lksudoku wrote:a BUG pattern can have an even number of solutions, that is 0,2,4 and so on

Therefore a BUG-lite can also have an even number of solutions

In a valid puzzle a BUG (as in BUG+n, all unsolved cells included) always has 0 solutions. As BUG-lite is a lighter version of BUG, a BUG-lite with 0 solutions should of course be accepted.

But in this case the 0 solutions is not caused by the BUG-lite, but by the impossible pattern on candidate c (see below). So essentially, any elimination made using that BUG-lite pattern could have been made as a non-uniqueness elimination using a smaller pattern.

- Code: Select all
` . c . | . . . | c . . `

. . . | . . c | . c .

c . . | . . c | . . .

----------+----------+--------

. c . | . . . | . c .

. . . | . . . | . . .

c . . | . . . | c . .

RW