The New Sudoku Players' Forum

by **mith** » Mon Dec 14, 2020 8:17 pm

Code: Select all: +-------+-------+-------+ | . . 3 | . 1 . | . . . | | . 4 1 | . . . | . 5 . | | . . 9 | 2 6 5 | . . 3 | +-------+-------+-------+ | 5 . . | . . 8 | . 9 . | | . . . | . 7 . | . . . | | . 9 . | 3 . . | . . 2 | +-------+-------+-------+ | 3 . . | 8 4 6 | 2 . . | | . 6 . | . . . | 4 3 . | | . . . | . 3 . | 8 . . | +-------+-------+-------+ ..3.1.....41....5...9265..35....8.9.....7.....9.3....23..8462...6....43.....3.8..

by **Cenoman** » Mon Dec 14, 2020 10:38 pm

Code: Select all: +-------------------+--------------------+----------------------+ | 6 5 3 | 479 1 479 | 79 2 8 | | 2 4 1 | 79 8 3 | 679 5 67 | | 7 8 9 | 2 6 5 | 1 4 3 | +-------------------+--------------------+----------------------+ | 5 17 67 | 146 2 8 | 3 9 1467 | | 148 3 2 | 1469 7 149 | 5 168 146 | |Aa148x 9 F'678y | 3 5 B14 |F'67y E'1678y 2 | +-------------------+--------------------+----------------------+ | 3 c1-7 5 | 8 4 6 | 2 D'17z 9 | | b18 6 G'78 | 157 9 2 | 4 3 157 | | 9 2 4 | 157 3 B17 | 8 C167 1567 | +-------------------+--------------------+----------------------+

Kraken cell (148)r6c1
(1)r6c1 - r8c1 = (1)r7c2
(4)r6c1 - (4=17)r69c6 - r9c8 = [(7)r7c8 = r6c8 - (7=68)r6c37 -(8=7)r8c3]
(8)r6c1 - (8=671)r6c378 - (1=7)r7c8
=> -7 r7c2; ste

by **SpAce** » Tue Dec 15, 2020 3:10 am

Code: Select all: .--------------------.---------------.--------------------. | 6 5 3 | 479 1 479 | 79 2 8 | | 2 4 1 | 79 8 3 | 679 5 67 | | 7 8 9 | 2 6 5 | 1 4 3 | :--------------------+---------------+--------------------: | 5 b17 67 | 146 2 8 | 3 9 1467 | | 148 3 2 | 1469 7 149 | 5 168 146 | | c148 9 a8-67 | 3 5 d14 | f67 cf1678 2 | :--------------------+---------------+--------------------: | 3 bcd17 5 | 8 4 6 | 2 be17 9 | | 18 6 a78 | 157 9 2 | 4 3 157 | | 9 2 4 | 157 3 d17 | 8 e167 1567 | '--------------------'---------------'--------------------'

(8,7)r68c3 = (7,1,1)r74c2,r7c8 - r7c2|r6c18 = (7,17)r7c2,r69c6 - r79c8 = (76)r6c87 => -67 r6c3; stte

Website · by **denis_berthier** » Tue Dec 15, 2020 6:13 am

The puzzle is in tW3, but it can also be solved with slightly longer reversible chains (of max length 4):

Code: Select all: (solve-sudoku-grid +-------+-------+-------+ ! . . 3 ! . 1 . ! . . . ! ! . 4 1 ! . . . ! . 5 . ! ! . . 9 ! 2 6 5 ! . . 3 ! +-------+-------+-------+ ! 5 . . ! . . 8 ! . 9 . ! ! . . . ! . 7 . ! . . . ! ! . 9 . ! 3 . . ! . . 2 ! +-------+-------+-------+ ! 3 . . ! 8 4 6 ! 2 . . ! ! . 6 . ! . . . ! 4 3 . ! ! . . . ! . 3 . ! 8 . . ! +-------+-------+-------+ )

Hidden Text: Show

Notice that, it t-whips are active, the only bivalue-chain[4] in the resolution path, i.e.:
biv-chain[4]: r2n9{c4 c7} - c7n6{r2 r6} - b4n6{r6c3 r4c3} - b5n6{r4c4 r5c4} ==> r5c4 ≠ 9
can be replaced by a t-whip[3]:
t-whip[3]: r2n9{c4 c7} - c7n6{r2 r6} - r5n6{c9 .} ==> r5c4 ≠ 9
and all the rest is unchanged.

I think this is a very good example of how to use a t-candidate instead of looking for more bivalue-cells.

by **SpAce** » Tue Dec 15, 2020 8:36 am

Speaking of t-whips, here's a single-stepper with one:

Code: Select all: .-------------------.---------------.-------------------. | 6 5 3 | 479 1 479 | 79 2 8 | | 2 4 1 | 79 8 3 | 679 5 67 | | 7 8 9 | 2 6 5 | 1 4 3 | :-------------------+---------------+-------------------: | 5 b17 67 | 146 2 8 | 3 9 1467 | | 148 3 2 | 1469 7 149 | 5 168 146 | | e148# 9 k678# | 3 5 f14 | j67 ei1678 2 | :-------------------+---------------+-------------------: | 3 ac1-7 5 | 8 4 6 | 2 dh17# 9 | | 18 6 l78 | 157 9 2 | 4 3 157 | | 9 2 4 | 157 3 g17 | 8 h167 1567 | '-------------------'---------------'-------------------'

t-whip[8t3]: n1c2{r7 r4} - n1r7{c2 c8} - n1r6{c8 c6} - r9c6{n1 n7} - n7c8{r9 r6} - r6c7{n7 n6} - r6c3{n6 n8} - r8c3{n8 n7} => -n7r7c2; stte

--
Denis, I tried to see what SudoRules would suggest for that elimination, but couldn't get anything out of the 'try-to-eliminate-candidates' function. It was the first time I tried it. This is what I wrote, as per the instructions here:

(try-to-eliminate-candidates ".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." 772)

All it did was print a few simple steps (including three biv-chains, if enabled) and then report it couldn't solve it. I tried other candidates and various combos of enabled techniques too, but the same result. Did I use the right syntax for the candidate?)

Website · by **denis_berthier** » Tue Dec 15, 2020 9:22 am

SpAce wrote:
(try-to-eliminate-candidates ".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." 772)

All it did was print a few simple steps (including three biv-chains, if enabled) and then report it couldn't solve it. I tried other candidates and various combos of enabled techniques too, but the same result. Did I use the right syntax for the candidate?)

As I've written, "try-to-eliminate-candidates" doesn't work if t-whips are active.

In addition,:
- c2n1{r7 r4} - r7n1{c2 c8} is not the start for a whip, let alone a t-whip
- what rules did you use to reach your PM?

Website · by **denis_berthier** » Tue Dec 15, 2020 12:05 pm

I re-tested function try-to-eliminate-candidates on another puzzle, Magictour #77 (chosen because it starts with long whips):

Code: Select all: 7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5

Activating only whips, the start of the normal resolution path is :

Code: Select all: *********************************************************************************************** *** SudoRules 20.1.s based on CSP-Rules 2.1.s, config = W *** Using CLIPS 6.32-r773 *********************************************************************************************** hidden-single-in-a-row ==> r9c8 = 3 215 candidates, 1351 csp-links and 1351 links. Density = 1.46815909584873 whip[1]: b9n4{r7c8 .} ==> r7c3 ≠ 4, r7c2 ≠ 4, r7c1 ≠ 4 whip[9]: c4n6{r2 r9} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c5n6{r3 r4} - r1n6{c5 c3} - b1n8{r1c3 r1c2} - b1n1{r1c2 r2c1} - b1n9{r2c1 .} ==> r2c6 ≠ 6 whip[11]: c3n4{r5 r2} - c6n4{r2 r4} - b5n6{r4c6 r4c5} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c4n1{r3 r5} - b5n8{r5c4 r6c4} - b5n9{r6c4 r6c5} - r6c2{n9 n5} - r3c2{n5 .} ==> r5c1 ≠ 4 whip[12]: r9c5{n8 n6} - r7c5{n6 n5} - r3c5{n5 n1} - r4c5{n1 n9} - r6c4{n9 n4} - r5c6{n4 n3} - b4n3{r5c1 r6c1} - c1n9{r6 r2} - r2c4{n9 n6} - r1n6{c4 c3} - b1n8{r1c3 r1c2} - b1n1{r1c2 .} ==> r6c5 ≠ 8 whip[15]: r5c6{n4 n3} - b4n3{r5c1 r6c1} - r6c5{n3 n9} - r4c6{n9 n6} - c6n4{r4 r2} - c3n4{r2 r4} - r6n4{c2 c8} - r6n2{c8 c7} - r4n2{c8 c1} - r9n2{c1 c4} - c6n2{r7 r1} - b2n3{r1c6 r1c5} - r1c9{n3 n1} - r5n1{c9 c7} - b6n5{r5c7 .} ==> r5c4 ≠ 4 whip[15]: b3n6{r2c7 r3c7} - b3n7{r3c7 r3c8} - r3n2{c8 c4} - c4n4{r3 r6} - r5c6{n4 n3} - r6c5{n3 n9} - r4c6{n9 n6} - r1n6{c6 c3} - b1n8{r1c3 r1c2} - r6c2{n8 n5} - c8n5{r6 r1} - r1c6{n5 n9} - r1c4{n9 n1} - r5c4{n1 n8} - r5c1{n8 .} ==> r2c4 ≠ 6

If instead of solve, I use (try-to-eliminate-candidates "7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5" 626), it gives as expected:

Code: Select all: hidden-single-in-a-row ==> r9c8 = 3 215 candidates, 1351 csp-links and 1351 links. Density = 5.87% whip[1]: r9n4{c2 .} ==> r7c3 ≠ 4, r7c1 ≠ 4, r7c2 ≠ 4 whip[9]: c4n6{r3 r9} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c5n6{r3 r4} - r1n6{c5 c3} - r1n8{c3 c2} - b1n9{r1c2 r2c1} - b1n1{r2c1 .} ==> r2c6 ≠ 6

It is not a priori sure that (try-to-eliminate-candidates "7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5" 451) would give the second whip: it might depend on the elimination done by the first whip[9]. But it gives:

Code: Select all: hidden-single-in-a-row ==> r9c8 = 3 215 candidates, 1351 csp-links and 1351 links. Density = 5.87% whip[1]: r9n4{c2 .} ==> r7c3 ≠ 4, r7c1 ≠ 4, r7c2 ≠ 4 whip[11]: c3n4{r5 r2} - c6n4{r2 r4} - r4n6{c6 c5} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c4n1{r3 r5} - c4n8{r5 r6} - b5n9{r6c4 r6c5} - r6c2{n9 n5} - r3c2{n5 .} ==> r5c1 ≠ 4

However, for the third whip of the normal resolution path, (try-to-eliminate-candidates "7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5" 454) doesn't give the whip[15] because it depends on the previous eliminations.

Notice that try-to-eliminate starts afresh from the given puzzle (not from the current PM).

However, you can try several candidates:
(try-to-eliminate-candidates "7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5" 626 451 865) gives as expected:

Code: Select all: hidden-single-in-a-row ==> r9c8 = 3 215 candidates, 1351 csp-links and 1351 links. Density = 5.87% whip[1]: r9n4{c2 .} ==> r7c3 ≠ 4, r7c1 ≠ 4, r7c2 ≠ 4 whip[9]: c4n6{r3 r9} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c5n6{r3 r4} - r1n6{c5 c3} - r1n8{c3 c2} - b1n9{r1c2 r2c1} - b1n1{r2c1 .} ==> r2c6 ≠ 6 whip[11]: c3n4{r5 r2} - c6n4{r2 r4} - r4n6{c6 c5} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c4n1{r3 r5} - c4n8{r5 r6} - b5n9{r6c4 r6c5} - r6c2{n9 n5} - r3c2{n5 .} ==> r5c1 ≠ 4 whip[12]: r9c5{n8 n6} - r7c5{n6 n5} - r3c5{n5 n1} - r4c5{n1 n9} - r6c4{n9 n4} - r5c6{n4 n3} - r6n3{c5 c1} - c1n9{r6 r2} - r2c4{n9 n6} - r1n6{c6 c3} - r1n8{c3 c2} - b1n1{r1c2 .} ==> r6c5 ≠ 8

Conclusion: I can't see any problem with this function.

by **SpAce** » Tue Dec 15, 2020 12:24 pm

denis_berthier wrote:As I've written, "try-to-eliminate-candidates" doesn't work if t-whips are active.

Ok, though I don't know where you've written it. I understood that the function only works for whips and braids, but didn't realize that everything else had to be turned off. Well, I did now, but still no luck.

In addition,:
- c2n1{r7 r4} - r7n1{c2 c8} is not the start for a whip, let alone a t-whip

Why not? As a matrix it sure looks like a t-whip to me:

8x8 TM: Show

No gaps -> not a braid; 3 t-candidates and no z-candidates -> t-whip. What is it if not that? Here's the same written a bit differently:

8x8 TM: Show

Would you accept that as a braid, with the 1r7c2 as a z-candidate?

- what rules did you use to reach your PM?

The normal rules of this section -- singles, intersections, subsets (no fish).

I re-tested the function on another puzzle : Magictour #77:
...
If instead of solve, I use (try-to-eliminate-candidates "7.....4...2..7..8...3..8..9...5..3...6..2..9...1..7..6...3..9...3..4..6...9..1..5" 626), it gives as expected:
...
whip[9]: c4n6{r3 r9} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - c5n6{r3 r4} - r1n6{c5 c3} - r1n8{c3 c2} - b1n9{r1c2 r2c1} - b1n1{r2c1 .} ==> r2c6 ≠ 6

Not for me. When I turned everything but whips off, I got nothing. However, before that I forgot to turn forcing braids off and got:

SudoRules wrote:whip[1]: r9n4{c2 .} ==> r7c3 ≠ 4, r7c1 ≠ 4, r7c2 ≠ 4
braid[8]: r1n6{c6 c3} - c4n6{r1 r9} - r9c5{n6 n8} - r7c5{n8 n5} - r3c5{n5 n1} - r1n8{c3 c2} - b1n1{r1c2 r2c1} - b1n9{r2c1 .} ==> r2c6 ≠ 6

I almost missed it because it took forever to compute and I thought Clips had hanged. I guess I have something wrong with the config file since the whip didn't show up. Unfortunately I don't think I have more patience to figure it out, as the debug cycle takes too long with this feature.

Website · by **denis_berthier** » Tue Dec 15, 2020 1:24 pm

SpAce wrote:
denis_berthier wrote:- what rules did you use to reach your PM?

The normal rules of this section -- singles, intersections, subsets (no fish).

There are no "normal rules of this section" or no "normal rules" of anywhere.
If your PM includes eliminations of candidates that can't be eliminated by the restricted whips, it's no wonder that SudoRules don't find them.

SpAce wrote:the debug cycle takes too long with this feature

There's no debug cycle here. It's you learning it the hard way for not following the stated restrictions.

I remind you also that this function is experimental and not at all the normal way of using CSP-Rules.

by **Mauriès Robert** » Tue Dec 15, 2020 2:53 pm

Hi all,
Two-step resolution with the same anti-track :
P'(7r7c8) : (-7r7c8)=>(1r7c7->1r4c2)->1r6c6->7r9c7->... => -7r9c89.
P'(7r7c8) : (-7r7c8)=>(7r6c8->6r6c7)->8r6c3->7r8c3->... => -7r8c9 =>r7c8=7, stte.

puzzle: Show

Obviously one can solve in a single step with this anti-track by considering the 8 elements composing the construction sequence:
P'(7r7c8) : (-7r7c8)=>(1r7c7->1r4c2)->1r6c6->7r9c7->(7r6c8->6r6c7)->8r6c3->7r8c3->... => -7b9p689.
Robert

by **SpAce** » Tue Dec 15, 2020 3:37 pm

denis_berthier wrote:
SpAce wrote:the debug cycle takes too long with this feature

There's no debug cycle here. It's you learning it the hard way for not following the stated restrictions.

<sigh> You wouldn't have to state the obvious so often if you tried to avoid misinterpreting my words all the time. I was talking about my own debug cycle with my own config file. I own any mistakes in that, though not the reasons why they're easy to make.

The New Sudoku Players' Forum

More Pi 8 (SE 7.7)

More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)

Re: More Pi 8 (SE 7.7)