Name this Turbot Fish type!

Advanced methods and approaches for solving Sudoku puzzles

Re: Name this Turbot Fish type!

Postby SpAce » Fri Nov 16, 2018 5:09 am

StrmCkr wrote:
SpAce wrote:In fact, should we make the connection explicit: LoadER Crane ? Or is that stupid?

nope, that was what i was hinting at with the ER separated.

Which question did you answer? :) (Btw, did you see my edited version in which I answered that question from my current point of view? I think the lower-case is neater and actually symbolizes the minimal ER better.)

this one will be tricky to explain but ill attempt. patterns understood conjugated cells in a different light cells in a sector both could be true but not both. so, we built patterns around that idea.
that a sector is locked to holding a truth.
...
if you want you can add fancier words and explain the connections and call it a "chain" i can defiantly see how you would do that and can admit they would be similar to a chain/coloring comparitivly.

Yes, it seems like they're just different ways of approaching the same thing. I think chains are a simple and compact way to express that "pattern" logic you described -- at least when linear chains are applicable. When nets are required, then things are a bit different. In those cases fishes or truths/links or templates approaches have more value. I guess what you mean by "pattern" logic can include any of them depending on the situation, so perhaps it's more of a hybrid approach.

like this which could also fall under the turbot header as 4 grouped links using ERI's

That's a nice pattern, but I'd just call it a grouped X-Loop or a Mutant Jellyfish (bbbb\rrcc). It's too long and complicated to belong to the Turbot Fish family, so I would not count it under that header. I guess the terms Turbot Chain, or extended Turbot Fish, were sometimes used for longer X-Chains, but I don't really see any reason for that. I think the generic X-Chain/Loop is a fine name for anything longer than four nodes. The Turbot Fish family is valuable exactly because the patterns are few and simple to recognize. Complicated generalizations of simple concepts tend to make simple things harder, especially if the same naming is used for both simple and complex cases (that was my point with keeping the Simple Coloring simple as well).
SpAce
 
Posts: 713
Joined: 22 May 2017

Re: Name this Turbot Fish type!

Postby StrmCkr » Fri Nov 16, 2018 9:36 am

Or is that stupid?
nope.

A side note: when you use digit filters in houdoku or manual you are actually swapped to nc, NR, Nb view
conversion of Nb back to row/col format from squares notation is a pain does it help not so much, I rarely even use it in my code out side subsets.

the one that helps more is this: i convert the whole grid to all 27 sectors.

N[sector] saving Cell * {Square technically} {using another program to convert the square to cell if i need }


Ps that link u mentioned is the bases for Nb, Nr, Nc space. that's prevalent in most of today's solvers.

Code: Select all
It's too long and complicated to belong to the Turbot Fish family,
only long and stupid when it uses the grouped links instead of ERI which only needs 4 links :)
but yeah smaller versions of turbots stick to conjugated links and grouped links and an eri {with in the loader crane/tower crane setup}
now if it could use "links" types then the eri variations are defiantly applicable to "turbots" and naming would balloon tremendously as pointed out by

Postby Jeff » 26 Dec 2005 23:20
Havard,

I can see your point, but this pattern can appear in many shapes that you would be very busy trying to give each one a different name.


i already know multiple that would work easily with the "3-4" link restriction. {i had all the ERI's classed & named for size 1 -> 7 before the crash of the old forum}

1 eri with opposite direction conjugate
1 eri with 2 same direction conjugated links {end points connected or not}
2 eri's with connecting conjugate
2 connected ERI with non connected conjugate
2 non connected eri's connected by a conjugate
2 non connected ERi's each with 1 conjugate connected strongly or weakly.
3 ERi's all connected {like a 1 digit xy-wing}
4 eri's all connected

so you can restrict its usage how you wish to keep it simple.

Perhaps I've missed some hidden benefit of that point of view?
no, manual solving has all the spaces superimposed onto what you call pencil marks and the human brain can separate stuff easily.
----if a program could highlight groups of digits then hidden subsets would also jump out easier then naked. ---- instead of filters for 1 digit at a time.


computer code however has no understanding of difference from RC space to Rn space {what digit is on} and NR space {what spot is still on}

having both points of view allows us to search smaller size volumes quicker, instead of say looking for a quintuple naked set to find a hidden quad.

the extra spaces helps us find things with computer aid quicker they aren't really for manual solver unless you have some kind of aiding device
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 889
Joined: 05 September 2006

Re: Name this Turbot Fish type!

Postby SpAce » Fri Nov 16, 2018 3:38 pm

StrmCkr wrote:only long and stupid when it uses the grouped links instead of ERI which only needs 4 links :)

I'm not sure if I understand how you use those ERIs and count the links. Seems to me that your ERI approach is something of a hybrid between chaining and fishing, but I might have misunderstood something. In any case, if efficiency is the goal, isn't the pure fish view here the shortest way to express the logic, and probably the simplest too for those who understand mutants? I'd still say the grouped X-Loop wins the first prize just because it's the most versatile tool and probably the easiest to understand for most players.

Jeff wrote:Havard, I can see your point, but this pattern can appear in many shapes that you would be very busy trying to give each one a different name.

Actually, I don't think Jeff really saw Havard's point :) The important thing about naming is that only relevant variations need their own names, which is exactly what Havard did (except, unfortunately, he didn't name one relevant type, which is why we're here). He'd found the exact sweet spot of naming without going too deep into subtyping. Too fine-grained an approach just makes things more complicated, which is why I've never bothered to memorize the various one-letter Wing-types and their zillion subtypes (to me they're just short AICs, unless I really feel like looking up the name). He managed to avoid that pitfall.

Related to that, I'd like to comment something else Jeff said right after that:

Jeff wrote:Other names for this pattern are turbot fish, fishy-cycle, multi-colouring, x-cycle, double chain, combination nice loop, advanced colouring, 3D-medusa, super-colouring and simple-colouring.

A bit ironic that most of those other names or techniques he mentioned are long dead, but Skyscrapers and 2-String Kites keep on living -- and hopefully Loader Crane will join them! :)

i already know multiple that would work easily with the "3-4" link restriction. {i had all the ERI's classed & named for size 1 -> 7 before the crash of the old forum}

I understand that back then that stuff was new and you were a trail-blazer, which required more detailed case-by-case analysis to come up with the generic rules. Now that we understand the generic rules better -- thanks to the respectable pioneering efforts by the likes of you -- there's no need for such detail, imho. Once you have so many different cases, classifying and naming them all runs the risk of what I said about getting too fine-grained for practical purposes, at least for a manual player who'd need a look-up table to document any solution.

the extra spaces helps us find things with computer aid quicker they aren't really for manual solver unless you have some kind of aiding device

I guess Denis Berthier would disagree :) Doesn't his p&p solving method use multiple grids with different spaces at the same time? Like I said, I tried a variation of that style (using nr & nc spaces instead of his rn & cn, and skipping the useless nb/bn) for a couple of times, but didn't really see the point as I could find pretty much the same stuff in all of the grids, just with different points of view (naked vs hidden, bivalue vs bilocation links, etc). I guess that approach only makes sense for a manual player whose skillset is limited to seeing things from just one perspective (e.g. naked stuff and bivalue links). The only thing it really did help with was finding (finned) basic fishes which turned into (A)LSs (like with the arcilla method).
Last edited by SpAce on Fri Nov 16, 2018 5:42 pm, edited 1 time in total.
SpAce
 
Posts: 713
Joined: 22 May 2017

Re: Name this Turbot Fish type!

Postby tarek » Fri Nov 16, 2018 5:12 pm

Going back a little and again branching regarding fish;

denis_berthier wrote:
arcilla wrote:I have tried to bring box-bound links into the picture (representing franken stuff?) but I didn't succeed. The boxnumber of a cell is not an independent coordinate. Also, I still know too little of the very advanced sudoku theory: I'm actually just a happy intermediate sudoku player.


It's normal that you could not extend your idea to boxes. Only block-free rules can be transposed to the rn- and cn- spaces.
In these spaces, I proved that only Latin Squares rules are valid.


It is appealing to have our 1d mapping to get our locked set but it has to be without boxes. Has anyone though of box free mutant fish to see if that it is still feasible???

Tarek
User avatar
tarek
 
Posts: 2761
Joined: 05 January 2006

Re: Name this Turbot Fish type!

Postby StrmCkr » Fri Nov 16, 2018 6:34 pm

Code: Select all
It is appealing to have our 1d mapping to get our locked set but it has to be without boxes. Has anyone though of box free mutant fish to see if that it is still feasible???


depends, if we can convert the sets of the fish construct to an equivalent that wont use boxes. ie to use Row/Col {only}

http://forum.enjoysudoku.com/about-the-arithmetic-of-ultimate-fish-t5256.html#p41058 <- for reference on how to do that.. {don't really have a better link atm}

however it must me noted the eliminations changes as the base/cover changes as well.


I'm not sure if I understand how you use those ERIs and count the links.


me either, was pretty late when i wrote it. I was counting the * cells as 1 cell which matches turbots.

Seems to me that your ERI approach is something of a hybrid between chaining and fishing, but I might have misunderstood something. In any case, if efficiency is the goal, isn't the pure fish view here the shortest way to express the logic, and probably the simplest too for those who understand mutants? I'd still say the grouped X-Loop wins the first prize just because it's the most versatile tool and probably the easiest to understand for most players.


ERI Space for the Mutant Jellyfish (bbbb\rrcc)
Code: Select all
+---------+---------+---------+
| .  .  . | .  .  . | .  .  . |
| .  1  . | .  .  . | .  1  . |
| .  .  . | .  .  . | .  .  . |
+---------+---------+---------+
| .  .  . | .  .  . | .  .  . |
| .  .  . | .  .  . | .  .  . |
| .  .  . | .  .  . | .  .  . |
+---------+---------+---------+
| .  .  . | .  .  . | .  .  . |
| .  1  . | .  .  . | .  1  . |
| .  .  . | .  .  . | .  .  . |
+---------+---------+---------+

shows up as an x-wing, you just have to remember you are in { BOX &{ R,C}} space then the eliminations rules are the same:
Eliminate from same Row/Col of a Box but not from the used boxes.
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 889
Joined: 05 September 2006

Re: Name this Turbot Fish type!

Postby StrmCkr » Tue Nov 20, 2018 8:33 am

another side note:

tow - Er Crane

{where the tow is a mixed up version of two : the number of cells in the minimal case one. )


load - Er Crane

{loaded could be the reminder its a full box "loaded" with more then 2 candidates}

bonus the shapes still match.

alright so this grid blown up into the extra spaces
Code: Select all
.-----------------.---------------------.-----------------------.
| 16   346  13468 | 2456  469    24569  | 12346  45679  1235789 |
| 7    5    468   | 1     3      2469   | 246    469    24689   |
| 2    9    1346  | 8     7      456    | 1346   456    13456   |
:-----------------+---------------------+-----------------------:
| 69   236  23569 | 246   14689  123489 | 146    4567   14567   |
| 4    1    3569  | 7     69     369    | 8      2      56      |
| 8    7    26    | 246   5      1246   | 9      3      146     |
:-----------------+---------------------+-----------------------:
| 169  246  12479 | 456   1468   14578  | 2346   469    23469   |
| 5    246  12469 | 3     146    146    | 7      8      2469    |
| 3    8    467   | 9     2      467    | 5      1      46      |
'-----------------'---------------------'-----------------------'


to read the chart:
Code: Select all
 Row/Col/Box are in 0 base  (0..8) isntead of (1..9)

 Row(Top-> down) -  #(Left -> Right) - col {listed in cell} 

position: (0 base)
( left -> right ) for a row
(Top -> down) for a col
( left -> right )  & (Top -> down) for a box




notice the ERI for digit 6. boxes: 0,1,3,5,8 @ position 3 {which is the hub for Row/Col in each box}
Code: Select all
+----------------------+------------------------+-------------------------+
| 1(6)  34(6)  1348(6) | 245(6)  49(6)  2459(6) | 12346   45679   1235789 |
| 7     5      48(6)   | 1       3      249(6)  | 246     469     24689   |
| 2     9      134(6)  | 8       7      45(6)   | 1346    456     13456   |
+----------------------+------------------------+-------------------------+
| 9(6)  23(6)  2359(6) | 246     14689  123489  | 14(6)   457(6)  1457(6) |
| 4     1      359(6)  | 7       69     369     | 8       2       5(6)    |
| 8     7      2(6)    | 246     5      1246    | 9       3       14(6)   |
+----------------------+------------------------+-------------------------+
| 169   246    12479-6 | 456     1468   14578-6 | 234(6)  49(6)   2349(6) |
| 5     246    12469   | 3       146    146     | 7       8       249(6)  |
| 3     8      467     | 9       2      467     | 5       1       4(6)    |
+----------------------+------------------------+-------------------------+
Attachments
pm.PNG
pm.PNG (64.28 KiB) Viewed 85 times
Last edited by StrmCkr on Fri Nov 23, 2018 11:33 am, edited 2 times in total.
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 889
Joined: 05 September 2006

Re: Name this Turbot Fish type!

Postby SpAce » Fri Nov 23, 2018 7:57 am

StrmCkr wrote:another side note:
tow - Er Crane
{where the tow is a mixed up version of two : the number of cells in the minimal case one. )
load - Er Crane
{loaded could be the reminder its a full box "loaded" with more then 2 candidates}
bonus the shapes still match.

I get your logic, but let's not complicate things with more word/letter plays and even switching our chosen names around! I think it's more important that the names and the shapes match, and the original order matches better, imho (i.e. minimal: Loader Crane; grouped: Tower Crane). I'm fine with the optional capitalization of ER in either or both, but I definitely don't want any weird spaces or hyphens in the names (not really sure if you were suggesting that or just highlighting the logic). It just gets ugly and complicated and ensures that no one will ever use those names (which might be the likely outcome anyway). I'm not too fond of the partial capitalization scheme either (i.e. LoadEr, TowEr). My suggestion is to stick with these:

Minimal (ungrouped) ER:

"Official" Turbot Name: Loader Crane (also acceptable: LoadER Crane, Turbot Crane)
Nicknames: Crane, Loader, LoadER
Discouraged: "Turbot Fish", "ER" (unless referring to family memberships) -- else we've gained nothing

Non-minimal (grouped) ER:

"Official" Grouped Turbot Name: TowER Crane (also acceptable: Tower Crane, Grouped Crane, ER)
Nicknames: TowER, Tower
Not Acceptable: plain "Crane" or "Turbot Crane" -- else we've gained an even worse chaos

I'm fine with many variants as long as they don't clash with each other or any other names currently in use. Since our original goal was to distinguish the minimal and non-minimal ER variants within the Turbot family, I think the critical rule is that the minimal ER is not called just "ER" or "Turbot Fish" and even more critically that the non-minimal ER is never called just "Crane" or "Turbot Crane". Any other listed variant and nickname should ensure everyone can understand which variant is being discussed. I'm willing to let the users decide which variants become the de facto standard, if the whole idea ever gets off the ground in the first place (which might be doubtful anyway, especially since the most popular software solvers will never get updated to use those names).

alright so this grid blown up into the extra spaces

Interesting chart!

notice the ERI for digit 6. boxes: 0,1,3,5,8 @ position 3 {which is the hub for Row/Col in each box}

Yep! Easy to see those eliminations. However, it misses the rest of the fish eliminations.
SpAce
 
Posts: 713
Joined: 22 May 2017

Re: Name this Turbot Fish type!

Postby StrmCkr » Fri Nov 23, 2018 11:27 am

just highlighting the logic)


side note:
i have the ERI displayed as a reduced format: its actually a 3d space per box{instead of 2d}, as each box should only be using the 4 boxes they can see indirectly instead of all space... makes reading it a lot harder then needed ...as you can manually check the nodes to see if they all link up. .......

notice the ERI for digit 6. boxes: 0,1,3,5,8 @ position 3 {which is the hub for Row/Col in each box}
Yep! Easy to see those eliminations. However, it misses the rest of the fish eliminations.

there is only 2 eliminations for that fish, so not sure what else it misses?

{eri space is a 3d representation of space we can add R/c/b or MIni R/C or Box R/C space back to it to expand eliminations}

if we add Row 9 to the eri boxes we get the bottom fish with 18 eliminations.

(just a note: all cells in a "row" have to do is see/force a direction change to an eri block, which explains how the over lap in R9C9 works)

using the chart its Row - # * col on Digit 6 {note the 2,5,8) covert the cordial for eri position 2 for each box and it changes to R123456789, COL 2,5,8 {which matches R9 digit 6 Cols}

Code: Select all
+-----------------------+--------------------------+--------------------------+
| 1(6)  34(6)  1348(-6) | 245(6)  49(6)   2459(-6) | 1234-6  4579-6  1235789  |
| 7     5      48(6)    | 1       3       249(6)   | 246     469     2489-6   |
| 2     9      134(6)   | 8       7       45(6)    | 1346    456     1345-6   |
+-----------------------+--------------------------+--------------------------+
| 9(6)  23(6)  2359(-6) | 24-6    1489-6  123489   | 14(6)   457(6)  1457(-6) |
| 4     1      359(6)   | 7       69      39-6     | 8       2       5(6)     |
| 8     7      2(6)     | 246     5       124-6    | 9       3       14(6)    |
+-----------------------+--------------------------+--------------------------+
| 169   246    12479-6  | 456     1468    14578-6  | 234(6)  49(6)   2349(-6) |
| 5     246    1249-6   | 3       146     14-6     | 7       8       249(-6)  |
| 3     8      47(6)    | 9       2       47(6)    | 5       1       4(6)     |
+-----------------------+--------------------------+--------------------------+
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 889
Joined: 05 September 2006

Re: Name this Turbot Fish type!

Postby SpAce » Tue Nov 27, 2018 10:07 pm

StrmCkr, I have a feeling that you'd really like the ER-connection to show in the Loader Crane, and after some rethinking there might be a way that I could find acceptable too. How about LoadeR Crane? One reason why I don't really like "LoadER" or "LoadEr" is that they aren't visually pleasing to me. I tracked the problem to the lack of contrast between the lower-case 'd' and the upper-case 'E' next to each other because of their same height. "TowER" doesn't suffer from that, and neither does "LoadeR", because both 'w' and 'e' are shorter than 'E' or 'R', which makes those words more readable. "LoadeR" would be logical too, because 'eR' would still mean "minimal ER" (but more explicitly than "er").

So, my latest suggestion is to use "LoadeR Crane" and "TowER Crane" as the "official" Turbot names for those two variants. Would that work for you?
SpAce
 
Posts: 713
Joined: 22 May 2017

Re: Name this Turbot Fish type!

Postby StrmCkr » Wed Nov 28, 2018 8:09 am

So, my latest suggestion is to use "LoadeR Crane" and "TowER Crane" as the "official" Turbot names for those two variants. Would that work for you?


how about just

Loader Crane andTower Crane

and those that use the name decide how they'd like to write it. :) .......... { like i use 2-string kyte, instead of two string kite.}

arranging the letters lower/upper is pretty much user choice.

i mostly did it to point out the ER was still in the name :)

i'll leave it in your hands to determine best approach as this is your thread.
Some do, some teach, the rest look it up.
User avatar
StrmCkr
 
Posts: 889
Joined: 05 September 2006

Re: Name this Turbot Fish type!

Postby SpAce » Wed Nov 28, 2018 11:10 am

StrmCkr wrote:how about just

Loader Crane and Tower Crane

Well, I'm personally very much fine with that. Simplicity and clarity are always good goals.

and those that use the name decide how they'd like to write it. :) .......... { like i use 2-string kyte, instead of two string kite.}

arranging the letters lower/upper is pretty much user choice.

I agree, which is why I already listed multiple acceptable variants earlier. However, I think it would be preferable to have some "official" name. I'd also like to discourage some variants, such as LoadEr and TowEr, because somehow they seem ugly to me.

i mostly did it to point out the ER was still in the name :)

This would have been simpler if it weren't or if you hadn't :D

i'll leave it in your hands to determine best approach as this is your thread.

Thanks... but if I had a clear idea what the best approach is, I would have already pushed it more strongly :) I guess the only way to know is to let the market forces decide.
SpAce
 
Posts: 713
Joined: 22 May 2017

Previous

Return to Advanced solving techniques