that I think is missing from the exemplars is rank 2 fish eliminations. I remember a while ago coming up with an extended list of fish exemplars; I think I had everything up to jellyfish size, basic, franken, mutant, finned, Siamese (like a skyscraper, not the Hodoku definition) and rank 2 fish. Then I got distracted by exocets. Ironically, my current thinking on exocets has been interrupted by these fish! I'll try and find my exemplars and post them, as I'm sure I would have missed some.

Given this fish guide appears to be in the process of significant revision, I will dig-out my old extended list of exemplars and post them in the next day or so. From memory, I used some different symbols to those used in the original guide because there are some additional concepts in my guide, but it shouldn't be a problem to come up with an agreed notation for any revision to this guide.

My advice to anyone learning about fish is that because the fish concept developed in an ad-hoc way as new fish concepts were found, some definitions are duplicates or redundant. Note I only found sudoku after the 'glory years' when this forum was at its peak. Most of the key contributors to this forum had moved on by then, but I was able to cherry-pick this excellent guide and sort out the different concepts in my own mind.

Finally, having studied fish extensively, I am happy to contribute to the revision of this guide and answer any questions others have about fish, although I'm sure I will need to refresh my own knowledge along the way given its been a couple of years.

SV

[tarek]

Hi SV and SV2 ,

I would advise any new fisherman to read the 1st post of this thread. Even in its current un-revised form would have more precise and Un-duplicated info. I would also advise therefore not to read anywhere else

Tarek

[sultan vinegar 2]

Below are my preliminary thoughts prior to posting exemplars in batches beginning with cyclops fish. Feedback (both good and bad) on the proposed structure is most welcome. It will be interesting to see how the revisions to this guide are received.

• All Sudokus can be solved in one step with a giant net; the restriction to considering the subset of a single digit (also known as a floor) only is a natural simplification especially for human players.
• The intention of this revised guide is to cover all single digit/floor eliminations, of which it may be argued that fish are one type, with other types being singles, locked candidates, x-chains (both open and closed, both grouped and ungrouped), and (single digit) Kraken fish.
• In their general form, fish represent nets, the simplest fish net being the swordfish made from three rows and three columns. Note some of these nets may be linearised to x-chains with the use of derived inferences, but others are unable to be simplified and remain as nets. Each player will have their own philosophy along some spectrum as to what methods constitute acceptable solving. Personally, I class these fish nets in the same category as Exocets; they comprise a subset of well defined patterns that human players are able to recognise, understand, and utilise.
• A difficulty with cataloguing a fish aquarium is that many accepted fish concepts change depending on how you look at the same fish; a situation that is not ideal for making robust definitions. These concepts include:
  • Base and cover sets
  • Fish size (e.g. swordfish vs jellyfish)
  • Fish shape (e.g. franken vs mutant)
  • Fins (e.g. exo-fins, endo-fins, remote or dorsal fins, whether a candidate is a fin or not)
  • Degeneracy (e.g. sashiminess, remora fish, cannibalism, Siamese fish)
• Another difficulty is that these accepted concepts were developed in an ad-hoc manner as new fish were discovered. Some of these concepts don't generalise well to more complex fish, and become redundant. If what is known about fish today was known from the start, things would be defined quite differently and be simpler to understand, but to some extent we remain bound to these legacy definitions.
• The intended structure of the guide is to consider one fish at a time, starting from the simplest and moving to the more advanced, noting the concept of simplicity is not able to be rigorously defined due to the points made above, and as such will be a personal opinion, chosen to make the tutorial as logical to follow as possible. This structure works particularly well for the fish at the beginning of the guide where it is common for one new concept to be introduced per fish. Each fish in the guide will:
  • Introduce a new fish concept if required, noting some fish may require multiple new concepts to be introduced together, while others may require no new concepts
  • Have the pattern defined with a template
  • Prove the logic of the elimination(s)
  • Be accompanied by a real example
  • Contain the nomenclature for how to notate the elimination(s)

[David P Bird]

SV,

Is this re-working dead or is it just all collaboration is being conducted via private messages/e-mails? As exotic fish are not my forte and some of my views are contentious, I've been happy to be one of the indians rather than a chief here, but in the hope that this project has not been abandoned, here are a few selected thoughts.

I am all for tidying up the fish descriptions as, frankly, the plethora of fancy terms is quite a burden for newcomers to absorb, and their payback in terms of extra puzzles that can be solved isn't that great.

In the Sudpedia Mirror a 'floor' is defined as "A part of the grid containing 3 rows and 3 boxes" so there's a clash. (I just call your floors digit maps.)

I favour simply treating the body of a fish as a one Boolean and the set of fin cells as another. In this way, whatever eliminations they provide can be notated as an AIC showing any additional links used. This eliminates the need to define kraken fish, remote fins, endo-fins etc. This view seems to agree with yours but I'm not certain.

This also applies to rival fish that have common eliminations that I also look for. I guess these could be combined into some sort of exotic fish but they would still need to be explained, so I've never tried.

When a fish has a single fin and a single potential elimination that see each other they can both be eliminated in a single step, but this is awkward to notate. One approach is to use an 'and so' construct:
(x)Fish:Cells = FinCell => PE cell <> x, &So => FinCell <> x

In addition to the examples that are planned, I also think that tips to help manual players to find fish would also be useful.

David PB
.

[sultan vinegar 2]

DPB,

It's not dead, it's just taking longer than I expected because I want a consistent approach between the easy fish and the hard fish, so I'm having to consider all the fish at once. The place where I'm struggling time-wise at the moment is finding the balance between describing the plethora of fancy terms (on one hand these need to be in the guide because people read about them and don't understand them, but on the other hand they really aren't necessary because they just over-complicate and confuse things), and writing a short, easy to follow guide.

Once I get the cyclops fish part posted, it should happen faster because the fundamentals (which I want to do a really good job with) will be covered by then.

The tips are a great idea. I know from extensive fishing that I don't have to rigidly look for each type of fish one by one like a machine; there are certain triggers that short-circuit what you should be looking for.

That, plus at the moment, not only can't I reply to PMs, but also my posts take time to be manually approved because SV foolishly lost his old account details.

SV

[David P Bird]

SV,

I'm pleased to hear you're still on the case, and I'm sorry that you have to endure the rigours of being considered a new member.

You also have my sympathy re trying to unify the approach across fish of different types as the problems are similar to the ones I faced in writing the JExocet Compendium.

One point about my 'rival fish' approach that I should have mentioned is that the links between them aren't confined to using only the fish digit which would be necessary to construct an exotic fish from them. This is only a minor consideration though as they are rare!

David
.

[blue]

When a fish has a single fin and a single potential elimination that see each other they can both be eliminated in a single step, but this is awkward to notate. One approach is to use an 'and so' construct:
(x)Fish:Cells = FinCell => PE cell <> x, &So => FinCell <> x

Every time someone brings this %$@# up, it does nothing but confuse me.
I finally figured out what you were on about, though.

Your claim, can be extended to any fish where every PE can see every fin.
It a fish like that exists, then (in the end) every PE (like usual) and every fin can be eliminated.

It's a completely different fish, though, that's responsible for the 2nd half of the eliminations.
They can also be done as follow on eliminations from a "same sized" finless fish, after eliminating the original PE's.

If you really think that the topic is worth covering then please explain things in detail, where it's covered, and don't encourage anyone to blow things off with a simple " &So " attribution.

Actually "please do XYZ" is irrelevant. I don't really care.
Instead, take: "IMHO, you should XYZ, and you shouldn't ..."

--

P.S.: I've been deliberately vague, here, as to what the 2nd fish really is, and how it's related to the first fish.
I'm curious whether anyone else, really understands what's going on here, and whether David remembers what base and cover sectors are.

[David P Bird]

When a fish has a single fin and a single potential elimination that see each other they can both be eliminated in a single step, but this is awkward to notate. One approach is to use an 'and so' construct:
(x)Fish:Cells = FinCell => PE cell <> x, &So => FinCell <> x

Every time someone brings this %$@# up, it does nothing but confuse me.
I finally figured out what you were on about, though.

Your claim, can be extended to any fish where every PE can see every fin.
It a fish like that exists, then (in the end) every PE (like usual) and every fin can be eliminated.

It's a completely different fish, though, that's responsible for the 2nd half of the eliminations.
They can also be done as follow on eliminations from a "same sized" finless fish, after eliminating the original PE's.

If you really think that the topic is worth covering then please explain things in detail, where it's covered, and don't encourage anyone to blow things off with a simple " &So " attribution.

Actually "please do XYZ" is irrelevant. I don't really care.
Instead, take: "IMHO, you should XYZ, and you shouldn't ..."

--

P.S.: I've been deliberately vague, here, as to what the 2nd fish really is, and how it's related to the first fish.
I'm curious whether anyone else, really understands what's going on here, and whether David remembers what base and cover sectors are.

Hi Blue,
Good to hear from you even if I have hit a nerve. I trust you feel better now that you have penned your response.

I don't claim to be an expert on every variety of fish but as such, perhaps my feedback gives those that are some idea of the confusion that can be caused by the existing coverage. If what I write is accurate then perhaps the way I present it may be of help. If not, then it reveals a point that I, and perhaps others, have failed to understand, which might help to improve the way it is presented.

It was a result of you correcting my use of the base and cover set terms that I realised that my working knowledge of fish was inadequate. I therefore went through the UFG and other fish based threads and made up my own notes which, after tidying up, I posted in my Fish Bones thread. In that post I filleted out terms that were only of interest to mathematical puritans and concentrated on the information that was useful to me in solving everyday puzzles or might be of use in the future. Two people, but not you, messaged me with comments and corrections which I used to edit it, but no-one posted a response.

To me base sets and cover sets are interchangeable. For a fish with fins and PE cells, if the base and cover sets are swapped fins become PEs and PEs become fins. So when I look for them I don't pre-judge which will be which. However, since your chastisement, I have tried to be careful in the way I write about them! I would add that the terms are not well chosen as a base set and a cover set are both cover sets except one is strong and the other weak. The fin term carries the fish analogy onwards but then we get PEs and EEs which don't. Being only two letters long, neither can be found by the forum's search engine. In my view mathematicians should never be trusted to coin their own terms, and other examples abound in the Sudoku world.

Finally, you are setting one rule for me and another for yourself when you say "If you really think that the topic is worth covering then please explain things in detail" and later "I've been deliberately vague, here". In other mathematical topics you have contributed to, various terms have been coined and later modified in scattered posts perhaps in different threads. This makes it impossible for anyone that is not deeply involved in the discussion to make head or tail of the subject. However, you never made the same criticism there.

Cheers David
.

[blue]

I trust you feel better now that you have penned your response.

No, I'm still miffed that I wasted time on this same thing ... yet again.
Thanks for the thought, though.

Finally, you are setting one rule for me and another for yourself when you say "If you really think that the topic is worth covering then please explain things in detail" and later "I've been deliberately vague, here".

No, it's just that I don't think it's worth covering, and I don't want any part of it.
Since my opinion is often in the minority, I'll just finish this post, and leave it at that.

To me base sets and cover sets are interchangeable.

When the "sets" are the usual "sector" sets, it's true that they can be used in either way. If we were talking about a strongly linked set of candidates where it is possible for more than just one of them to be true, then it could only be used as a base set. As you no doubt know, the usual elimination rules remain valid, if we allow sets like that to be used in place of base sectors. Fins and potential eliminations can be defined in an analagous way, and so on. My view of fish, has always been one that allows a strongly linked set of candidates (like that) to be used in place of a base sector (and a set of pairwise weakly linked candidates to be used in place of a cover sector). In that view -- which admittedly "off topic" here -- the "2nd fish" -- the one that you left unmentioned -- wouldn't necessarily exist.

Cheers,
Blue.

[David P Bird]

Hi again Blue,

If we were talking about a strongly linked set of candidates where it is possible for more than just one of them to be true, then it could only be used as a base set.

Now we have identified one bone of contention - I wasn't and you were.

It's also possible to consider an NxN fish with a single fin (node) and a single PE (node) that are (fully) in sight of each other to be a recognisable pattern lacking a descriptor that allows both to be eliminated.

Since my opinion is often in the minority ...

Hey that's my line! - except it's 'always' not 'often'.

Cheers, David
.

[champagne]

studying the fish guide to have it coded in my program, I got this fish coming in a puzzle from the interactive games
...........12.34...2.1.5.6..75...18...........89...74..6.3.9.5...47.19...........;6.50;4.20;4.00

with only 4 truths and 11 eliminations, it looks like a "franken jelly fish.
2 of the eliminations are cannibalistic (2 links for a candidate in the truths).

p6,5 fish.png (32.53 KiB) Viewed 1002 times

[yzfwsf]

My solver can find this as shown in the figure below

fish.PNG (29.45 KiB) Viewed 999 times

[champagne]

My solver can find this as shown in the figure below

Hi yzfwsf,
completely different, but same number of truths,
also cannibalism, but in other places.
Interesting
A side question is "what is the ranking of this compared to other rules"??

[yzfwsf]

HI champagne:
For me, when POM can't find the equivalent fish, the solver will go through all combinations and become inefficient, so it is placed in a very low priority, just before force Net.