.
.
First, to make things clear: I'm not speaking of the generation process; I'm talking from the point of view of a user of a database.
One problem with the current state of affairs is, puzzles have been stuffed into the ph2010 database without anybody caring to have a look at the contents - except for the highest SERs and for adding exotic patterns to it. That's how the tridagon pattern could escape discovery for 12 years.
AFAIK, I've been the only one to analyse the large databases.
Things have been different with your T&E(3) database, because we talked of what I needed to run such analyses more easily. Unfortunately, there was this misunderstanding about the min-expands.
1) Minimal puzzles have always been the primary concept:
- the definition is self-contained: a puzzle can be checked to be minimal without any reference to any other puzzle, let alone to any database;
- minimal puzzles is what all the puzzle generators generate;
- this is what all the big databases contain;
- this is what (almost) all the published puzzles are;
- this is what allows meaningful statistical analyses.
2) Min-expands in my sense are a derived concept. The associated primary concept is the process of expansion by Singles.
Expansion of a puzzle P (minimal or not) by Singles:
- is uniquely defined;
- requires only trivial computations;
- is intrinsic (doesn't need any reference to any database);
- preserves all the reasonable classifications (but possibly not the SER).
3) Min-expands are the result of this trivial expansion process applied to minimals:
- no one is supposed to ask if a puzzle is a min-expand: min-expands are produced as the expansion of minimals; there's nothing to check about them;
- they are extremely useful for practical classification purposes: they allow to drastically reduce computation times. In order to get an idea of the gain, I started from a sample of 1000 puzzles in your max-expand database, I generated all their minimals, then I computed their min-expands and I finally reduced them by isomorphism: the gain between the minimals and the min-expands is > 7.
When starting from a database of minimals, the first step in studying them is this reduction process of their min-expands by isomorphisms.
-