We can salvage that starred potential xy-wing or xz-almost locked set pattern using POM.
Here is a complete POM merge grid, although we only really needed to work out the labels for the 5's and 7's
- Code: Select all
+-----------------------+--------------------+-----------------+
| 9 6 4ab | 3 1 8 | 2 4cd 7 |
| 5ab | | 5cdef |
| | | |
| 2 5cd 5ef | 9 4 6 | 3abc 3de 1 |
| 7a 7bcdefg | | 8b 5ab |
| 8a | | |
| | | |
| 1 3 4cd | 5ace 5bdf 2 | 8a 4ab 6 |
| 8b | 7abcde 7fg | 9a 9bcde |
+-----------------------+--------------------+-----------------+
| 3b 8 5cd | 1a 2a 1b | 3d 2b 4 |
| 5abef 7a | 7fg 7de 3ce | 6b 3a |
| 6a | 9bc 9d | 9e 9a |
| 7bc | | |
| | | |
| 3ade 4 9 | 6 2b 5 | 1 2a 8 |
| 7defg | 7abc | 3bc |
| | | |
| 3c 1 2 | 4 8 3abd | 3e 7 5 |
| 6b | 9ae | 6a |
| | | 9bcd |
+-----------------------+--------------------+-----------------+
| 8 *7bdf 3 | 1b *5ace 1a | 4bd 6 2 |
| *9b | 5bdf *9ade 4ac | 7aceg |
| | 9c | |
| | | |
| 4 2 6 | 8 3 7 | 5 1 9 |
| | | |
|*5cd 5abef 1 | 2 6 4bd | 4ac 8 3 |
|*7a 7ceg | 9b | 7bdf |
| 9acde | | |
+-----------------------+--------------------+-----------------+
Cells r2c2, r4c3, and r9c1 give you the equations
5
cd = 7
bcdefg; and 5
ab
ef = 7a;
while cells r3c4 and r3c5 give you the equations
5
ace = 7
fg; and 5b
df = 7a
bcde;
These 4 equations can only be true if
5
c = 7
fg;
5
d = 7
bcde;
5bf = 7a;
and 5
a & 5
e are invalid.
Now we bring in the starred almost locked sets xz-pattern
A = {r7c2, r7c5}; B = {r9c1}; z=7; x=5 (or maybe vice-versa).
Because of the xz-pattern, we know that the only valid patterns for 5's must either be in A, or B. Thus 5b and 5f are also invalid, leaving 5cd as the only valid labels for 5's. This kills a whole bunch of 5 candidates and basically solves the puzzle.
Edit - actually, since there is another xz-pattern using cells r4c3, r4c4, and r7c4, that also invalidate labels 5ae, we don't need the labels for 7's after all.