The New Sudoku Players' Forum

[StrmCkr]

Dynamic forcing networks that are nested chains pretty much the end of the Linefor solving techniques...

Having the rating get harder the more depth it explores: sure you could go that route, but your still using the same technique just going deeper
(Which is why it caps at 12)
There is already some degree differences from 10-12 based on different styles of forcing chains, anything post 9.8 se ratings usually are in the realm of needing aids to solver for even hardened veterans.
Occasionally these have exotic patern that make it easier but many don't)

[Alpxcx]

try this:

5....78...1..8..2...46....7..34.6...............2.86..1....59...5..6..4...23....1

Pat # 289 # (258812) #

Code: Select all


 5 . . | . . 7 | 8 . .
 . 1 . | . 8 . | . 2 .
 . . 4 | 6 . . | . . 7
-------+-------+------
 . . 3 | 4 . 6 | . . .
 . . . | . . . | . . .
 . . . | 2 . 8 | 6 . .
-------+-------+------
 1 . . | . . 5 | 9 . .
 . 5 . | . 6 . | . 4 .
 . . 2 | 3 . . | . . 1



This one actually works quite well with the solvers. The quad was not very visible.
Despite it was valid after 3 moves (hidden pair 15r56c3, locked candidates then hidden single 8r9c8), I wouldn't do it in this way.
The locked candidates and pairs after this were more visible so they would be applied before the quad.
Actually, they were not skippable because applying the quad earlier wouldn't make it taking less steps to be solved.
I would probably find a finned X-wing (5r34c58 fr4c9) before the quad, but that would be an extra move.
Interesting.

[Alpxcx]

Pss. Some of us can solve advance moves without any pms: can everyone definitely not.

more advance players increase bi-locals to include all 5 types of strong links on a single digit.

before moving into full pencil markups {for als moves}

problem with this markup approach is that it can be easy to mistake bivalves for overlapping strong links. {or a player doesn't move to the next step finalizing all pms} --- i see this frequently on reddit where players stall on synder notes and lock them selves up.

Yes, the way people solve puzzles are very different. Some do synder notes while some don't. Even with synder notes, only a small portion of people can master them.

A very important thing I learnt was that, not everyone solves puzzles in the same way. This makes a uni-rating reflecting every individual's effort needed to solve a puzzle a non-existance.

If I make a synder note based rating, probably it won't be as popular as SER, because it can go even less creditable.
Because of this, I won't make it a priority anymore if I say I will make another own solver.

[Alpxcx]

Dynamic forcing networks that are nested chains pretty much the end of the Linefor solving techniques...

Having the rating get harder the more depth it explores: sure you could go that route, but your still using the same technique just going deeper
(Which is why it caps at 12)
There is already some degree differences from 10-12 based on different styles of forcing chains, anything post 9.8 se ratings usually are in the realm of needing aids to solver for even hardened veterans.
Occasionally these have exotic patern that make it easier but many don't)

I understand that any puzzle which need dynamic forcing chains to solve where at least one DFC isn't covered by any exsiting simpler technique, is quite impractical.
Because, you can't usually find it anywhere in the real world, like in a commercial sudoku app, on newspapers or even in a sudoku championship.
Let's talk about something simpler.
Let's say an XY-chain has a length of 10 (10 different cells) and an ALS-XZ has a length of 5, how would you compare them?
On another occasion, there is an XY-chain involving 4 cells and an ALS-XZ involving 9 cells, how would you compare them?
SudokuExplainer/SudokuWiki/HoDoKu/YZF all do infinite depth search on each technique, and it can be worse than the above cases.
This was why I said it won't be just fixed by swapping the orders. Priority wasn't the only issue.

[StrmCkr]

Als xz can also written as aic chain wth 1 weaklink . (N cells at each link)

À xy chain with 10 length has 10 cells with 10 strong links and 9 weak links
An xy chain follows a visible path via Bi vavles in the pencil marks....

Same thing for the other example als has 1 weaklink, xy chain has 4 strong limks and 3 weak links

Comparitivly als would be shorter, however
Als moves require knowing: n cells with n+1 candidates using the internal strong link to connect to the next als

This concept is outside of most people's ability without practice as the als size can scale easily with understanding this concept and many don't know how they work let alone there eliminations, nor do most understand a naked pair is the simplistic als xz possible.

Compared to following a or b amongst bivavles and worrying only about the first and last cell having the same digits
Is easier then checking a group of cells from the als to make the eliminations.
And again even fewer know that a xy chain is also an als chain.

Overlap in constructs is huge as there is 2 ways to solve a grid:

Graphing theory (a. I. C)
Set theory (subsets, als, DDs, fish)

And these two overlap 100% being able to translate from one to the other

But I know this dosent mean your average joe solver will

Ps hodoku doesnt do infite depth it's limited (20) and uses templates to build backwards for eliminations plus defaults to each elimination to having 1 chain only..
yzfs also had limits in place to stop infite depth and duplicates etc.

[Alpxcx]

Als xz can also written as aic chain wth 1 weaklink . (N cells at each link)

À xy chain with 10 length has 10 cells with 10 strong links and 9 weak links
An xy chain follows a visible path via Bi vavles in the pencil marks....

Same thing for the other example als has 1 weaklink, xy chain has 4 strong limks and 3 weak links

Comparitivly als would be shorter, however
Als moves require knowing: n cells with n+1 candidates using the internal strong link to connect to the next als

This concept is outside of most people's ability without practice as the als size can scale easily with understanding this concept and many don't know how they work let alone there eliminations, nor do most understand a naked pair is the simplistic als xz possible.

Compared to following a or b amongst bivavles and worrying only about the first and last cell having the same digits
Is easier then checking a group of cells from the als to make the eliminations.
And again even fewer know that a xy chain is also an als chain.
...

Scaled up ALS structrues are generally hard to find. However, there is an exception. If you find an ALS structure and it has some deductions, then you may find a slightly larger ALS structure based on the one you already found.
This is easier than finding the large ALS at the start. Again, one may solve in a diffeent way but this is what i think.

XY chains can be easily found if there are a lot of bi-value cells, but the start and the end isn't always clear. If there is an XY chain involving 10 cells, one may need to do a throughout search to find what deductions can be made with a start and an end, as there are multiple combinations of starts and ends. One may start writing an another piece of paper to fill both assumptions in, but it would be a forcing chain view by doing this. I don't think this is much easier than finding a few ALSs.
There are rectangular shaped XY loops. There are 4 XY chains inside one, but most software only make deductions on only one row/column and move on. If there are simpler techniques available, they use it, and finally they may come back to this XY loop and recognise it as another XY chain. Those "simpler" techniques aren't always more visible than an already found structure. Not all manual solvers know XY loops or XY loops as the simplest form of MSLS, but they will know there are actually 4 XY chains inside this kind of structures immediately when those dumb "machines" don't (well, YZF's approached it by adding an XY-Cycle technique but less priotised than XY-Chain, which is confusing).

If one can find a larger ALS quickly based on an already found one, or discover 4 XY chains at the same time, why would one not apply these but instead looks for "simpler" techniques?

Now about concepts. I believe that only a small portion of people even knows how a naked subset works. They can understand this by doing cell forcing chains. However, it's more likely that they don't even try to understand, instead they just memorise this and make all deductions. Same with all other techniques. You don't have to understand how it works to work around it. For example, laws are extremely complex but you don't need to read thousands of pages to follow them.

The way how a lot of manual solvers do is interesting. I saw a lot of posts on reddit, where some people give hints about a hard technique when there is no need to search for those.
You may have seen this, so I probably don't need examples here.
Some people do everything in a forcing chain way, then name the technique of the forcing chain they find. This is very different from the technique based approach. What would exsiting ratings mean to them?

[StrmCkr]

Not all manual solvers know XY loops or XY loops as the simplest form of MSLS

they aren't msls

MSLS are ALS and AHS combined

What would exsiting ratings mean to them?

what would a rating mean for anyone that doesn't have any faint understanding of what they are doing.

it be just gibberish, as much as their attempts at guessing and testing cells and digits and following the path to solution or contradiction.
which is what all techniques can fall down to " i tried this and x happened"

[Alpxcx]

Not all manual solvers know XY loops or XY loops as the simplest form of MSLS

they aren't msls

MSLS are ALS and AHS combined

I can show you this example.
The XY loop as an image:

It can be expressed as 4 ALSs (in anticlockwise order):
8r1c28
4r15c2
3r5c28
6r15c8

nor do most understand a naked pair is the simplistic als xz possible.

If you really mean this, then you mean a single digit in two cells can be seen as an ALS. This is based on the same logic.
I don't see why this can't be seen as an MSLS, because the above "table form" is exclusive to MSLS.

[Alpxcx]

Well, I apologise for not explaining the terms I used. "XY loop" is not a very common term for rectangular shaped ring-type XY chains. Any of these structures is equivalent to 4 XY chains, and even for anyone who knows XY chains but not these structures, they can find all 4 XY chains at the same time.

[StrmCkr]

XY loop" is not a very common term for rectangular shaped ring-type XY chains.

i know what you ment, no need to applogize it is either 4 cells as a xy-ring or greater then and a xy loop of x cells.

the xy -ring can be reduced to this:
Almost Locked Set XZ-Rule: A=r1c28 {468},B=r5c28 {346}, X=46, Z=/ => r5c5<>3 r4c2<>4 r6c8<>6

or as a stand alone 4 cell als chain, or xy-chain etc.

where is the HIDDEN set ? this is N cells sharing +2 digits to lock the sets to them selves.

msls is a balance of Naked and hidden sets so that N sectors is locked for N digits in N cells. { ie the hidden set is locked x times and the naked set is also locked x times for the balance}

Code: Select all

+-----------------+---------------+----------------+
| 3   4(8)    2   | 9    1      5 | 46  6(8)    7  |
| 1   9       7   | 46   68     3 | 2   5       48 |
| 5   6       48  | 24   28     7 | 9   1       3  |
+-----------------+---------------+----------------+
| 6   -4(38)  458 | 1    35     2 | 45  7       9  |
| 27  4(3)    459 | 567  569-3  8 | 1   6(3)    24 |
| 27  1       59  | 567  3569   4 | 56  -6(38)  28 |
+-----------------+---------------+----------------+
| 4   7       3   | 25   25     6 | 8   9       1  |
| 9   2       6   | 8    7      1 | 3   4       5  |
| 8   5       1   | 3    4      9 | 7   2       6  |
+-----------------+---------------+----------------+

this is the hidden set version of the same loop
AHS- xz

a msls combines both types to balance the sectors selected.
as its balancing full sets of digits 1-9 as a set of naked & a complementary set of hidden in each sector.

[Alpxcx]

I will try to understand the MSLS part later. I thought it was one because it follows the same deduction rule.
I haven't seen a clear and common definition of an MSLS yet (with a document and demonstration).
So this was how I learnt:
pattern rule → deduction rule
rather than:
pattern rule → internal logic → pattern rule → deduction rule
In this example of the XY pattern, I think the pattern rule is very similar to MSLS.
Something relating this happened on 2019, when I wrote my SimpleStrategicSudokuSolver.
I completed the code for finned swordfish, but because I learnt the rule by looking at a few examples generated by HoDoKu, I only considered the most common case.
The pattern rule I used was swordfish + 1 fin only. I didn't understand there can be 2 fins on a row or there can be 2 rows with fins when they are all in the same block.
Also, all finned X-Wings can be finned swordfishes on another view. This has caused a serious bug which I didn't understand.
However, if I was only solving a puzzle, this may have worked. My route can still work for the most common case.

Now about concepts. I believe that only a small portion of people even knows how a naked subset works. They can understand this by doing cell forcing chains. However, it's more likely that they don't even try to understand, instead they just memorise this and make all deductions. Same with all other techniques. You don't have to understand how it works to work around it.

I don't think I was the only one who did this.

[StrmCkr]

your last common, pretty much most that aren't learned will place x and follow it and find a conclusion and later on try to figure out what actually caused it: i see this frequently on reddit

MSLS

as for your fish i can help fix the bug if you have a link for the source code

my working fish code is found here
broken down by class and a find all near the bottom

most fish in my point of view are N sectors with N covers and at most +2 extra covers to solve {as a Nxn+ K solver } or another way is nxn + fins seen in the U.F.G
then its a 2 phase elimination rule setup for exo fins {external fins} and endo fins {internal}
the endo fish version are the ones that cause the most headaches for me when i first coded it