The New Sudoku Players' Forum

[Mauriès Robert]

Hi Eleven,
It's your choice, it's your opinion, I respect it. But they have no more value than mine to solve with the TDP, so accept it without denigrating it.
Sincerely
Robert

[champagne]

I am an admirer of the "Art of Sudoku Solving", as it has been practised by RW, SteveK, MythJellies, ..., David Bird and totuan. It is characterized by nice ideas showing the beauty and variety of puzzles, and therefore including many techniques, they developed through their insight of sudoku properties. Not missing a plan (which might change through the solving process), how to effectively crack a puzzle.
The opposite is backtracking, the most effective way of solving puzzles with a program.
Systems like TDP or Resolution Rules are between these approaches, and for my taste too mechanical. This does not change much, when other ideas (like uniqueness) are integrated.

Hi eleven,

I am somewhere in between both positions.
Programming is my hobby, so I use intensively computers.
My first program was designed using a multi colouring technique (full tagging). It worked well, but I noticed that skill players did not work in this way.
Having solving techniques not used by skill manual players is of small interest.
(but skfr was written using this frame).
To day, in the solving field, I apply (part of) the so called “basic moves” and some of the “extended basic moves”, but I also work to find new solving rules that a manual solver can apply when the set of known rules fails or leads to a boring path made of tens/hundreds of long single eliminations.

The exocet and later the Jexocet have been designed with the cooperation of programs and manual solvers (Allan Barker model, “abi” work,” ttt/”totuan” examples of nets ...)
The TLG0/rank 0 logic has been progressively enriched in similar ways.

In many cases, the TLG0 was described in former times as “pigeon hole matrix”. This powerful technique had small success. I suspect that the main reason is that the engine to produce them has never been described.

David as you write made a terrific job with the JExocets..

All programs are more or less doing some T&E plus back tracking to produce nice paths. I see small chances to get out of this if we accept that a computer must produce a path that a manual player has a chance to find. I have been impressed years ago studying the eliminations produced by ”abi” and “ttt/totuan”.

At this point, I give small chances to the TDP rules to be used intensively by manual players, and I understand the frustration of “totuan” who spends hours to write “in diagram” his path when the alternative appears as a list of TDP “titles”.

I am not a specialist of the “names” and “notations”, I understand that manuals players like to have an agreed condensed notation to describe the path. Often, I can’t read it without the commented pm.

“totuan” has it’s own way to do the job, and most of us have no problem to understand the logic of the diagram, but again, it is time consuming.

This to say that use of other ways to decribe the logic is accepted. But most of us are asking for a feeling of the complexity of an elimination done.

[Mauriès Robert]

[url][/url]Hi Champagne,
I read with interest what you write, but I would like to comment on the following sentence.

At this point, I give small chances to the TDP rules to be used intensively by manual players, and I understand the frustration of “totuan” who spends hours to write “in diagram” his path when the alternative appears as a list of TDP “titles”.

Make no mistake, the TDP is mainly for manual players (In France, I count many players who use it). It's very easy to draw two tracks together with a pencil, and on the puzzles usually used in the public one or two tracks hunts are enough. These players are the ones for whom advanced techniques are difficult to memorize and their patterns difficult to spot. With TDP there is little to remember and only a little experience to gain in order to choose the right track generators.
You talk about Tetouan's frustration with the lack of explanation of resolutions by TDP. I recognize that TDP is a rather visual process, which is practiced directly by tracing on the puzzle, and that the textual description of the sequences constituting a track is not the priority. I tried most of the time in my contributions on forum.enjoysudoku.com, seeing that this was the rule here, to make these visual descriptions, but with a notation better adapted to that than the Eureka language: a->b->c that everyone understands, and with Tetouan-type diagrams (see "My resolution width TDP" http://forum.enjoysudoku.com/robert-s-puzzles-2019-12-25-t36989.html#p285899). But one must also make the effort to understand that in TDP one builds with basic techniques and not by alternating strong and weak links.
Sincerely
Robert

[champagne]

Hi Robert

I read with interest what you write, but I would like to comment on the following sentence.

At this point, I give small chances to the TDP rules to be used intensively by manual players, and I understand the frustration of “totuan” who spends hours to write “in diagram” his path when the alternative appears as a list of TDP “titles”.

Make no mistake, the TDP is mainly for manual players (In France, I count many players who use it). It's very easy to draw two tracks together with a pencil, and on the puzzles usually used in the public one or two tracks hunts are enough. These players are the ones for whom advanced techniques are difficult to memorize and their patterns difficult to spot. With TDP there is little to remember and only a little experience to gain in order to choose the right track generators.
Robert

This sounds like the situation where I was years ago with the full tagging process. I had also my small “club of fans” and I could push some players far from having “abi” and “ttt” skills to solve puzzles till the area of rating 8.5 (SE rating).

I recognize that TDP is a rather visual process, which is practiced directly by tracing on the puzzle, and that the textual description of the sequences constituting a track is not the priority. I tried most of the time in my contributions on forum.enjoysudoku.com, seeing that this was the rule here, to make these visual descriptions.....

Again, I see no compulsory rule and except for AICs where I try to follow the agreed notation, I use many other ways to explain what I am doing.

The only constraint, to be convincing, is to have a clear explanation of what is done. You have seen recently some images produced by “yzfwsf”’s solver, it was perfect.

[denis_berthier]

That's a relatively hard one, as it requires whips[7]

Hidden Text: Show

Code: Select all

finned-swordfish-in-rows: n6{r3 r9 r6}{c2 c7 c6} ==> r4c6 ≠ 6
whip[1]: c6n6{r9 .} ==> r7c4 ≠ 6
whip[4]: r5n5{c7 c6} - r9n5{c6 c5} - r8n5{c6 c2} - r1n5{c2 .} ==> r7c7 ≠ 5
whip[5]: c2n4{r8 r2} - r7n4{c2 c8} - c8n6{r7 r1} - r3n6{c7 c2} - c2n2{r3 .} ==> r8c3 ≠ 4
whip[4]: r6c5{n1 n8} - r9c5{n8 n5} - r8c4{n5 n9} - r8c3{n9 .} ==> r8c5 ≠ 1
whip[5]: c5n9{r3 r8} - r8c3{n9 n1} - r9c3{n1 n8} - r9c5{n8 n5} - r8c4{n5 .} ==> r3c5 ≠ 1
whip[5]: c5n9{r3 r8} - r8c3{n9 n1} - r8c4{n1 n5} - c5n5{r8 r1} - c5n7{r1 .} ==> r3c5 ≠ 8
whip[6]: r5n5{c6 c7} - r5n9{c7 c3} - r8c3{n9 n1} - r9c3{n1 n8} - r7c1{n8 n4} - r5n4{c1 .} ==> r7c6 ≠ 5
whip[5]: c3n6{r4 r1} - c8n6{r1 r7} - r7c7{n6 n1} - r7c6{n1 n8} - b7n8{r7c1 .} ==> r4c3 ≠ 8
whip[6]: r3c5{n9 n7} - r8n7{c5 c6} - c6n2{r8 r9} - c7n2{r9 r2} - r3n2{c9 c2} - r3n6{c2 .} ==> r3c7 ≠ 9
whip[6]: r6n3{c9 c4} - r6n6{c4 c2} - c3n6{r4 r1} - c8n6{r1 r7} - c8n4{r7 r8} - r8n3{c8 .} ==> r4c9 ≠ 3
whip[6]: c7n7{r3 r4} - c3n7{r4 r2} - c6n7{r2 r8} - c6n2{r8 r9} - r9n6{c6 c7} - b3n6{r1c7 .} ==> r1c8 ≠ 7
whip[6]: r8n3{c9 c8} - r6n3{c8 c4} - r6n6{c4 c2} - c3n6{r4 r1} - c8n6{r1 r7} - c8n4{r7 .} ==> r2c9 ≠ 3
whip[7]: c3n6{r4 r1} - c3n7{r1 r2} - c3n4{r2 r5} - c3n9{r5 r8} - c5n9{r8 r3} - r3n7{c5 c7} - r3n6{c7 .} ==> r4c3 ≠ 1
whip[7]: c3n7{r2 r4} - c3n6{r4 r1} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 .} ==> r2c1 ≠ 7
whip[7]: r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - c3n7{r2 r4} - c3n6{r4 .} ==> r1c2 ≠ 6
whip[7]: c3n6{r4 r1} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - b4n7{r6c1 .} ==> r4c3 ≠ 9
whip[4]: r6n6{c4 c2} - r4c3{n6 n7} - r6c1{n7 n1} - r6c5{n1 .} ==> r6c4 ≠ 8
whip[4]: r6n6{c4 c2} - r4c3{n6 n7} - r6c1{n7 n8} - r6c5{n8 .} ==> r6c4 ≠ 1
whip[4]: r6n7{c8 c1} - r4c3{n7 n6} - r6c2{n6 n1} - r6c5{n1 .} ==> r6c8 ≠ 8
whip[4]: r6n7{c8 c1} - r4c3{n7 n6} - r6c2{n6 n8} - r6c5{n8 .} ==> r6c8 ≠ 1
whip[4]: r6c5{n1 n8} - r6c1{n8 n7} - r4c3{n7 n6} - r6c2{n6 .} ==> r6c9 ≠ 1
whip[4]: r6c5{n8 n1} - r6c1{n1 n7} - r4c3{n7 n6} - r6c2{n6 .} ==> r6c9 ≠ 8
biv-chain[5]: b6n2{r6c9 r6c8} - r6n7{c8 c1} - r4c3{n7 n6} - b1n6{r1c3 r3c2} - b1n2{r3c2 r2c2} ==> r2c9 ≠ 2
whip[5]: r7n9{c4 c2} - b7n5{r7c2 r8c2} - b7n4{r8c2 r7c1} - r5n4{c1 c3} - c3n9{r5 .} ==> r7c4 ≠ 5
whip[1]: r7n5{c2 .} ==> r8c2 ≠ 5
whip[5]: r4c3{n6 n7} - b6n7{r4c7 r6c8} - r6n2{c8 c9} - r6n3{c9 c4} - r6n6{c4 .} ==> r4c2 ≠ 6
whip[5]: r3n6{c2 c7} - r3n2{c7 c9} - r6c9{n2 n3} - r6c4{n3 n6} - c2n6{r6 .} ==> r3c2 ≠ 8
whip[5]: r3n6{c2 c7} - r3n2{c7 c9} - r6c9{n2 n3} - r6c4{n3 n6} - c2n6{r6 .} ==> r3c2 ≠ 1
whip[6]: r8n3{c8 c9} - r6c9{n3 n2} - b9n2{r8c9 r9c7} - r3n2{c7 c2} - r3n6{c2 c7} - r7c7{n6 .} ==> r8c8 ≠ 1
whip[7]: b4n7{r6c1 r4c3} - c3n6{r4 r1} - c8n6{r1 r7} - r7c7{n6 n1} - r7c6{n1 n8} - b5n8{r5c6 r4c4} - r4n6{c4 .} ==> r6c1 ≠ 8
whip[7]: r3n2{c9 c2} - r3n6{c2 c7} - r7c7{n6 n1} - r9c7{n1 n5} - r9c9{n5 n2} - r8c9{n2 n3} - r6c9{n3 .} ==> r2c7 ≠ 2
whip[5]: r8n3{c8 c9} - r6n3{c9 c4} - r6n6{c4 c2} - r3c2{n6 n2} - b3n2{r3c9 .} ==> r2c8 ≠ 3
whip[6]: c7n2{r9 r3} - c8n2{r2 r6} - r6n7{c8 c1} - r3n7{c1 c5} - r8n7{c5 c6} - r8n2{c6 .} ==> r9c9 ≠ 2
hidden-pairs-in-a-row: r9{n2 n6}{c6 c7} ==> r9c7 ≠ 5, r9c7 ≠ 1, r9c6 ≠ 8, r9c6 ≠ 5, r9c6 ≠ 1
whip[1]: b9n5{r9c9 .} ==> r4c9 ≠ 5
finned-x-wing-in-rows: n8{r9 r6}{c5 c3} ==> r5c3 ≠ 8
finned-x-wing-in-rows: n8{r6 r9}{c5 c2} ==> r7c2 ≠ 8
whip[4]: b7n8{r9c3 r7c1} - c1n5{r7 r2} - r2n4{c1 c2} - b7n4{r7c2 .} ==> r2c3 ≠ 8
whip[5]: r9c3{n8 n1} - r8c3{n1 n9} - r5c3{n9 n4} - c1n4{r5 r2} - c1n5{r2 .} ==> r7c1 ≠ 8
hidden-single-in-a-block ==> r9c3 = 8
whip[3]: r8c3{n1 n9} - r8c4{n9 n5} - r9c5{n5 .} ==> r8c6 ≠ 1
whip[5]: c4n6{r4 r6} - c4n3{r6 r2} - c4n5{r2 r8} - c5n5{r8 r1} - c5n8{r1 .} ==> r4c4 ≠ 8
whip[5]: r6n1{c2 c5} - c5n8{r6 r1} - c4n8{r3 r7} - r7n9{c4 c2} - b4n9{r4c2 .} ==> r5c3 ≠ 1
whip[4]: r8c3{n1 n9} - r5c3{n9 n4} - c1n4{r5 r2} - c1n5{r2 .} ==> r7c1 ≠ 1
whip[5]: r7c7{n1 n6} - r7c6{n6 n8} - b5n8{r5c6 r6c5} - r6c2{n8 n6} - r3n6{c2 .} ==> r7c2 ≠ 1
whip[1]: b7n1{r8c3 .} ==> r8c4 ≠ 1, r8c9 ≠ 1
whip[5]: r6n1{c2 c5} - r9c5{n1 n5} - r8c4{n5 n9} - c3n9{r8 r5} - r5n4{c3 .} ==> r5c1 ≠ 1
whip[3]: c3n6{r1 r4} - b4n7{r4c3 r6c1} - c1n1{r6 .} ==> r1c3 ≠ 1
naked-pairs-in-a-column: c3{r1 r4}{n6 n7} ==> r2c3 ≠ 7
biv-chain[5]: r1c3{n7 n6} - r3n6{c2 c7} - c7n2{r3 r9} - c6n2{r9 r8} - b8n7{r8c6 r8c5} ==> r1c5 ≠ 7
hidden-pairs-in-a-column: c5{n7 n9}{r3 r8} ==> r8c5 ≠ 5
whip[6]: b6n5{r4c7 r5c7} - c7n9{r5 r2} - b3n3{r2c7 r1c8} - c6n3{r1 r2} - r2n7{c6 c8} - b6n7{r4c8 .} ==> r4c7 ≠ 3
whip[1]: c7n3{r2 .} ==> r1c8 ≠ 3
whip[6]: r5c1{n4 n8} - r5c8{n8 n1} - r7c8{n1 n6} - r1c8{n6 n8} - c2n8{r1 r2} - c2n4{r2 .} ==> r7c1 ≠ 4
naked-single ==> r7c1 = 5
whip[1]: b7n4{r8c2 .} ==> r2c2 ≠ 4
biv-chain[5]: r8n3{c8 c9} - c9n5{r8 r9} - c5n5{r9 r1} - b1n5{r1c2 r2c2} - r2n2{c2 c8} ==> r8c8 ≠ 2
biv-chain[4]: c8n2{r6 r2} - b1n2{r2c2 r3c2} - c2n6{r3 r6} - r6c4{n6 n3} ==> r6c8 ≠ 3
biv-chain[5]: b9n2{r9c7 r8c9} - c9n5{r8 r9} - c5n5{r9 r1} - b1n5{r1c2 r2c2} - r2n2{c2 c8} ==> r3c7 ≠ 2
singles ==> r9c7 = 2, r9c6 = 6, r8c6 = 2, r8c5 = 7, r3c5 = 9
finned-x-wing-in-rows: n7{r3 r6}{c1 c7} ==> r4c7 ≠ 7
whip[1]: b6n7{r6c8 .} ==> r2c8 ≠ 7
hidden-triplets-in-a-block: b6{r6c8 r6c9 r4c8}{n2 n3 n7} ==> r4c8 ≠ 8, r4c8 ≠ 1
biv-chain[3]: r3n7{c7 c1} - r1c3{n7 n6} - r3n6{c2 c7} ==> r3c7 ≠ 1
biv-chain[4]: r7c6{n8 n1} - r7c7{n1 n6} - r3c7{n6 n7} - r2n7{c7 c6} ==> r2c6 ≠ 8
whip[4]: b5n5{r5c6 r4c4} - r4n6{c4 c3} - c3n7{r4 r1} - b2n7{r1c6 .} ==> r2c6 ≠ 5
biv-chain[4]: r2n5{c4 c2} - r2n2{c2 c8} - c9n2{r3 r6} - r6n3{c9 c4} ==> r2c4 ≠ 3
whip[1]: c4n3{r6 .} ==> r4c6 ≠ 3
hidden-pairs-in-a-block: b5{r4c4 r6c4}{n3 n6} ==> r4c4 ≠ 5, r4c4 ≠ 1
whip[1]: b5n5{r5c6 .} ==> r1c6 ≠ 5
hidden-pairs-in-a-column: c6{n3 n7}{r1 r2} ==> r2c6 ≠ 1, r1c6 ≠ 8, r1c6 ≠ 1
hidden-pairs-in-a-row: r2{n3 n7}{c6 c7} ==> r2c7 ≠ 9, r2c7 ≠ 1
hidden-single-in-a-block ==> r2c9 = 9
naked-pairs-in-a-block: b6{r4c9 r5c8}{n1 n8} ==> r5c7 ≠ 1, r4c7 ≠ 1
finned-x-wing-in-columns: n1{c7 c4}{r7 r1} ==> r1c5 ≠ 1
whip[1]: b2n1{r3c4 .} ==> r7c4 ≠ 1
biv-chain[3]: b3n2{r2c8 r3c9} - c9n8{r3 r4} - b6n1{r4c9 r5c8} ==> r2c8 ≠ 1
biv-chain[3]: r4c9{n8 n1} - r9n1{c9 c5} - r6c5{n1 n8} ==> r4c6 ≠ 8
biv-chain[3]: r2n5{c2 c4} - r1c5{n5 n8} - r6n8{c5 c2} ==> r2c2 ≠ 8
finned-swordfish-in-columns: n8{c2 c5 c9}{r4 r6 r1} ==> r1c8 ≠ 8
hidden-pairs-in-a-block: b3{r2c8 r3c9}{n2 n8} ==> r3c9 ≠ 1
whip[1]: b3n1{r1c8 .} ==> r1c2 ≠ 1
x-wing-in-rows: n8{r1 r6}{c2 c5} ==> r4c2 ≠ 8
stte

For those who don't like patterns like whips that depend on the target, there's also a solution with t-whips, but then the length goes up to 9 (instead of 7 for whips).

Hidden Text: Show

Code: Select all

finned-swordfish-in-rows: n6{r3 r9 r6}{c2 c7 c6} ==> r4c6 ≠ 6
whip[1]: c6n6{r9 .} ==> r7c4 ≠ 6
t-whip[5]: r7n4{c2 c8} - c8n6{r7 r1} - r3n6{c7 c2} - c2n2{r3 r2} - c2n4{r2 .} ==> r8c3 ≠ 4
t-whip[5]: c5n9{r3 r8} - r8c3{n9 n1} - r8c4{n1 n5} - c5n5{r8 r1} - c5n7{r1 .} ==> r3c5 ≠ 1
t-whip[5]: c5n9{r3 r8} - r8c3{n9 n1} - r8c4{n1 n5} - c5n5{r8 r1} - c5n7{r1 .} ==> r3c5 ≠ 8
t-whip[6]: r8n3{c9 c8} - c8n4{r8 r7} - c8n6{r7 r1} - r3n6{c7 c2} - r6n6{c2 c4} - r6n3{c4 .} ==> r2c9 ≠ 3
t-whip[6]: r8n3{c9 c8} - c8n4{r8 r7} - c8n6{r7 r1} - r3n6{c7 c2} - r6n6{c2 c4} - r6n3{c4 .} ==> r4c9 ≠ 3
t-whip[7]: c7n7{r3 r4} - r6n7{c8 c1} - r3n7{c1 c5} - c6n7{r2 r8} - c6n2{r8 r9} - r9n6{c6 c7} - c8n6{r7 .} ==> r1c8 ≠ 7
t-whip[7]: r3c5{n9 n7} - c6n7{r2 r8} - c6n2{r8 r9} - r9n6{c6 c7} - r3n6{c7 c2} - c2n2{r3 r2} - c7n2{r2 .} ==> r3c7 ≠ 9
t-whip[7]: c3n6{r4 r1} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - c3n7{r2 .} ==> r4c3 ≠ 1
t-whip[7]: c3n6{r4 r1} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - c3n7{r2 .} ==> r4c3 ≠ 8
t-whip[7]: c3n6{r4 r1} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - c3n7{r2 .} ==> r4c3 ≠ 9
t-whip[5]: r4c3{n6 n7} - r6n7{c1 c8} - r6n2{c8 c9} - r6n3{c9 c4} - r6n6{c4 .} ==> r4c2 ≠ 6
t-whip[5]: r6n6{c4 c2} - r4c3{n6 n7} - b6n7{r4c7 r6c8} - r6n2{c8 c9} - r6n3{c9 .} ==> r6c4 ≠ 1
t-whip[5]: r6n6{c4 c2} - r4c3{n6 n7} - b6n7{r4c7 r6c8} - r6n2{c8 c9} - r6n3{c9 .} ==> r6c4 ≠ 8
t-whip[5]: r7n9{c4 c2} - c3n9{r8 r5} - r5n4{c3 c1} - b7n4{r7c1 r8c2} - b7n5{r8c2 .} ==> r7c4 ≠ 5
t-whip[4]: b7n5{r7c2 r8c2} - c1n5{r7 r2} - c4n5{r2 r4} - b6n5{r4c7 .} ==> r7c7 ≠ 5
t-whip[5]: r6n2{c9 c8} - r6n7{c8 c1} - r4c3{n7 n6} - c4n6{r4 r6} - r6n3{c4 .} ==> r6c9 ≠ 1
t-whip[5]: r6n2{c9 c8} - r6n7{c8 c1} - r4c3{n7 n6} - c4n6{r4 r6} - r6n3{c4 .} ==> r6c9 ≠ 8
biv-chain[6]: r6c9{n2 n3} - b9n3{r8c9 r8c8} - c8n4{r8 r7} - c8n6{r7 r1} - r3n6{c7 c2} - b1n2{r3c2 r2c2} ==> r2c9 ≠ 2
t-whip[6]: r3n6{c2 c7} - c8n6{r1 r7} - c8n4{r7 r8} - r8n3{c8 c9} - r6c9{n3 n2} - r3n2{c9 .} ==> r3c2 ≠ 1
t-whip[6]: r3n6{c2 c7} - c8n6{r1 r7} - c8n4{r7 r8} - r8n3{c8 c9} - r6c9{n3 n2} - r3n2{c9 .} ==> r3c2 ≠ 8
t-whip[7]: c3n7{r2 r4} - b4n6{r4c3 r6c2} - r3n6{c2 c7} - r9n6{c7 c6} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 .} ==> r2c1 ≠ 7
t-whip[7]: r3n2{c9 c2} - r3n6{c2 c7} - c8n6{r1 r7} - c8n4{r7 r8} - r8n3{c8 c9} - r6c9{n3 n2} - b9n2{r8c9 .} ==> r2c7 ≠ 2
t-whip[5]: c7n3{r2 r4} - r6n3{c9 c4} - r6n6{c4 c2} - r3c2{n6 n2} - b3n2{r3c9 .} ==> r2c8 ≠ 3
t-whip[7]: r3n6{c2 c7} - c7n2{r3 r9} - c6n2{r9 r8} - r8n7{c6 c5} - r3n7{c5 c1} - b4n7{r6c1 r4c3} - c3n6{r4 .} ==> r1c2 ≠ 6
t-whip[7]: c8n4{r7 r8} - r8n3{c8 c9} - r6c9{n3 n2} - b9n2{r8c9 r9c7} - r3n2{c7 c2} - b1n6{r3c2 r1c3} - c8n6{r1 .} ==> r7c8 ≠ 1
t-whip[7]: c8n6{r1 r7} - c8n4{r7 r8} - r8n3{c8 c9} - r6c9{n3 n2} - c8n2{r6 r2} - c2n2{r2 r3} - b1n6{r3c2 .} ==> r1c7 ≠ 6
t-whip[6]: r7n5{c2 c6} - b5n5{r5c6 r4c4} - r4n6{c4 c3} - r1n6{c3 c8} - r7c8{n6 n4} - b7n4{r7c1 .} ==> r8c2 ≠ 5
whip[1]: b7n5{r7c2 .} ==> r7c6 ≠ 5
t-whip[7]: c6n7{r2 r8} - c6n2{r8 r9} - c7n2{r9 r3} - b3n6{r3c7 r1c8} - c3n6{r1 r4} - b4n7{r4c3 r6c1} - r3n7{c1 .} ==> r1c5 ≠ 7
hidden-pairs-in-a-column: c5{n7 n9}{r3 r8} ==> r8c5 ≠ 5, r8c5 ≠ 1
t-whip[7]: c7n2{r9 r3} - b3n6{r3c7 r1c8} - c3n6{r1 r4} - b4n7{r4c3 r6c1} - r3n7{c1 c5} - r8n7{c5 c6} - c6n2{r8 .} ==> r9c9 ≠ 2
hidden-pairs-in-a-row: r9{n2 n6}{c6 c7} ==> r9c7 ≠ 5, r9c7 ≠ 1, r9c6 ≠ 8, r9c6 ≠ 5, r9c6 ≠ 1
whip[1]: b9n5{r9c9 .} ==> r4c9 ≠ 5
t-whip[7]: c6n2{r8 r9} - c7n2{r9 r3} - b3n6{r3c7 r1c8} - c3n6{r1 r4} - b4n7{r4c3 r6c1} - r3n7{c1 c5} - r8n7{c5 .} ==> r8c6 ≠ 1
t-whip[7]: c6n2{r8 r9} - c7n2{r9 r3} - b3n6{r3c7 r1c8} - c3n6{r1 r4} - b4n7{r4c3 r6c1} - r3n7{c1 c5} - r8n7{c5 .} ==> r8c6 ≠ 5
t-whip[8]: b4n7{r6c1 r4c3} - r4n6{c3 c4} - r6n6{c4 c2} - b1n6{r3c2 r1c3} - c8n6{r1 r7} - r7c7{n6 n1} - r7c6{n1 n8} - b5n8{r5c6 .} ==> r6c1 ≠ 8
t-whip[4]: r6c5{n8 n1} - r6c1{n1 n7} - r4c3{n7 n6} - r6c2{n6 .} ==> r6c8 ≠ 8
finned-x-wing-in-rows: n8{r9 r6}{c5 c3} ==> r5c3 ≠ 8
finned-x-wing-in-rows: n8{r6 r9}{c5 c2} ==> r7c2 ≠ 8
t-whip[4]: b7n8{r9c3 r7c1} - c1n5{r7 r2} - c1n4{r2 r5} - c3n4{r5 .} ==> r2c3 ≠ 8
t-whip[6]: c3n8{r9 r1} - c3n6{r1 r4} - c3n7{r4 r2} - r3c1{n7 n1} - r1c2{n1 n5} - b7n5{r7c2 .} ==> r7c1 ≠ 8
hidden-single-in-a-block ==> r9c3 = 8
t-whip[8]: r8c6{n2 n7} - r8c5{n7 n9} - b7n9{r8c2 r7c2} - r8c3{n9 n1} - r8c4{n1 n5} - c5n5{r9 r1} - c2n5{r1 r2} - r2n2{c2 .} ==> r8c8 ≠ 2
biv-chain[4]: c8n2{r6 r2} - b1n2{r2c2 r3c2} - c2n6{r3 r6} - r6c4{n6 n3} ==> r6c8 ≠ 3
biv-chain[5]: c8n2{r6 r2} - c2n2{r2 r3} - b1n6{r3c2 r1c3} - r4c3{n6 n7} - r6c1{n7 n1} ==> r6c8 ≠ 1
biv-chain[5]: r7c7{n1 n6} - r3n6{c7 c2} - r6n6{c2 c4} - r6n3{c4 c9} - b9n3{r8c9 r8c8} ==> r8c8 ≠ 1
t-whip[9]: r9c7{n2 n6} - r7n6{c8 c6} - b8n8{r7c6 r7c4} - r7n9{c4 c2} - r8c3{n9 n1} - b8n1{r8c4 r9c5} - c5n5{r9 r1} - c2n5{r1 r2} - r2n2{c2 .} ==> r3c7 ≠ 2
singles ==> r9c7 = 2, r9c6 = 6, r8c6 = 2, r8c5 = 7, r3c5 = 9
finned-x-wing-in-rows: n7{r3 r6}{c1 c7} ==> r4c7 ≠ 7
whip[1]: b6n7{r6c8 .} ==> r2c8 ≠ 7
biv-chain[4]: r3n7{c7 c1} - b4n7{r6c1 r4c3} - c3n6{r4 r1} - b3n6{r1c8 r3c7} ==> r3c7 ≠ 1
t-whip[6]: r7c6{n8 n1} - r9c5{n1 n5} - r8c4{n5 n9} - r7c4{n9 n8} - r3c4{n8 n1} - r1c5{n1 .} ==> r2c6 ≠ 8
t-whip[6]: r7c6{n8 n1} - r9c5{n1 n5} - r8c4{n5 n9} - r7c4{n9 n8} - r3c4{n8 n1} - r1c5{n1 .} ==> r1c6 ≠ 8
t-whip[6]: b3n3{r2c7 r1c8} - c8n6{r1 r7} - r7c7{n6 n1} - r1c7{n1 n7} - b2n7{r1c6 r2c6} - c6n3{r2 .} ==> r4c7 ≠ 3
whip[1]: c7n3{r2 .} ==> r1c8 ≠ 3
hidden-triplets-in-a-block: b6{r6c8 r6c9 r4c8}{n2 n3 n7} ==> r4c8 ≠ 8, r4c8 ≠ 1
t-whip[3]: c8n1{r2 r5} - b6n8{r5c8 r4c9} - c9n9{r4 .} ==> r2c9 ≠ 1
t-whip[6]: r6n1{c2 c5} - r9c5{n1 n5} - r1c5{n5 n8} - r3c4{n8 n1} - r8c4{n1 n9} - r8c3{n9 .} ==> r5c3 ≠ 1
t-whip[6]: r6c5{n8 n1} - r9c5{n1 n5} - r1c5{n5 n8} - r3c4{n8 n1} - r8c4{n1 n9} - r7c4{n9 .} ==> r4c4 ≠ 8
t-whip[6]: r5n4{c1 c3} - b4n9{r5c3 r4c2} - r7n9{c2 c4} - b8n8{r7c4 r7c6} - b5n8{r5c6 r6c5} - b4n8{r6c2 .} ==> r5c1 ≠ 1
t-whip[4]: b4n1{r6c2 r6c1} - c1n7{r6 r3} - r3c7{n7 n6} - r7c7{n6 .} ==> r7c2 ≠ 1
t-whip[4]: b7n1{r8c3 r7c1} - r6c1{n1 n7} - r3c1{n7 n8} - r3c4{n8 .} ==> r8c4 ≠ 1
t-whip[7]: b7n1{r8c3 r7c1} - c1n5{r7 r2} - c1n4{r2 r5} - c1n8{r5 r3} - r1c2{n8 n1} - r6n1{c2 c5} - r9n1{c5 .} ==> r8c9 ≠ 1
whip[1]: r8n1{c3 .} ==> r7c1 ≠ 1
t-whip[3]: c1n1{r3 r6} - b4n7{r6c1 r4c3} - c3n6{r4 .} ==> r1c3 ≠ 1
naked-pairs-in-a-column: c3{r1 r4}{n6 n7} ==> r2c3 ≠ 7
t-whip[4]: b5n5{r5c6 r4c4} - r4n6{c4 c3} - c3n7{r4 r1} - c6n7{r1 .} ==> r2c6 ≠ 5
t-whip[4]: b2n7{r2c6 r1c6} - c3n7{r1 r4} - r4c8{n7 n3} - c6n3{r4 .} ==> r2c6 ≠ 1
t-whip[7]: b4n8{r6c2 r5c1} - r5n4{c1 c3} - r2c3{n4 n1} - r3c1{n1 n7} - r6n7{c1 c8} - c8n2{r6 r2} - c8n8{r2 .} ==> r1c2 ≠ 8
biv-chain[3]: r6n8{c2 c5} - r1n8{c5 c8} - b6n8{r5c8 r4c9} ==> r4c2 ≠ 8
finned-x-wing-in-columns: n8{c2 c5}{r6 r2} ==> r2c4 ≠ 8
hidden-triplets-in-a-column: c2{n2 n6 n8}{r2 r3 r6} ==> r6c2 ≠ 1, r2c2 ≠ 5, r2c2 ≠ 4, r2c2 ≠ 1
whip[1]: c2n4{r8 .} ==> r7c1 ≠ 4
stte

[denis_berthier]

Berthier's whips are equivalent to invalid tracks from the target,

No. Whips rely on a much more elaborated background and logical expression than tracks. And they have an inherent length.

[totuan]

Hi denis,

That's a relatively hard one, as it requires whips[7]

Nice to see you back and thanks for your detail solution.
I can translate your eliminations to AICs or Diagrams but can you explain some things for me:
1. Your solution is solver’s solution or yourself?
2. Can you optimize your solution? Then I can learn from your path - it seems some elimitions that is not necessary…

Thanks again,
ttt/totuan.

[denis_berthier]

Hi Totuan,
Thanks for your interest

I can translate your eliminations to AICs or Diagrams

I don't think whips can be translated to AICs, unless you extend the meaning of "AIC" in some unexpected direction. Whips are basically non reversible, whereas, AFAIK, the credo of AICs is reversibility. Now, if you only mean that there is an alternance of negative and positive candidates, then the left-linking and right-linking candidates can be the support for this.
On the other hand, diagrams hide the fundamental linearity of whips. t-candidates and z-candidates are not part of the whip! I've introduced the nrc-notation for some reason.

Your solution is solver’s solution or yourself?

What's the difference? Everybody here is using a solver. The solver that produced the solution is mine (SudoRules, part of CSP-Rules).

Can you optimize your solution? Then I can learn from your path - it seems some elimitions that is not necessary…

The resolution path is optimised in some very precise sense: at any point, one of the applicable rules is randomly chosen among those with the highest priority (i.e. among the simplest ones). One could call this local optimisation. But it also leads to global optimisation of the rating.
But I guess you're asking about global optimisation of the path. Then, the answer is no. But I've never seen any consistent proposal for optimising a full path.
If you only mean that some steps in my resolution path could be avoided, it's probably true, but the only way to check this would be to scan the resolution path backwards step by step.

[totuan]

Hi denis,
Thank you for quickly response.

I don't think whips can be translated to AICs, unless you extend the meaning of "AIC" in some unexpected direction.

I meant, from your eliminations I can use AICs or Diagrams to present them (maybe, my English is too bad on expression ).

I will study and learn from your path.

totuan

[Mauriès Robert]

Berthier's whips are equivalent to invalid tracks from the target,

No. Whips rely on a much more elaborated background and logical expression than tracks. And they have an inherent length.

What I wanted to say Denis, is that with a track starting from the target, we can get the same result (elimination of the target) as the whip associated with that target and with the same number of sequences (same length).
Sincerely
Robert

[yzfwsf]

Hi Denis,
I think your whips is similar to XSUDO's Truths \ links. Each Whip is equivalent to a Truth.Truths is equivalent to strong Inferences in AIC.
Because if I input the whip as the truth into XSUDO, it will get the same elimination as yours.

The first whips(7) is shown in the figure below：

RTD0000.png (29.67 KiB) Viewed 602 times

[StrmCkr]

TDP is trial and error, by looking at the out come of two massive paths:

Ive already given you examples of the path choices to leading to full solution which is blatantly obvious in any of the examples you have given to-date.
via 1 side of it
the other gives direct contradictions and [u]nested memory chains[/u ]contradictions.
ignore the solution and take the contradictions or equal truths, then apply those to the puzzle.
which is trial and error technically as its a two point guess and adrenal's thread.
the common truths of the solution and the contradiction path are hard truths in the puzzle and your trial paths can be much shorter.
but I've never seen you admit that or use the information.

if you want to show how its a:
and/not logic gate

or a set-wise mathematical construct

then be my guest and I'll believe its anything other then what i suggested above.

but
use it as you will as this is my opinion of what it is doing.

realistically you can just use the 1 side of the chain and build the forcing nets/ & or logic gate networks without ever needing the other side of the chain that lead to a solution.
strmckr

[denis_berthier]

Hi Denis,
I think your whips is similar to XSUDO's Truths \ links. Each Whip is equivalent to a Truth.Truths is equivalent to strong Inferences in AIC.
Because if I input the whip as the truth into XSUDO, it will get the same elimination as yours.

Hi yzfwsf,
I had this discussion long ago with Allan Barker. Partly transposing one pattern to another framework is often possible.

But nobody knows exactly how the XSUDO code works. What is sure is, It is incomplete wrt the idea of set covers and it's highly likely it relies on chains (as shown by Allan's logically inconsistent idea of "varying rank"). That shouldn't be a surprise, because it was from the start inspired by whips and Allan's wish to exploit the whip structure for more eliminations, as he has stated himself many times. His concept of "Truth" is just a name change for my concept of CSP-Variable. For whoever knows what reason, he made many such name changes; I've written the bilingual dictionary somewhere in one of his threads. I would say that XSUDO is a mix (in undefined proportions) of whips and set covers. In the mix, any logical foundation for each of the two approaches is lost.

One point is sure: XSUDO doesn't find all the whips or g-whips (we also had this discussion with Allan).
If you want to compare XSUDO (the part with rank 0) to some pattern I analysed in detail, it should be to Subsets or g-Subsets (see PBCS). It is easy to see that Subsets of some size are much less powerful than whips of the same size (i.e. using the same number of CSP-Variables). This is shown in great detail in PBCS for Sudoku.

And no, a CSP-Variable or a Truth is not "equivalent" to strong inferences in AICs. Of course, one can use CSP-Variables to make "strong" inferences when possible, but using them only for this is a very reductive view of my approach. A CSP-Variable is there from the start. Inferences using it eventually become possible in some resolution states.

Now, more generally, there's one thing I don't like in such graphical representation of whips as complicated nets: it totally hides the continuous and linear structure of whips. As long as you need such representations, you can't really grasp whips.

[denis_berthier]

Berthier's whips are equivalent to invalid tracks from the target,

No. Whips rely on a much more elaborated background and logical expression than tracks. And they have an inherent length.

What I wanted to say Denis, is that with a track starting from the target, we can get the same result (elimination of the target) as the whip associated with that target and with the same number of sequences (same length).

A track starting from the target will produce many more virtual eliminations and assertions that the whip (exactly as T&E does). So, what you really do is map the whip to a part of the track, namely the small part corresponding to the whip. This is useless work, as I have proven long ago that whips are a special case of braids and a braid can be mapped into a part of a T&E path (what you call a track).
The definition of a whip gives precise instructions on how to build one, much more precise than those of a track or T&E.

[Mauriès Robert]

Hi StrmCkr,
I don't quite understand what you're explaining to me, certainly for translation reasons. But I can already dispute your first sentence:

TDP is trial and error, by looking at the out come of two massive paths:

No, TDP is not a trial and error technique, or all the resolutions I see on this forum are trial and error. You're not going to make me believe, for example, that the resolutions proposed in the tarek puzzles by each other are obtained the first time without having tried something else beforehand that leads to nothing or a partial result. Even if a pattern is spotted, it would be a lucky guess if it eliminates the anti-door.
The trial-and-error in its basic definition is to take one candidate at random from the puzzle and build a chain to see what it leads to, and more intelligently to choose that candidate from a pair of two strongly related candidates to see if it does not lead to a contradiction.
What I propose with TDP is more subtle and consists of several approaches:
- the use of an anti-trace which is the equivalent of your AICs, which are for the most part chains of contradiction.
- the use of two conjugated tracks whose interactions allow eliminations and validations. If there are tests at this level, it is on the choice of tracks in the same way that you test the models. The Kraken very often to be used on this forum proceeds in the same way and you do not say that it is trial and error!
- the use of the extension (or bifurcation), what you call OR, which is found in the Totuan diagrams, and there again you don't talk about trial and error for Totuan.
- You can choose to make long or short sequences, or even to go as far as contradiction, but these choices remain free.
Anyway, I mean..,
- TDP is also very easy to use as a visual coloring method when you don't need to explain to everyone the paths you follow.
- TDP frees you from all those numerous modeling techniques that require a lot of memory and practice reserved for specialists.
I don't think you understand what I'm doing, or does it bother you, I don't know?
Sincerely
Robert