Start of the scan on solution grids.
More than a test, this is a start of the scan of solution grids to try to detect a new “loki” family.
We expect in average 0.5 “loki patterns” per solution grid. The scan start with the first million of min lexical grids, all of them starting with the band 1 (index 0).
Here, the number of “loki patterns” is much higher in average 3.32 per solution grid for the first 100 000.
As the band 1 has a high potential to produce the first band or the first stack of the magic square, this was expected.
The ratio decreases slowly in this first million grids, but remains over 2.5. It goes down to 1.96 for the second million solution grids.
I run the test on a mix of old computers, so the runtime per solution grid is not easy to establish, but on an old i7 4770 3.4Gh, it seems to be around 2.5 seconds per target with 5 cores working in parallel.
The output is a list of grids not solved in T&E(1) and not solved with some easy steps of a quick solver.
The grids are cleared if they can be expanded to more than 32 clues (nearly no chance to get then a high rating).
The output is the expanded form with most of the redundancy cleared.
2 chunks are closed:
The first 100 000 solution grids
The solution grids 100 001 to 400 000
The chunk 400 001 to 1 000 000 in ongoing.
The first chunk produced 266 431 records, 2546 of them rating skfr >=10.5
The second chunk produced 488 954 records, 7906 of them rating skfr >=10.5
I just keep here the ratings >=11.3
- Code: Select all
2.4.5..............7....63..8.13..56...7.81.3....6578.5...83.1....6.1...8..57..6.;11.3/2.0/2.0
.3..47...6.1....9.8.........2.45..79.5.7.92.3....2354.....94...2..3.5..7...27..3.;11.6/10.4/3.4
.......8...4.5......23........87....1.35.68.7.8..1356.....81.766..7.53.8.7863.1..;11.4/11.0/2.6
.62...8...4.5.9...............31..544..9.83.11...4598..14.53.98.9.4.1..3...89....;11.4/11.4/3.4
.......4.....39...6.2.......8.31..54.5.9.83.1....4598.5.1.84.3....1.3...3.859....;11.4/2.0/2.0
...2.1...34....1..8.1...5....679..12...1.26.5....6597.....27......5.9...97.61..5.;11.5/10.4/2.6
More can be sent by mail to anybody willing to work on a big file.
The third chunk could be closed end of this week.