denis_berthier wrote:Do you mean the SER will now appear in the min-expand list instead the minimal list? That's good for me.
For the T&E(3) database, yes. The expanded forms are already rated, the minimals are not. Just haven't included it in the past because I know it's not particularly relevant for you and you're the main one using the updates.
Do you mean all the intermediate puzzles between the min-expands and max-expands. I don't see the point.
It's really just for convenience. As I said, they could be generated easily (as can the minimals), but it's already been done and I have the space so there's no point requiring anyone else to generate them.
One thing you could add to the min-expand list is some cabalistic sign on lines, meaning it's a new puzzle wrt to the previous version.
I'll provide this in a README or something rather than in the file itself, sure. Because the set of puzzles is closed for a given update (all puzzles related by operations "transform", "expand", "minimize", and "add" are generated before any update), the new puzzles will just be those after a particular ID.
It'd be much better to have a lower SER cutoff in it. As I said before, restricting it to 11.6+ would eliminate potentially interesting puzzles. Also, 11.6 would make no sense wrt to the types of chains/nets in SER:
- Code: Select all
10.2 Dynamic + Forcing Chains (37-48 nodes) CRCD Forcing Chains
10.3 Dynamic + Forcing Chains (49-72 nodes) CRCD Forcing Chains
10.4 Dynamic + Forcing Chains (73-96 nodes) CRCD Forcing Chains
10.5 Dynamic + Forcing Chains (97-144 nodes) CRCD Forcing Chains
10.6 Dynamic + Forcing Chains (145-192 nodes) CRCD Forcing Chains
10.7 Dynamic + Forcing Chains (193-288 nodes) CRCD Forcing Chains
10.8 Dynamic + Forcing Chains (289-384 nodes) CRCD Forcing Chains
10.9 Dynamic + Multiple Forcing Chains (73-96 nodes) CRCD Forcing Chains
11.0 Dynamic + Multiple Forcing Chains (97-144 nodes) CRCD Forcing Chains
11.1 Dynamic + Multiple Forcing Chains (145-192 nodes) CRCD Forcing Chains
11.2 Dynamic + Multiple Forcing Chains (193-288 nodes) CRCD Forcing Chains
11.3 Dynamic + Multiple Forcing Chains (289-384 nodes) CRCD Forcing Chains
11.4 Dynamic + Multiple Forcing Chains (385-576 nodes) CRCD Forcing Chains
11.4 [Dynamic + Dynamic Forcing Chains (73-96 nodes) Region/Contradiction Forcing Chains]
11.5 [Dynamic + Dynamic Forcing Chains (97-144 nodes) Region Forcing Chains]
11.6 [Dynamic + Dynamic Forcing Chains (145-192 nodes) Cell Forcing Chains]
11.7 [Dynamic + Dynamic Forcing Chains (193-288 nodes) Double Forcing Chains]
Note that most of the terms used make no sense to me (Dynamic, Multiple Forcing, Contradiction Forcing, Double Forcing)
But what's sure is,
something happens within SER level 11.4: a new type of chains (Dynamic Forcing Chains) starts to be considered.
Oh, it's much worse than this, as I alluded to in the previous post. The range for Dynamic + Multiple Forcing Chains (DFC+MFC) is actually 10.6-11.7 (in known puzzles), and the range for Dynamic + Dynamic Forcing Chains (DFC+DFC) is 11.2-11.9 (in known puzzles). (DFC+FC is 10.1-11.1). There's a huge overlap, and I would find it *much* easier to understand an 11.2 DFC+DFC (up to 48 nodes) than an 11.2 DFC+MFC (up to 256 - the 288 is an error in that list) for example.
The rating is just a sum of the base value (which is +0.5 per level for Dynamic chains; 8.5 for DFC, 9.0 for DFC+, 9.5 for DFC+FC, 10.0 for DFC+MFC, 10.5 for DFC+DFC) and the value from the node count, which corresponds to 0.2 for every doubling of the node count (but awkwardly alternating between 3/2 and 4/3 for steps, rather than sqrt(2) every time - if they were sqrt(2) we'd already have at least one more 11.9 in mike's 20c). The specific type of Dynamic Forcing Chain (Contradiction, Region, Cell, Double) is not related to the rating at all other than that some types are more likely to have more nodes. The expanded form 11.8s have all four types represented (the 11.9s are all DCFC+DFC, Contradiction). But if I ever use D*FC+DFC that's why.
I'll try to write up some definitions of these at some point, but I'm going to have to consult the code to figure out exactly how SE approaches them. (For example, a Forcing Chain should be equivalent to an Alternating Inference Chain? But I don't think SE is coded as AICs. Multiple Forcing Chains include Region and Cell Forcing Chains. Not sure about Double.)
As for me, until very recently, I had used (occasionally) only the "real" SudokuExplainer. Experience with PGXplainer shows it's much faster (1to9only says 3 times; I have no precise timings, but 3 times seems to be a modest estimate IMO).
I don't take the idea that PGXplainer solves all the morph problems, but it does solve some of them. Which makes it a better candidate for replacing the "real" SudokuExplainer.
I'm personally switching to PGXplainer. Which nobody should care about, as I'm not searching for hard puzzles. I'm only working on analysing them. As a result, what I'm interested in is the min-expands, in order not to waste time on (possibly non-isomorphic but nevertheless) redundant puzzles.
I currently use SukakuExplainer (tarek's fork), but really the only reason is that it supports multithreading. At some point I will just switch to PGXplainer and set up some parallelization via scripts.