YZF_Sudoku

Programs which generate, solve, and analyze Sudoku puzzles

Re: YZF_Sudoku

Postby denis_berthier » Sun Dec 13, 2020 4:27 am

ghfick wrote:Hi Denis,
To run YZF_Sudoku on either a Mac or with Linux, you need to install Wine. [winehq.org]

Wine is 32-bit and doesn't run on MacOS Catalina or above - that's 1+ year old.
As far as I know, there's no plan to make Wine 64-bit.


ghfick wrote:To my knowledge, YZF_Sudoku does not currently have many of the techniques in SudoRules. If you are willing, it would be very useful of have some form of a thesaurus to detail the corresponding namings in SudoRules into Sudopedia namings. I do recognize that in some cases there will not be equivalent namings. I do think that SudoRules' z-chains are distinct from anything in YZF_Sudoku [I could be wrong here]. I think that SudoRules' whips [partially] overlap with most of the techniques in YZF_Sudoku [again, I may be mistaken

Except some techniques (Subsets, sk-loops) clearly identified as "classical", most of SudoRules techniques are original - there's no point of giving them other names in Sudopedia, except maybe for hiding plagiarism. I have no objection if YZF_Sudoku implements whips but then call them whips. It wouldn't be the first alternative implementation of whips.
BTW, I'm not aware of any persisting existence of Sudopedia.
denis_berthier
2010 Supporter
 
Posts: 3967
Joined: 19 June 2007
Location: Paris

Re: YZF_Sudoku

Postby denis_berthier » Sun Dec 13, 2020 5:05 am

SpAce wrote:
ghfick wrote:So far, I have not been able to run SudoRules on either Mac or Linux. My understanding of the inner workings of Mac and Linux holds me back. A very detailed, step-by-step set of instructions might help.

That’s weird. SudoRules worked out of the box on my Mac. I did compile Clips, though, as instructed. I didn’t even try the bundled executable, so I don’t know if that would have worked or not.

Apart from not launching because of security reasons on Catalina and higher, I see no reason why the provided executable wouldn't work. As you have a Mac, could you try it, to make sure that it does?
denis_berthier
2010 Supporter
 
Posts: 3967
Joined: 19 June 2007
Location: Paris

Re: YZF_Sudoku

Postby StrmCkr » Sun Dec 13, 2020 5:46 am

Actually Denis you have posted mutiple times on my threads where discontinuous nice loops are z, t chains and a few others are the same thing as already named techniquesso there is much overlaping in the names you use vs the more common names.


Or x2y2 chains ie belts or whips /virus patterns land on skloops


Anyway my 2 cents
Strmckr
Andrew steward owns sudopedia and has a limited version of stuff on it.
the forms maintain.a copy of the orginal not updated version befor andrew bought it for his website scanraid renaming.
So it has persisted in a sense.
Some do, some teach, the rest look it up.
stormdoku
User avatar
StrmCkr
 
Posts: 1417
Joined: 05 September 2006

Re: YZF_Sudoku

Postby SpAce » Sun Dec 13, 2020 5:51 am

denis_berthier wrote:Apart from not launching because of security reasons on Catalina and higher, I see no reason why the provided executable wouldn't work. As you have a Mac, could you try it, to make sure that it does?

I just downloaded the new version and tried. I also tried (presumably the same) executable that came with the previous version. In both cases:

macOS Catalina Version 10.15.7 wrote:“clips” cannot be opened because the developer cannot be verified.

It did start after overriding the lock, so I expect it would work normally.

Btw, there's a slight difference in size between the bundled and my compiled version: 918K vs 938K. Different sources or compilers/options? (I used the makefile that came with the sources.)
User avatar
SpAce
 
Posts: 2671
Joined: 22 May 2017

Re: YZF_Sudoku

Postby denis_berthier » Sun Dec 13, 2020 6:49 am

SpAce wrote:
denis_berthier wrote:Apart from not launching because of security reasons on Catalina and higher, I see no reason why the provided executable wouldn't work. As you have a Mac, could you try it, to make sure that it does?

I just downloaded the new version and tried. I also tried (presumably the same) executable that came with the previous version. In both cases:

macOS Catalina Version 10.15.7 wrote:“clips” cannot be opened because the developer cannot be verified.

It did start after overriding the lock, so I expect it would work normally.

Btw, there's a slight difference in size between the bundled and my compiled version: 918K vs 938K. Different sources or compilers/options? (I used the makefile that came with the sources.)


I used the makefile also. I can't explain the difference in size. Except that the version provided with CSP-Rules may be a slightly older version of Clips core -(bit don't worry, the modifications are irrelevant to CSP-Rules).

TO ALL: beware that, if need be (i.e. if you have any problem with the executables delivered with CSP-Rules) CLIPS must be compiled withe the provided Makefile, i.e. with "make". Using only "gcc" without the proper options would make it significantly slower.
denis_berthier
2010 Supporter
 
Posts: 3967
Joined: 19 June 2007
Location: Paris

Re: YZF_Sudoku

Postby yzfwsf » Sun Dec 13, 2020 2:49 pm

ghfick wrote:To my knowledge, YZF_Sudoku does not currently have many of the techniques in SudoRules.

My superficial understanding is that the chain of denis_berthier is a kind of AIC that contains grouped nodes and ALS nodes and has branches, but he only lists the strong links, and the weak links are hidden.
yzfwsf
 
Posts: 845
Joined: 16 April 2019

Re: YZF_Sudoku

Postby denis_berthier » Sun Dec 13, 2020 3:26 pm

yzfwsf wrote:
ghfick wrote:To my knowledge, YZF_Sudoku does not currently have many of the techniques in SudoRules.

My superficial understanding is that the chain of denis_berthier is a kind of AIC that contains grouped nodes and ALS nodes and has branches, but he only lists the strong links, and the weak links are hidden.

Totally wrong
denis_berthier
2010 Supporter
 
Posts: 3967
Joined: 19 June 2007
Location: Paris

Re: YZF_Sudoku

Postby ghfick » Sun Dec 13, 2020 4:31 pm

YZF_Sudoku now has Symmetrical Placement including Sticks Symmetry. Very impressive!
I see you have set the default level as 'Medium' [Green] in All Possible Steps. I think you are suggesting that human solvers can 'easily' find these moves. I am now motivated to try them. Maybe you have some 'tips' as to how to do the manual search?
ghfick
 
Posts: 232
Joined: 06 April 2016
Location: Calgary, Alberta, Canada youtube.com/@gordonfick

Re: YZF_Sudoku

Postby yzfwsf » Sun Dec 13, 2020 4:38 pm

denis_berthier wrote:Totally wrong

Ok,I really don’t understand your method, it’s just that after I entered your problem-solving steps into xsudo, I understood it myself. I have tried to read your publications, but I am not familiar with English, and there are many mathematical symbols in your publications , It's too difficult for me.
yzfwsf
 
Posts: 845
Joined: 16 April 2019

Re: YZF_Sudoku

Postby yzfwsf » Sun Dec 13, 2020 4:41 pm

ghfick wrote:YZF_Sudoku now has Symmetrical Placement including Sticks Symmetry. Very impressive!
I see you have set the default level as 'Medium' [Green] in All Possible Steps. I think you are suggesting that human solvers can 'easily' find these moves. I am now motivated to try them. Maybe you have some 'tips' as to how to do the manual search?

I can only suggest that you modify the GSP difficulty level in the options, because I am a poor manual solver.
yzfwsf
 
Posts: 845
Joined: 16 April 2019

Re: YZF_Sudoku

Postby ghfick » Sun Dec 13, 2020 4:57 pm

Thanks so much to Denis for making his SudoRules available. After several 'poor' attempts at installation, I was successful. I was able to use the given Clips executable. My Mac is a few years old and currently runs OS 10.11.3.
Denis' instructions are clear and detailed. You do have to follow the instructions carefully. I must have muddled something in my earlier attempts and then initially gave up too soon.
My plan is to try some puzzles with both SudoRules and YZF_Sudoku. The goal being to first determine common techniques. Some of the common techniques have the same names so that part is easy.
Then, I plan to consider those techniques that do not [easily, at least] compare. The reality is that it may not be straightforward to develop 'similar' solution paths / resolution paths and so technique comparison may not be available.
It is clear that, with some puzzles with 'high' SE ratings, YZF_Sudoku has an 'easy' path while SudoRules does not appear to have an easy path. Sometimes, SudoRules has a path that is 'easy' while YZF_Sudoku does not. A while back, I came across a puzzle in which the SudoRules' path used a number of 'lengthy' z-chains and then the path was 'easy' while YZF_Sudoku needed 'Insane' steps in its paths.
ghfick
 
Posts: 232
Joined: 06 April 2016
Location: Calgary, Alberta, Canada youtube.com/@gordonfick

Re: YZF_Sudoku

Postby ghfick » Sun Dec 13, 2020 5:24 pm

yzfwsf wrote:I can only suggest that you modify the GSP difficulty level in the options, because I am a poor manual solver.

Using the defaults, the displayed path shows GSP before an X Wing [for example]. This may be the right place. I need to study more. I recall that Andrew Stuart places GSP very early in his 'Strategy' order.
ghfick
 
Posts: 232
Joined: 06 April 2016
Location: Calgary, Alberta, Canada youtube.com/@gordonfick

Re: YZF_Sudoku

Postby SpAce » Tue Dec 15, 2020 11:00 am

ghfick wrote:Sometimes, SudoRules has a path that is 'easy' while YZF_Sudoku does not. A while back, I came across a puzzle in which the SudoRules' path used a number of 'lengthy' z-chains and then the path was 'easy' while YZF_Sudoku needed 'Insane' steps in its paths.

What made you think those z-chains were easy? Did you try to find them manually without knowing the targets? I'm pretty sure some of them could be quite accurately described as "insane" steps, too. In general, z-chains are far from easy to find for a manual solver, even though Denis' notation makes them look very simple. In fact, they look deceptively simple even as matrices.

From a practical solving point of view that simplicity is an illusion. Without assumptive methods, z-chains are much harder to find than t-whips or even t-braids. In any complex case it's practically impossible without trying the target. I guess Denis has his reasons for putting z-chains before t-patterns in his hierarchy, but it does not reflect their relative difficulties for a manual solver who's not using T&E. If someone claims otherwise, the burden of proof is on them.

Some z-chains from this puzzle, and the simplest Hodoku-techniques for the same eliminations:

Hidden Text: Show
***********************************************************************************************
*** SudoRules 20.1.s based on CSP-Rules 2.1.s, config = Z+S
***********************************************************************************************
.3..7..2.8..4.9..5..9...8...9..8..4.7..2.6..8.8..5..3...1...6..9..7.3..2.2..1..9.
28 givens, 178 candidates
178 candidates, 956 csp-links and 956 links. Density = 6.07%
starting non trivial part of solution
whip[1]: c8n6{r3 .} ==> r3c9 ≠ 6, r1c9 ≠ 6
swordfish-in-columns: n6{c2 c5 c8}{r3 r8 r2} ==> r8c3 ≠ 6, r3c4 ≠ 6, r3c1 ≠ 6, r2c3 ≠ 6
swordfish-in-rows: n1{r2 r5 r8}{c8 c2 c7} ==> r6c7 ≠ 1, r4c7 ≠ 1, r3c8 ≠ 1, r3c2 ≠ 1, r1c7 ≠ 1
biv-chain[3]: r7n2{c6 c5} - c5n9{r7 r5} - b5n4{r5c5 r6c6} ==> r7c6 ≠ 4
biv-chain[3]: c8n8{r7 r8} - b7n8{r8c3 r9c3} - b7n7{r9c3 r7c2} ==> r7c8 ≠ 7
whip[1]: c8n7{r3 .} ==> r2c7 ≠ 7, r3c9 ≠ 7
hidden-pairs-in-a-block: b3{r2c8 r3c8}{n6 n7} ==> r2c8 ≠ 1
1. z-chain[4]: c6n2{r7 r3} - r2n2{c5 c3} - c3n7{r2 r9} - r9n8{c3 .} ==> r7c6 ≠ 8
2. z-chain[6]: r4c4{n1 n3} - r3c4{n3 n5} - r3c6{n5 n2} - r7n2{c6 c5} - c5n9{r7 r5} - r6c4{n9 .} ==> r1c4 ≠ 1
3. z-chain[11]: r7c6{n5 n2} - r3c6{n2 n1} - r3c4{n1 n3} - r2n3{c5 c7} - r2n1{c7 c2} - r1n1{c1 c9} - c9n9{r1 r6} - c4n9{r6 r7} - c4n5{r7 r9} - c4n6{r9 r1} - r1n8{c4 .} ==> r1c6 ≠ 5
4. z-chain[12]: r7c8{n5 n8} - r8n8{c8 c3} - r8n5{c3 c2} - r8n6{c2 c5} - r9c4{n6 n8} - r9c6{n8 n4} - b5n4{r6c6 r5c5} - r5n9{c5 c7} - r1c7{n9 n4} - r8c7{n4 n1} - c8n1{r8 r5} - c8n5{r5 .} ==> r9c7 ≠ 5

naked-triplets-in-a-block: b9{r7c9 r9c7 r9c9}{n3 n4 n7} ==> r8c7 ≠ 4
5. z-chain[7]: c5n9{r7 r5} - r5n3{c5 c3} - r5n4{c3 c2} - r8n4{c2 c3} - r8n8{c3 c8} - r7n8{c8 c4} - r7n9{c4 .} ==> r7c5 ≠ 4
hidden-triplets-in-a-row: r7{n3 n4 n7}{c9 c1 c2} ==> r7c2 ≠ 5, r7c1 ≠ 5
6. z-chain[5]: b9n5{r8c8 r7c8} - r5c8{n5 n1} - r5c2{n1 n4} - c5n4{r5 r8} - r8n6{c5 .} ==> r8c2 ≠ 5
naked-pairs-in-a-row: r8{c2 c5}{n4 n6} ==> r8c3 ≠ 4
biv-chain[4]: c8n6{r2 r3} - r3n7{c8 c2} - c2n5{r3 r5} - c2n1{r5 r2} ==> r2c2 ≠ 6
7. z-chain[5]: r2n1{c2 c7} - r2n3{c7 c5} - r3c4{n3 n5} - c2n5{r3 r5} - c2n1{r5 .} ==> r3c1 ≠ 1
8. z-chain[5]: r5n4{c3 c5} - r5n9{c5 c7} - r1c7{n9 n4} - c3n4{r1 r9} - c6n4{r9 .} ==> r6c1 ≠ 4
9. z-chain[4]: c3n7{r9 r2} - b1n2{r2c3 r3c1} - c1n4{r3 r1} - c7n4{r1 .} ==> r9c3 ≠ 4
10. z-chain[5]: r7c2{n4 n7} - c3n7{r9 r2} - b1n2{r2c3 r3c1} - r3n4{c1 c9} - r7n4{c9 .} ==> r8c2 ≠ 4

naked-single ==> r8c2 = 6
naked-single ==> r8c5 = 4
hidden-single-in-a-block ==> r6c6 = 4
hidden-single-in-a-block ==> r4c6 = 7
hidden-single-in-a-block ==> r9c4 = 6
whip[1]: b5n1{r6c4 .} ==> r3c4 ≠ 1
biv-chain[4]: r2n1{c2 c7} - r2n3{c7 c5} - r3c4{n3 n5} - c2n5{r3 r5} ==> r5c2 ≠ 1
hidden-single-in-a-column ==> r2c2 = 1
naked-single ==> r2c7 = 3
whip[1]: b3n1{r3c9 .} ==> r4c9 ≠ 1, r6c9 ≠ 1
naked-single ==> r4c9 = 6
biv-chain[4]: c2n5{r5 r3} - r3c4{n5 n3} - c5n3{r3 r5} - r5n9{c5 c7} ==> r5c7 ≠ 5
biv-chain[4]: c3n4{r5 r1} - r1c7{n4 n9} - r5c7{n9 n1} - r5c8{n1 n5} ==> r5c3 ≠ 5
biv-chain[4]: r4n1{c1 c4} - b5n3{r4c4 r5c5} - r5c3{n3 n4} - r5c2{n4 n5} ==> r4c1 ≠ 5
biv-chain[4]: r3c4{n5 n3} - c5n3{r3 r5} - r5c3{n3 n4} - r5c2{n4 n5} ==> r3c2 ≠ 5
;stte
GRID 0 SOLVED. rating-type = Z+S, MOST COMPLEX RULE TRIED = z-chain[12]
***********************************************************************************************

Code: Select all
1.  z-chain[4]  -> DNL
2.  z-chain[6]  -> ALS-DNL or ALS-XY-Wing
3.  z-chain[11] -> Forcing Net
4.  z-chain[12] -> Forcing Net
5.  z-chain[7]  -> Forcing Net
6.  z-chain[5]  -> Forcing Net
7.  z-chain[5]  -> ALS-XZ
8.  z-chain[5]  -> Finned Swordfish or ALS-DNL
9.  z-chain[4]  -> ALS-DNL
10. z-chain[5]  -> ALS-AIC

Of those ten z-chains only the very first one is trivial to find without assuming the target. Five others are doable, but four map to Forcing Nets -- good luck finding them without T&E. Sure, even they are relatively easy if you know the target in advance, but how would you?
User avatar
SpAce
 
Posts: 2671
Joined: 22 May 2017

Re: YZF_Sudoku

Postby denis_berthier » Tue Dec 15, 2020 11:50 am

SpAce wrote:Of those ten z-chains only the very first one is trivial to find without assuming the target. Five others are doable, but four map to Forcing Nets -- good luck finding them without T&E. Sure, even they are relatively easy if you know the target in advance, but how would you?

The most absurd argument I've seen about z-chains: they map to forcing nets. So, because some other solver is unable to find z-chains other than via hyper-complex forcing nets, z-chains must the as complex as forcing-nets????
I think it's time for you to take (and give us) some rest.
denis_berthier
2010 Supporter
 
Posts: 3967
Joined: 19 June 2007
Location: Paris

Re: YZF_Sudoku

Postby SpAce » Tue Dec 15, 2020 3:06 pm

denis_berthier wrote:The most absurd argument I've seen about z-chains: they map to forcing nets. So, because some other solver is unable to find z-chains other than via hyper-complex forcing nets, z-chains must the as complex as forcing-nets????

Sorry to inform you, but z-chains are nothing but forcing nets. They're just a specific type of them that you have named, and one that happens to be hard to find without T&E despite the apparent simplicity. I wouldn't expect you to be an expert on such mappings, though, because you obviously don't care much about anything else but your own system.

And no, my judgments aren't based on what Hodoku finds. They're based on what I and other decent manual solvers can find and how. Fact is, if neither I nor Robert can see a practical way to find complex z-chains manually without trying specific targets, there probably isn't any. On the other hand, both of us can find very complex t-whips/braids and forcing t-whips/braids without fixing a target.

You can freely keep your theoretical hierarchy, but don't be so certain that it necessarily matches the practical realities of manual solving. In that realm you'd be wise to listen to those who have a proven track record. If you do, I'm not aware of it.

I think it's time for you to take (and give us) some rest.

Or maybe it's time for you to stop and think. If you want to be arrogant about matters of manual solving, you should find weaker opponents or otherwise show us the goods. Merely claiming something without evidence, and disparaging people who are better experts in their own niches, is not an effective strategy, no matter how many times you repeat it. I'm pretty sure most people can see through it at this point.
User avatar
SpAce
 
Posts: 2671
Joined: 22 May 2017

PreviousNext

Return to Software