denis_berthier wrote:

Hi Defise,

Good, you have fewer rule applications than in the simplest-first strategy (with the same max length of whips).

Can you remind me your criterion for choosing the next target (within a fixed Wn)? I can't find the post where you told it. Was it the maximum number of candidates that would be subsequently eliminated by Singles and ECP (or maybe also Whips[1] and Subsets) if the target was eliminated? Did you also consider the number of candidates turned into solution values?

Hi Denis,

Here is the link of my post :

robert-s-puzzles-2020-11-14-t38411-15.html#p297633

- Yes, my first criterion is the number of candidates that would be subsequently eliminated by basic techniques if the target was eliminated (here basics = singles + alignments + naked & hidden pairs + naked triplets).

- I don’t consider the number of candidates turned into solution values, because the placement of a candidate leads to a lot of deletions of other candidates. But I could easily add that as second criterion or even as first criterion.

- Note that I added a 3rd criterion which is a random number, which leads to different choices of "ex aequo" targets (modulo criteria 1 et 2) from one execution to another.

For example for this 4leaf puzzle, 20 executions gave:

1 solution with 5 whips [<= 5],

8 solutions with 4 whips [<= 5]

11 solutions with 3 whips [<= 5].