P.O. wrote:Hi AnotherLife, i am not a human solver i use algorithms; i understand the expression 'human means' as meaning a pattern that a human can find as oppose to a pattern that only a algorithm can find; but you can't know that a algorithm was used or not for the patterns you class as 'human'; of course remains the question of defining the class of human patterns;
concerning this puzzle i have always seen its rating as an anomaly or an error as it is easily solved with a shortest-chain first strategy, only 12 chains of which six are intersections;
here a 4-chains solution in the first state of the grid after intersections:
c3n2{r9 r5} - r5c7{n2 n7} - r1{c6c7}{n3n6} - r1c3{n3n6 n4} => r9c3 <> 4
r1c4{n4 n8} - r7c4{n8 n2} - c7n2{r7 r4r5} - c8n2{r4r5 r2} - r2n4{c8 c1c3} => r1c3 <> 4
b1n8{r3c2 r2c1} - c5n8{r2 r7} - r7c4{n8 n2} - b9n2{r7c7r7c8r7c9 r8c9r9c9} - r2n2{c9 c8} - r2n4{c8 c3} => r3c2 <> 4
r2n4{c8 c1c3} - r1{c3c6}{n3n6} - r1c7{n3n6 n7} - r5c7{n7 n2} - r5c3{n2 n4} - c2n4{r5 r9} - c1n4{r8r9 r2} => r2c3 <> 4
Hi P.O.,
You see, the difference between human solutions and solutions found by programs is not only in the different choice of the methods but in the different order of steps. For example, after your steps 1 and 2 any human would spot a naked pair in r1,
- Code: Select all
.-----------------.----------------.----------------------.
| 1 2 36* | 48 5 36* | 7-36 4789-36 789-36 |
| 468 5 346 | 7 368 9 | 1 23468 2368 |
| 7 3468 9 | 148 2 136 | 5 3468 368 |
:-----------------+----------------+----------------------:
| 269 1 7 | 39 4 8 | 236 2369 5 |
| 3 489 24 | 6 1 5 | 27 2789 2789 |
| 689 689 5 | 39 7 2 | 4 3689 1 |
:-----------------+----------------+----------------------:
| 5 369 1 | 28 3689 4 | 2367 2367 2367 |
| 246 7 8 | 12 36 136 | 9 5 2346 |
| 2469 3469 236 | 5 369 7 | 8 1 2346 |
'-----------------'----------------'----------------------'
which leads to 5 singles and intersections, and then he might apply your step 3 leading to a sequence of singles. So there in no need in step 4 from a human point of view.
As to the SER of this puzzle, I don’t consider it as an anomaly or an error. Rather, this is a drawback of the SE ratings that they don’t take account of group- and ALS-nodes in AICs and forcing chains, and if we restrict ourselves to such chains, we will need dynamic forcing chains to solve this puzzle, so SER 9.0 will be a correct estimate.
To Eleven and TotuanThanks for your solutions! I thought that ALS-nodes were needed to solve this puzzle, but it comes to be that group nodes suffice. Now I will have to make corrections
to my recent post.. Unfortunately, HoDoKu can reach -4 r1c3 only via forcing chains, so it is a bug in the program. Let me write my variant, which is based on your solutions.
- Code: Select all
.----------------- -.----------------.---------------------.
| 1 2 36-4 | g48 5 36 | 367 346789 36789 |
| a468 5 a346 | 7 368 9 | 1 b23468 c2368 |
| 7 3468 9 | 148 2 136 | 5 3468 368 |
:-------------------+----------------+---------------------:
| 269 1 7 | 39 4 8 | 236 2369 5 |
| 3 489 24 | 6 1 5 | 27 2789 2789 |
| 689 689 5 | 39 7 2 | 4 3689 1 |
:-------------------+----------------+---------------------:
| 5 369 1 | f28 3689 4 | e2367 e2367 d2367 |
| 246 7 8 | 12 36 136 | 9 5 d2346 |
| 2469 3469 2346 | 5 369 7 | 8 1 d2346 |
'-------------------'----------------'---------------------'
1. AIC with groups: (4)r2c13 = (4-2)r2c8 = r2c9 - r789c9 = r7c78 - (2=8)r7c4 - (8=4)r1c4 => -4 r1c3; naked pair, 5 singles, lcste
- Code: Select all
.---------------.----------------.-------------------.
| 1 2 36 | 48 5 36 | 7 489 89 |
| c48 5 36 | 7 368 9 | 1 d23468 e2368 |
| 7 48 9 | 148 2 136 | 5 3468 368 |
:---------------+----------------+-------------------:
| 2 1 7 | 39 4 8 | 36 369 5 |
| 3 89 4 | 6 1 5 | 2 789 789 |
| 689 689 5 | 39 7 2 | 4 389 1 |
:---------------+----------------+-------------------:
| 5 369 1 | 28 3689 4 | 36 2367 2367 |
| b46 7 8 | 12 36 136 | 9 5 a346-2 |
| 469 3469 2 | 5 369 7 | 8 1 346 |
'---------------'----------------'-------------------'
2. Basic AIC: (4)r8c9 = r8c1 - r2c1 = (4-2)r2c8 = r2c9 => -2 r8c9; ste
P.S. Denis, thanks for the puzzle.