The New Sudoku Players' Forum

[Paquita]

Some new puzzles, T&E(2) https://drive.google.com/file/d/1u8ol3BdPIMLYwVRyFGxr3G77DUFqlP-0/view?usp=sharing

[denis_berthier]

.
hi Paquita,

Using the SHC (Sudoku Hierarchical Classifier), I computed the BpB classification of your last 3 batches. I found no puzzle in B7B.
.

[Paquita]

And a new batch https://drive.google.com/file/d/1LkEEzEmsNK5sb50bq-GH7PB2423tdhXD/view?usp=sharing

[denis_berthier]

And a new batch https://drive.google.com/file/d/1LkEEzEmsNK5sb50bq-GH7PB2423tdhXD/view?usp=sharing

Hi Paquita,

I've launched the calculations for the BpB ratings (still running). I observe that most of the puzzles (all but 4 in the first 3000) are in B6B.
Have you used the Sudoku Hierarchical Classifier (SHC) as your filter, instead of SER ?

[Paquita]

Denis,

I have never used SHC. I use PGXplainer now to rate after a preliminary test for higher rates with skfr (skfr rates above 11.2). I use previously found puzzles as seed. I don't know why they are almost all in B6B.

[denis_berthier]

Denis,
I have never used SHC. I use PGXplainer now to rate after a preliminary test for higher rates with skfr (skfr rates above 11.2). I use previously found puzzles as seed. I don't know why they are almost all in B6B.

Hi Paquita,
Thanks for your answer.
This is very strange. PGX or SER is the same thing (except in very rare cases). The BpB distribution you get in this batch is completely different from your previous batches. Have you changed any of your filters? Or are your seeds different in any way from your previous seeds?

BTW, you should try SHC, it's orders of magnitude faster than SER or PGX. And it is as easy to use.

[Paquita]

Hi Denis

I will try SHC, thanks for the tip. I have no idea just now why I have these results. I use a few methods to generate new puzzles, one is to change one digit in a puzzle, another is to expand the seeds. That has not changed. The seeds now are the puzzles from my previously published files, before the seeds were mainly the puzzles from mith and hendrik added to the T&E(2) collection. (I have already used most of the ph2010 puzzles as seeds a couple of years ago). I use T&E(2) puzzles as seeds to obtain new t&E(2) puzzles, as the T&E(3) area is well covered by mith and hendrik.

[denis_berthier]

.
Hi Paquita,

Using SHC, the BpB calculations for the whole file of 6769 puzzles took 10 hrs on my Mac M1.

Among the 6769 puzzles (all with SER ≥ 11.4), 5372 have BpB = 6 and 397 BpB = 5. None has BpB = 4 or less.

But the most surprising result is, among the 3853 with SER = 11.7, 3825 (99.3 %) have BpB = 6. This is much higher than any previous T&E(2) collection.

It'd be interesting to understand how this is possible by trying only to maximise the SER, but at this point, I don't.

[denis_berthier]

.
For people involved in the search for the "hardest" puzzles, these comparisons of computation times may be interesting:
http://forum.enjoysudoku.com/pgexplainer-a-minimal-sudokuexplainer-in-56-712-bytes-t39049-52.html
.

[denis_berthier]

.
Paquita's last list of 6769 puzzles is definitely very special for a T&E(2) collection. In addition to its BpB distribution:
- all the puzzles in it have a non-degenerate tridagon;
- only 270 of them are not solved in W8+Trid-OR5W8.

[denis_berthier]

.
Finally, I have some explanation. As it doesn't involve SER but tridagons, I put it in the tridagon thread: http://forum.enjoysudoku.com/the-tridagon-rule-t39859-120.html
.

[Paquita]

Hi

A new batch https://drive.google.com/file/d/1dBsPcIRSsMgY_4GUZ-w3is6wPUUKh6yN/view?usp=sharing

I have not checked BpB or tridagons yet. It is a T&E(2) collection.

[Paquita]

And another batch TE(2) https://drive.google.com/file/d/1mqlI6A3rosPrkLUk9vgkKxadAs4nC2Si/view?usp=sharing

[mith]

Hoping to get back to this in the next week or two. Busy busy.

Well, this didn't happen for a number of reasons, including that I've had to have surgery for a torn labrum and am in a sling until the end of this month. I'll get to it eventually! (Recovery is going well, back to work and slowly catching up on things in general.)

[eleven]

oops, wish you a good and complete recovery !