The New Sudoku Players' Forum

[ronk]

When I run SE 1.2.1 against the two puzzles, the first elimination is r9c3<>1 and r9c3<>9, respectively. However, if you examine the explanation behind each, then you discover that different logic is used to derive each elimination.

It appears that the set of strong inferences is the same, after allowing for the swap of candidate values <1> and <9>, of course.

One must remember that this is a net, and that the description order of the chain segments of the net might differ. In this case, steps (9) through (19) say the same thing, just in a different order.

[daj95376]

My apologies. I was on a different page (figuratively) than you and Iksudoku.

Now that I'm (hopefully) working with the same information as you and IK, I don't agree with IK's assessment of the situation.

I agree that the first seven steps of Chain 1 is (essentially) the same in both explanations. When we get to step 8, the explanations differ.

(8) If R8C6 does not contain the value 3 and R8C1 does not contain the value 3 (6), then R8C2 must contain the value 3 (only remaining possible position in the row)

(8) If R8C6 does not contain the value 3, then R9C5 must contain the value 3 (only remaining possible position in the block)

Everything else being equal with respect to <3> in these two puzzles, I contend that both explanations must be true at this step ... not as IK surmised. I also doubt if "being a net" explains the selection of one over the other. If I'm correct, then we're lacking an explanation for the divergence at this step.

Regards, Danny

[lksudoku]

If you take inference 8 in the 9.2 chain

Code: Select all

8) If R8C6 does not contain the value 3 and R8C1 does not contain the value 3 (6), then R8C2 must contain the value 3 (only remaining possible position in the row)


And inferemce 11 in the 9.3 chain

Code: Select all

(11) If R9C3 does not contain the value 3, R9C2 does not contain the value 3 (9) and R8C1 does not contain the value 3 (6), then R8C2 must contain the value 3 (only remaining possible position in the block)


You can see that the conclusion R8C2 must contain the value 3 is found in inference 11 for the 9.3 case and inference 8 in the 9.2 case

This is the cause for the rate difference for the same elimination

Note that step 11 in 9.3 chain could be the same as step 8 in 9.2 chain since it also use the fact "R8C1 does not contain the value 3" and it is based indirectly on the inference "R8C6 does not contain the value 3", however the block inferences are searched before the row inferences so the longer chain step 11 is found for the 9.3 rate chain

If the inference "R8C1 does not contain the value 3" was found before checking the "R8C6 does not contain the value 3" inference, the 9.2 inference 8 would be found at that time

[daj95376]

Iksudoku,

I don't care about comparing step 8 in the first puzzle with step 11 of the second puzzle ... or any of the other spaghetti you added to the explanation.

I want to know if I'm correct that both explanations at step 8 are true at the same time. If so, then I would like to know why they were selected differently.

Regards, Danny

[lksudoku]

I want to know if I'm correct that both explanations at step 8 are true at the same time.

No, not both explanations are true at step 8

What you see at step (6) of both chains

Code: Select all

(6) If R8C1 contains the value 1, then R8C1 cannot contain the value 3 (the cell can contain only one value)


Is processed before step 8 in the search for chain 9.2

In chain 9.3, step 8 with "R8C6 does not contain the value 3" is processed before the step "R8C1 cannot contain the value 3" and so, the deduction "R8C2 must contain the value 3" cannot be found at the search processing in that time

The order of the steps in the displayed solution is not the order of the search

[daj95376]

[Withdrawn: I initially missed IK's last statement.]

[lksudoku]

No, not both explanations are true at step 8

Okay, then where did I make a mistake while creating these grids?

Code: Select all

 puzzle 1 steps (1-7):   (*) Singles in [band 3] for <3>
 (6) If R8C1 contains the value 1, then R8C1 cannot contain the value 3
 (the cell can contain only one value)
 *-----------------------------------------------------------------------------*
 | 5       349     379     | 14678   4679    1468    | 2       1378    34678   |
 | 29      1       79      | 3       245679  4568    | 4568    578     45678   |
 | 8       234     6       | 147     2457    145     | 145     1357    9       |
 |-------------------------+-------------------------+-------------------------|
 | 7       3569    1359    | 2       3456    34568   | 14589   13589   3458    |
 | 2369    8       359     | 467     1       3456    | 459     23579   23457   |
 | 23      235     4       | 78      357     9       | 158     6       23578   |
 |-------------------------+-------------------------+-------------------------|
 | 69      7       89      | 5       46      146     | 3       289     268     |
 | 1      *356     2       | 9       8       7       | 56      4       56      |
 | 4       3569    3589    | 6      *36      2       | 7       589     1       |
 *-----------------------------------------------------------------------------*


Code: Select all

 puzzle 2 steps (1-7):   (*) Singles in [band 3] for <3>
 (6) If R8C1 contains the value 9, then R8C1 cannot contain the value 3
 (the cell can contain only one value)
 *-----------------------------------------------------------------------------*
 | 5       134     137     | 46789   1467    4689    | 2       3789    34678   |
 | 12      9       17      | 3       124567  4568    | 4568    578     45678   |
 | 8       234     6       | 479     2457    459     | 459     3579    1       |
 |-------------------------+-------------------------+-------------------------|
 | 7       1356    1359    | 2       3456    34568   | 14589   13589   3458    |
 | 1236    8       135     | 467     9       3456    | 145     12357   23457   |
 | 23      235     4       | 78      357     1       | 589     6       23578   |
 |-------------------------+-------------------------+-------------------------|
 | 16      7       18      | 5       46      469     | 3       128     268     |
 | 9      *356     2       | 1       8       7       | 56      4       56      |
 | 4       1356    1358    | 6      *36      2       | 7       158     9       |
 *-----------------------------------------------------------------------------*


Regards, Danny

Your mistake is that R8C1 contains the candidate 3 in the 9.3 search when that search reaches the "R8C6 does not contain the value 3" step while R8C1 does not contain the candidate 3 at that time in the 9.2 chain

Your puzzle description does not describe the state of the puzzle when the "R8C6 does not contain the value 3" inference is processed

Add 3 to R8C1 in the second case, and you will see the difference

[daj95376]

The order of the steps in the displayed solution is not the order of the search

Beautiful !!!

Then, basically, the explanations are useless for understanding the actual logic used in SE.

Why am I suddenly not surprised ???

[lksudoku]

The explanations for dynamic contradiction chains are still valid and explain how a contradiction can be achieved

They just do not always follow the algorithmic search order

For instance, suppose you have search order:
(1) A->B, (2) A->C, (3) B,C->D

It can have a display order
(1) A->C, (2) A->B, (3) C,B->D

if A->C have a side effect (dynamic change for instance) you may think that at step 2 you had that dynamic change while in the search step 2 is step 1 without that dynamic change

[m_b_metcalf]

Here's a real oddity-- this patently ill-formed puzzle is accepted by serate (but not by SE in interactive mode):

Code: Select all

 1 . . 2 . . 2 . .
 . 3 . . 4 . . 5 .
 . . 6 . . 2 . . 7
 7 . . . . 8 5 . .
 . 8 . . 2 . . 4 .
 . . 3 4 . . . . 9
 5 . . 2 . . 6 . .
 . 1 . . 9 . . 7 .
 . . 8 . . 1 . . 5   ED=7.2/7.2/7.1

Regards,

Mike Metcalf

[Pat]

Here's a real oddity --
this patently ill-formed puzzle is accepted by serate
(but not by SE in interactive mode):

Code: Select all

 1 . . 2 . . 2 . .
 . 3 . . 4 . . 5 .
 . . 6 . . 2 . . 7
 7 . . . . 8 5 . .
 . 8 . . 2 . . 4 .
 . . 3 4 . . . . 9
 5 . . 2 . . 6 . .
 . 1 . . 9 . . 7 .
 . . 8 . . 1 . . 5

               ED=7.2/7.2/7.1

amazing

just to clarify,
the problem is in the original software

Code: Select all

java -cp SudokuExplainer.jar diuf.sudoku.test.Tester p.TXT r.TXT

1..2..2...3..4..5...6..2..77....85...8..2..4...34....95..2..6...1..9..7...8..1..5

Analyzing Sudoku #1
52 Hidden Single
1 Direct Hidden Pair
1 Direct Hidden Triplet
6 Pointing
3 Naked Pair
3 X-Wing
3 Hidden Pair
1 Naked Triplet
2 XY-Wing
1 Unique Rectangle type 1
1 Forcing X-Chain
2 Bidirectional Cycle
9 Forcing Chain
Hardest technique: Bidirectional Cycle
Difficulty: 7.2

[champagne]

Hi all,

I would first react to a statement form daj95376.
Some are still working on cloning serate, I know at least three of them including me.

The main problem we face is that we find new problems at each step.

Here one on which lksoduku will certainly comment.

I know for long that serate is not performing well in the area of Y cycles and Y chains.

The deviations in ratings can be very large.

in that puzzle from the ongoing game 145

9....8..7.6..9..5...47..3..8.....2...9..6..1...2.....5..5..14...4..3..7.3..4....1

skfr, the program we are building following serate rules, finds a 6.8 diamond.

serate rates the same puzzle 7.2 7.2 7.1

here is the start on each side

Code: Select all

9    1235 13   |12356 1245 8     |16   246   7     
127  6    1378 |123   9    234   |18   5     248   
125  1258 4    |7     125  256   |3    2689  2689 
--------------------------------------------------
8    1357 1367 |1359  1457 34579 |2    3469  3469 
457  9    37   |2358  6    23457 |78   1     348   
1467 137  2    |1389  1478 3479  |6789 34689 5     
--------------------------------------------------
267  278  5    |2689  278  1     |4    23689 23689
126  4    1689 |25689 3    2569  |5689 7     2689 
3    278  6789 |4     2578 25679 |5689 2689  1     

start form skfr, 3 "y chains" rated 6.8

[] 1r1c7 -> ~1r1c3 -> 3r1c3 -> ~3r5c3 -> 7r5c3 -> ~7r5c7 -> 8r5c7 -> ~8r2c7 -> 1r2c7 -> ~1r1c7

[] 1r2c1 -> ~1r1c3 -> 3r1c3 -> ~3r5c3 -> 7r5c3 -> ~7r5c7 -> 8r5c7 -> ~8r2c7 -> 1r2c7 -> ~1r2c1

[] 1r2c3 -> ~1r1c3 -> 3r1c3 -> ~3r5c3 -> 7r5c3 -> ~7r5c7 -> 8r5c7 -> ~8r2c7 -> 1r2c7 -> ~1r2c3


start from serate, a "xy chain" rated 7.1

(1) If A2 = 1, then A2 <> 7
(2) If A2 <> 7, then C2 = 7
(3) If C2 = 7, then C5 <> 7
(4) If C5 <> 7, then C5 = 3
(5) If C5 = 3, then C1 <> 3
(6) If C1 <> 3, then C1 = 1
(7) If C1 = 1, then A2 <> 1

difficult to figure out why serate did not locate the "y chains";

champagne

[ronk]

difficult to figure out why serate did not locate the "y chains"

Ratings 6.6 through 7.0 apply to xy-loops, but not xy-chains. These rating numbers are shared by other techniques, but xy-chains is not one of them.

[champagne]

difficult to figure out why serate did not locate the "y chains"

Ratings 6.6 through 7.0 apply to xy-loops, but not xy-chains. These rating numbers are shared by other techniques, but xy-chains is not one of them.

rating base 6.5 apply to x loop, y loop
rating base 6.6 apply to x chain y chain

champagne

[ronk]

rating base 6.5 apply to x loop, y loop
rating base 6.6 apply to x chain y chain

Really! Please post one xy-chain ("y-chain") that Explainer finds with rating 6.6, 6.7, 6.8, 6.9 or 7.0.