coloin wrote:BXB-Mastermins1234 - that is the 6259 puzzles
BXB-Mastermins5 - these are the puzzles of hendriks at bottom of page 14
BXB-Mastermins678910 - these are the new ones from the new max expands [Paquita's included]
So, BXB-Mastermins5 and BXB-Mastermins678910 must be added in order to get all the new ones.
As all the calculations I do take much time, I'm thinking of writing scripts specifically intended to the extension of collections by new puzzles, that would do only the necessary calculations. This is based on my splitting of collections by solution grids (a separate folder per solution grid). Examples:
- if puzzles correspond to new solution grids, just add them to the collection in their new solution folders;
- if a minimal puzzle is new, but its BRT expand is not, then only add the minimal to the right file in the right folder;
- if a minimal is new and has a new BRT-expand, then see which of its BRT+1 and BRT+1+BRT expands are new and continue the calculations only with the new ones...
coloin wrote:What would be good perhaps is to identify which puzzles / max expands are the more complex ones which a vicinity search can be more worthwhile done ?
I haven't yet tried systematic expansions of T&E(3) expands, but there may appear complexity problems, as some of them have many clues and may therefore have too many minimals.
I'd be curious to see if minimising all the T&E(3)-expands would lead to many more minimals than just expanding the BRT-expands (as done by mith). Do you have results about this?
Currently, I see no way of saying in advance which expansions are most promising.
.