The New Sudoku Players' Forum

[urhegyi]

solution:

Hidden Text: Show

Code: Select all

37618425
54823167
76432581
21584736
15267348
83741652
62375814
48156273


I was able to solve this sudoku in 4 steps basicly with contradiction and finding an assumption which led to an empty cell so it can be eliminated. Like to know how to do it in a better way.
My solvepath was basicly to prove (in 3 steps) that R5C3 can't be 1 so that R5C1 must be 1, and further assuming that R7C6=4 which leads (through a chain) to a contradiction, so R7C6=8.
Any more methodical approach?

Code: Select all

..618.2............3258..158..3..5..734..3741............8.562..

Wondering why with so many cluess given this sudoku is so difficult to solve? Or am I missing something?

[urhegyi]

My conclusion when starting to solve 2X3 and 2X4 sudokus is that through the symmetry they are less difficult to solve as the 3X3 one's.
I found a post on 2X3 where all different grids were calculated and stated that the maximum difficulty was around SE 8.3.
Is there simular work done on 2X4 and what's the hardest puzzle possible?

[1to9only]

Code: Select all

..618.2............3258..158..3..5..734..3741............8.562.. 2Rx4C


The difficulty level is about right for this grid.

Code: Select all

Pencilmarks, after basics, a couple naked pairs and a naked triplet:
+-------------------------+-------------------------+
| 35    47    6     1     | 8     47    2     35    |
| 358   247   38    27    | 35    1     6     47    |
+-------------------------+-------------------------+
| 67    67    4     3     | 2     5     8     1     |
| 2     1     5     8     | 4     67    3     67    |
+-------------------------+-------------------------+
| 168   5     128   26    | 7     3     4     68    |
| 68    3     7     4     | 1     68    5     2     |
+-------------------------+-------------------------+
| 13467 267   123   267   | 35    48    17    3458  |
| 1347  8     13    5     | 6     2     17    34    |
+-------------------------+-------------------------+

Either:
6.6, Turbot Fish (w/4 nodes): R7C1.3 off: r7c1<>3
6.8, Bidirectional Y-Cycle (w/12 nodes): R1C6,R4C6,R6C6,R5C8,R5C4,R2C4,R1C2: r5c1<>6, r2c2<>7, r7c4<>2
7.1, Forcing Chain (w/6 nodes): R3C2.6 on: r3c2<>7, r3c2=6
1.0, Hidden Single: R3C1: 7 in block: r3c1=7
1.5, Hidden Single: R8C7: 7 in row: r8c7=7
1.0, Hidden Single: R7C7: 1 in column: r7c7=1
4.2, XY-Wing: Cells R7C1,R7C6,R6C1 on value 8: r6c6<>8
stte
ED=7.1/1.2/1.2

Or:
3.7, Two-String Kite: Cells R1C1,R1C8,R2C5,R7C5: 3: r7c1<>3
6.8, Bidirectional Y-Cycle (w/12 nodes): R1C6,R4C6,R6C6,R5C8,R5C4,R2C4,R1C2: r5c1<>6, r7c4<>2, r2c2<>7
4.1, W-Wing: Cells R3C1,R3C2,R7C2,R7C4: 7 and 6: r8c1<>7, r7c1<>7
1.5, Hidden Single: R3C1: 7 in column: r3c1=7
1.0, Hidden Single: R3C2: 6 in block: r3c2=6
1.5, Hidden Single: R8C7: 7 in row: r8c7=7
1.0, Hidden Single: R7C7: 1 in column: r7c7=1
4.2, XY-Wing: Cells R7C1,R7C6,R6C1 on value 8: r6c6<>8
stte
ED=6.8/1.2/1.2


I think very long chains are not possible in 6x6 and 8x8 grids, so the max difficulty is in the ED=8.x region.

Some hard 8x8 grids, from about 1000 randomly generated grids:

Code: Select all

..........2..456.4...6....3.2.....54..387..8....28.........51.2. 2Rx4C 19 ED=8.4/1.2/1.2
.53....7..8.1....8...2..1...5.6...45.......8.4.3.2..8........17. 2Rx4C 19 ED=8.4/1.2/1.2
....2.....5..6..1.........8..74....7...3...3..8462.....58..5..7. 2Rx4C 18 ED=8.3/1.2/1.2
12.74.6....5.......8....2...3..17......4.5.46....7...2...6.2..1. 2Rx4C 20 ED=8.3/1.2/1.2
..28.....3.7...2.........6.3.5.4....1...7.....25.27.51.31...4... 2Rx4C 20 ED=8.3/1.2/1.2
3.7.8......1..76..2..813.....2........4.1......77.8....12.....5. 2Rx4C 19 ED=7.7/1.2/1.2
6.1......3..6.8.7.2....4...3..5.32..4........5.....23.6.4......2 2Rx4C 19 ED=7.6/1.2/1.2


[1to9only]

Here is a Hard Sudoku 8x8:

Code: Select all

..........6.47...3....2..8............6..2....7...85.6.......... 2Rx4C-DG-X 12 ED=6.5/1.5/1.5


And a couple of harder ones:

Code: Select all

......7.6..8.2..2...7......................3...1..2.4..3.4...... 2Rx4C-DG-X 12 ED=7.1/1.5/1.5
..8..3...................1....8..5...7..4.2.1.........2......... 2Rx4C-DG-X 10 ED=8.4/1.2/1.2


[urhegyi]

After my first attempts with the hard level, I tried four evil ones.
Most of all solve with up to 4 chains.

Code: Select all

.1.45....5.3...2.8.1...3...5.8....3.6...8...3.7.4...2.1....78.6.
..3...2...873......8..4.31...6....2...65.6..4......124...3...7..
.8..6...6.4..7......2..8.6...57..76...4.8..2......8..3.4...1..5.
......6...7.83.2...15.2...8....43....5...8.42...1.43.7...6......


[urhegyi]

Here is a Hard Sudoku 8x8:

Code: Select all

..........6.47...3....2..8............6..2....7...85.6.......... 2Rx4C-DG-X 12 ED=6.5/1.5/1.5


And a couple of harder ones:

Code: Select all

......7.6..8.2..2...7......................3...1..2.4..3.4...... 2Rx4C-DG-X 12 ED=7.1/1.5/1.5
..8..3...................1....8..5...7..4.2.1.........2......... 2Rx4C-DG-X 10 ED=8.4/1.2/1.2


For the first I was able to place all the 6's then all the 2's.

[urhegyi]

Here is a Hard Sudoku 8x8:

Code: Select all

..........6.47...3....2..8............6..2....7...85.6.......... 2Rx4C-DG-X 12 ED=6.5/1.5/1.5


And a couple of harder ones:

Code: Select all

......7.6..8.2..2...7......................3...1..2.4..3.4...... 2Rx4C-DG-X 12 ED=7.1/1.5/1.5
..8..3...................1....8..5...7..4.2.1.........2......... 2Rx4C-DG-X 10 ED=8.4/1.2/1.2


For the first I was able to place all the 6's then all the 2's.

Edit:
I'm I correct?

Hidden Text: Show

Code: Select all

84715236
35624781
43176528
58267413
17532864
62481375
71853642
26348157

[1to9only]

Here is a Hard Sudoku 8x8:

Code: Select all

..........6.47...3....2..8............6..2....7...85.6.......... 2Rx4C-DG-X 12 ED=6.5/1.5/1.5


And a couple of harder ones:

Code: Select all

......7.6..8.2..2...7......................3...1..2.4..3.4...... 2Rx4C-DG-X 12 ED=7.1/1.5/1.5
..8..3...................1....8..5...7..4.2.1.........2......... 2Rx4C-DG-X 10 ED=8.4/1.2/1.2


Solutions: Show

Code: Select all

8471523635624781431765285826741317532864624813757185364226348157 2Rx4C-DG-X 12 ED=6.5/1.5/1.5

5142637867381245281675347354281682613457457386211625478334875162 2Rx4C-DG-X 12 ED=7.1/1.5/1.5
1287436536452871283671547154628365138742472815365471362883625417 2Rx4C-DG-X 10 ED=8.4/1.2/1.2



[urhegyi]

Any idea if Disjoint groups(offset sudoku) and diagonals also works with 2X3(6x6) sudoku?
If so, can it be added to the existing Sukaku6Explainer? For now there is no DG support.

[1to9only]

I think 6x6-DG generates some valid grids, but they are all singles only, ED=1.2/1.5 [edit: this is from a small sample. on a much larger sample, about 2% are non-singles only!].
6x6-DG-X is probably not possible, as I've failed so far to generate a solution grid!!
This probably means DG will not make it into Sukaku6Explainer.

[999_Springs]

My conclusion when starting to solve 2X3 and 2X4 sudokus is that through the symmetry they are less difficult to solve as the 3X3 one's.
I found a post on 2X3 where all different grids were calculated and stated that the maximum difficulty was around SE 8.3.
Is there simular work done on 2X4 and what's the hardest puzzle possible?

the relevant thread is here. there are nine 6x6 SE 8.4's after singles

Code: Select all

1...56...1..21...5.35..13.1.6..62.1.
....5.4....3...5.1...34.3.2.1..6..3.
.2......61..2.4...61.....4.6.5.6..3.
.2.4....6...2.4...61......2..5..123.
.....64...2.23...1..5..2...61....2.4
1....6....2..3..6.6.5.4....61....2.4
..3....5.1....456...5..234....5.1...
..34...5.....3..6..1.3.234....5.1...
1...5......3...5.1...34...2..556...4

the first one has 16 clues. the other 8 all have 12 clues and the complements of the clue grid in the solution grid are all isomorphic, but i'm not sure if the 8 puzzles are all isomorphic.

come to think of it i only rated about 70% of the 542257 no-first-single 6x6 puzzles because they have a 9x9 representation and i could plug them into champagne's skfr. after two years, i never got around to doing the other 30% of them with 1to9only's 6x6 explainer! i'll get that done at some point in the hope that another 8.4 or harder puzzle will show up

i highly doubt that an exhaustive search of all 8x8 puzzles would be possible as my guess is that the search space would be too big

I think 6x6-DG generates some valid grids, but they are all singles only, ED=1.2/1.5 [edit: this is from a small sample. on a much larger sample, about 2% are non-singles only!].
6x6-DG-X is probably not possible, as I've failed so far to generate a solution grid!!
This probably means DG will not make it into Sukaku6Explainer.

no 6x6-DG-X grid exists. i did an exhaustive pencil-and-paper search this morning because of boredom (don't ask)

boring manual proof of non existence: Show

assume 2Rx3C boxes, and fill in row 1 with 123456
notation: "1/" and "1\" mean a hidden single or locked candidate elimination by looking at the 1s in the "/" and "\" diagonals respectively. "ns" means naked single(s)

split into cases, based on the order of 4,5,6 in r2c123

case r2c123=456:
6\ r34c5-6
6b4 r4c4=6
(*) 14/ r4c2-14
ns r4c2=3
3\ r6c4-3
oddagon 3r2c4,r2c6,r6c6,r6c1,r3c4 => no solution

case r2c123=465:
5\ r4c4=5
same as (*) from above

case r2c123=546:
r4c4=6 as from case 456
5/ r3c3-5
ns r3c3=2, r5c5=3, r5c2=1, no 1 in c3

case r2c123=564:
r3c3=2 as from case 546
4\ r5c5=4
4/ r6c1=4
4b3 r4c2=4
ns r4c4=3, r6c6=5, r3c4=5, no 5 in c2

case r2c1=6:
6\ r5c5=6, no 6 in b4

[urhegyi]

After my first attempts with the hard level, I tried a few evil ones.
Most of all solve with up to 4 chains.

Code: Select all

.1.45....5.3...2.8.1...3...5.8....3.6...8...3.7.4...2.1....78.6.
..3...2...873......8..4.31...6....2...65.6..4......124...3...7..
.8..6...6.4..7......2..8.6...57..76...4.8..2......8..3.4...1..5.
...7.3....6....4.7....81..14..6..1..86..85....3.1....7....3.5...
......6...7.83.2...15.2...8....43....5...8.42...1.43.7...6......


The last one needed an AIC, a discontinuous nice loop and two xy-chains.

Conclusion of solving above examples is that after basics the further solvepath highly depends on finding a nice opening set. The question now is how to find out this was the most elegant solvepath or eventually there were other better solutions. So taking another startposition can determine the way a sudoku solves. A computer program usually starts with the simplest step that it finds. But the human solver can( sometimes with a bit of luck) find an elegant way to break open the puzzle. For the last puzzle I like to know which are the first steps SudokuExplainer(2X4) finds after basics? I estimate the rating around SE 7.3 based on following solvepath.

My way to start was whatever the value of R2C2(1 or not 1) R1C5 is always 7. Second step was proving that R5C7<>1 by xy-chain. Third step was starting with R5C4=6 a contradiction chain led to R5C4<>6. Then a locked candidate and a few singles followed by another xy-chain proving that R6C3<>1=6. A naked triple in row 5 finally solved this puzzle. Curious to compare this with a solver based solution.

[urhegyi]

Code: Select all

5..8...1.37.5...7.....1..2.6.5....3.4.5..5.....3...2.84.8...7..2


[urhegyi]

I could solve it when assuming R5C8=7 leads to R4C1<>1/3/4 is an empty cell and so R4C8 must be 7.
But there must be better ways to solve this.
The solution I found was a lucky shot, but for me it's not the logical way I prefer.

[urhegyi]

When I compare an evil 8x8 sudoku like

Code: Select all

5..8...1.37.5...7.....1..2.6.5....3.4.5..5.....3...2.84.8...7..2

with a hard one like

Code: Select all

..4736.13..6....1......4.4...12..71...4.8......2....2..84.2873..

I find the hard one much harder then the evil one based on the way I solved them.
Can the rating prove this or am I missing some obvious steps in the hard one that force me to use harder logic then necessary?