This is the current list of 898 puzzles which have singles expanded forums with T&E depth 3 and SER 11.8+:
google doc
The small deviation in the rating of different morphs in the high ratings belongs to the second category. This is something I already often had to write. In the low ratings, the search of a contradiction is exhaustive. In the very high ratings, it is not. As a consequence, if the order of the candidates in the search is changed, the process can be stopped with a different rating. This is what happens with different morphs. Which morph will give the highest rating is unpredictable;
mith wrote:The small deviation in the rating of different morphs in the high ratings belongs to the second category. This is something I already often had to write. In the low ratings, the search of a contradiction is exhaustive. In the very high ratings, it is not. As a consequence, if the order of the candidates in the search is changed, the process can be stopped with a different rating. This is what happens with different morphs. Which morph will give the highest rating is unpredictable;
Given improvements in computing power, I would suggest this voluntary limit category *could* be removed and puzzles above some threshold be re-rated accordingly. (I haven't looked at the serate code enough myself to have an idea how much of a task that would be, never mind whether it's worth the effort in the first place.)
champagne wrote:I am highly interested in the TH property breaking this recent family, less by the back doors. I have never seen a manual solver using a back door to solve a puzzle.
987.........96.8......5....74...32...93.4278..........47.....923.....4...28...37. ED=11.8/1.2/1.2
987.........96.8......5....8.........94.3278..3...42..4.....3..37.....92.28...47. ED=11.8/1.2/1.2
987......6...........85....4.32..1..2.1.453......1.......4.128.....8.4.3....23.15 ED=11.7/1.2/1.2
987......6...........85....4.32..1..2.1.453.....3........4.128.....8.4.3....23.15 ED=11.7/1.2/1.2
987......6...........85....4.32.81..2.1.453......1.......4.128.....23.15......4.3 ED=11.7/1.2/1.2
987......6...........85....4.32.81..2.1.453.....3........4.128.....23.15......4.3 ED=11.7/1.2/1.2
987......6...........85....4.32..1..2.1.453......1.....4..23.51....8.43......12.8 ED=11.7/1.2/1.2
987......6...........85....4.32..1..2.1.453.....3......4..23.51....8.43......12.8 ED=11.7/1.2/1.2
987......6...........85....4.32.81..2.1.453......1.....4..23.51.....12.8......43. ED=11.7/1.2/1.2
987......6...........85....4.32.81..2.1.453.....3......4..23.51.....12.8......43. ED=11.7/1.2/1.2
98.76.5..75..498.....8.....32.......1.....6...9.6.4......4..98.....87.45....5.7.6 ED=11.7/1.2/1.2
987......6..8.......5...4..5.32...1.1.238.5.4.......3....53.84....1....2....421.3 ED=11.7/1.2/1.2
987......6..8.......5...4..5.32...1.1.238.5.4......2.....53.84....1....2....421.3 ED=11.7/1.2/1.2
987......6..8.......5...4..5.32...811.23..54.........3...53...4...1.8.2.....4213. ED=11.7/1.2/1.2
987......6..8.......5...4..5.32...811.23..54.......2.....53...4...1.8.2.....4213. ED=11.7/1.2/1.2
987......6..85......4......3.62..1..1.2.653......1.......58.63.....23.51.....12.8 ED=11.7/1.2/1.2
987......6..85......4......3.62..1..1.2.653.....3........58.63.....23.51.....12.8 ED=11.7/1.2/1.2
987......6..85......4......3.62.81..1.2.653......1.......5..63.....23.51.....12.8 ED=11.7/1.2/1.2
987......6..85......4......3.62.81..1.2.653.....3........5..63.....23.51.....12.8 ED=11.7/1.2/1.2
9876.........958.......4...76........392..78...23..6..6.....29.2.8...3.7.93....68 ED=11.7/1.2/1.2
9876.........958.......4...76........932..78...23..6..6.....29.2.8...3.7.3.....68 ED=11.7/1.2/1.2
9876.........958.......4...76........932..78...23..6..6.....39.3.8...2.7.2.....68 ED=11.7/1.2/1.2
98.76.5..7.49..6....5.84...67..9....4.8.57.9..5...........487.........6........32 ED=11.7/1.2/1.2
98.76.5..7.59.4....64.8.9..69.8.....5.........475.6.8....4.76.........9........32 ED=11.7/1.2/1.2
98.76.5..7.49..6....5.48...67..9....4.8.57.9..5...........847.........6........32 ED=11.7/1.2/1.2
987......6..8...........5..4.32...811..3..45.......2.....43.......1.8.24....5213. ED=11.7/1.2/1.2
987......6..8...........5..4.32...1.1..38.4.5......2.....43.8.....1...42....521.3 ED=11.7/1.2/1.2
987......6..8...........5..4.32...811..3..45.......2...4..5213....43.......1.8.2. ED=11.7/1.2/1.2
987......6..8...........5..4.32...1.1..38.4.5......2...4..521.3...43.8.....1....2 ED=11.7/1.2/1.2
98.76.5..7.5.......468.......84.6.75..795...6....8......4.....3....94..........21 ED=11.7/1.2/1.2
98.76.5..7.5.4......68.....4.795..6...8..6.57....8......4....3.....94..........21 ED=11.7/1.2/1.2
98.76.5..75...98.......5...43.......2.....1.....6.1......95..86...1..7.5....8791. ED=11.6/1.2/1.2
98.76.5..75...98.....8.....43.......2.....1.....6.1......95..86...1..7.5....8791. ED=11.6/1.2/1.2
98.76.5..75...98.......5...43.......2.....1.....6.1.....795..86...1..7.5....8.91. ED=11.6/1.2/1.2
98.76.5..75...98.....8.....43.......2.....1.....6.1.....795..86...1..7.5....8.91. ED=11.6/1.2/1.2
98.76.5..75...98.......5...43.......2.....1...7.6.1......95..86...1.67.5....8.91. ED=11.6/1.2/1.2
98.76.5..75...98.....8.....43.......2.....1...7.6.1......95..86...1.67.5....8.91. ED=11.6/1.2/1.2
98.76.5..75..498.....8.....32.......1.....4...7.6.4......95..86....8.94......67.5 ED=11.6/1.2/1.2
98.76.54.75.9....8........75.....4.632........1.6........8...9....57.6.4....4987. ED=11.6/1.2/1.2
98.76.54.75.9....8......9..5.....4.632........1.6........8...9....57.6.4....4987. ED=11.6/1.2/1.2
98.76.5..7.4.......56.......9..8..5..6745......89.6.4....89.3.2....7...1......... ED=11.6/1.2/1.2
98.76.5..7.45.......6......5.89.6.4..9..8..5..674........89.3.2....7...1......... ED=11.6/1.2/1.2
Paquita wrote:Denis I am using your program to rate my previous 11.8 puzzles from august 13th. It takes a long time, I think I have too many print options enabled, have not figured that out yet. I am now rating them with the T&E(3) option, so far the ones I did are all T&E(BRT,3). But I wonder, if a puzzle is in T&E(2) or T&E(1), can it be rated by T&E(3)? In other words, should one try to solve the puzzle in T&E(1) first, if not solved, in T&E(2), if not solved still, in T&E(3)? I am not sure yet my ratings of T&E(BRT,3) are right. And is there an option to print to file? (if possible, to print just the rating, not all the intermediate steps).
(bind ?*print-actions* FALSE)
(bind ?*print-levels* FALSE)
(bind ?*print-main-levels* FALSE)
(bind ?*print-solution* FALSE)
(bind ?*print-hypothesis* FALSE)
(bind ?*print-RS-after-Singles* FALSE)
(bind ?*print-RS-after-whips[1]* FALSE)
(bind ?*print-final-RS* FALSE)
(bind ?*print-hypothesis* FALSE)
denis_berthier wrote:mith wrote:The small deviation in the rating of different morphs in the high ratings belongs to the second category. This is something I already often had to write. In the low ratings, the search of a contradiction is exhaustive. In the very high ratings, it is not. As a consequence, if the order of the candidates in the search is changed, the process can be stopped with a different rating. This is what happens with different morphs. Which morph will give the highest rating is unpredictable;
Given improvements in computing power, I would suggest this voluntary limit category *could* be removed and puzzles above some threshold be re-rated accordingly. (I haven't looked at the serate code enough myself to have an idea how much of a task that would be, never mind whether it's worth the effort in the first place.)
In order to avoid misleading newbies into hopeless coding efforts, let it be clear that dealing with this problem on non-exhaustive search in SER or with any bug encountered in it, will not free SER of its possible variations with morphs. I've explained the reasons in a previous post: the absence of the confluence property for the set of rules defining a fixed SER rating.
champagne wrote:Hi mith,
I don't know what is the average rating time for these puzzles with so many clues, but in the 19-22 area, a puzzle rated around 11.7 could require hours of computation with serate. This has been the reason to create skfr having a quick pre rating for the interactive game.
[98.7.....7.....6....6.5.....4...5.3...79..5......2...1..85..9......1...4.....3.2.] 11.9 11.9 11.8 DCFC+DFC
time 121.67701625823975
[98.7.....6.....87...7.....5.4..3.5....65...9......2..1..86...5.....1.3.......4..2] 11.9 11.9 11.6 DCFC+DFC
time 63.65245199203491
[12..3....4....1.2...52..1..5..4..2......6..7......3..8.5....9....9.7..3......8..6] 11.9 11.9 11.3 DCFC+DFC
time 54.75340390205383
[.......39.....1..5..3.5.8....8.9...6.7...2...1..4.......9.8..5..2....6..4..7.....] 11.9 11.9 11.3 DCFC+DFC
time 141.90189957618713
[.2.4...8.....8...68....71..2..5...9..95.......4..3.........1..7..28...4.....6.3..] 11.9 11.9 9.9 DCFC+DFC
time 116.47096490859985
[........1....23.45..51..2....25...1..6...27..8...9......42....7.3...6...9...8....] 11.9 11.9 9.9 DLFC+DFC
time 70.52767658233643
[12.3.....4.5...6...7.....2.6..1..3....453.........8..9...45.1.........8......2..7] 11.9 11.9 2.6 DCFC+DFC
time 160.4379918575287
[5.6...7...1.3.....8...5.9.....1...2.....8.6.7.....2.4.7...9...6.3...42....5......] 11.9 1.2 1.2 DCFC+DFC
time 81.70554542541504
[..3..6.8....1..2......7...4..9..8.6..3..4...1.7.2.....3....5.....5...6..98.....5.] 11.9 1.2 1.2 DCFC+DFC
time 57.40283441543579
champagne wrote:Some members of the forum having a good practice of JAVA fixed some bugs in serate, but nobody, AFAIK, tried to reach a rating morph neutral although this oddity has been source of trouble in many case in the game.
dobrichev wrote:champagne wrote:Some members of the forum having a good practice of JAVA fixed some bugs in serate, but nobody, AFAIK, tried to reach a rating morph neutral although this oddity has been source of trouble in many case in the game.
At least I tried.
Essentially, the problem is when at some point several elimination techniques have exactly the same minimal rounded rating, calculated by the non-linear expressions spread in the code.
The first elimination step wins and determines later resolution path.
"First" is determined by some hash value and here the morphs play their role.
This could be fixed, when different possible techniques are internally rated differently, so that the techniques with the exactly same final rounded rating can be ordered by this hidden property.
Finally, if several techniques still can't be distinguished by their combination of rounded and hidden rating, then all they must be applied at once, before restarting the process in finding the next cheaper elimination.
Of course, there is no consensus on rating changes, and no volunteers to code the new rating rules.
dobrichev wrote:Of course, there is no consensus on rating changes, and no volunteers to code the new rating rules.
mith wrote:
This sounds like a different issue from the one champagne was talking about previously (of the contradiction search not being exhaustive). Or is this part of what is causing the search to not be exhaustive?
I'd be curious if you have an example of a puzzle where this hash "tiebreaker" is causing the puzzle to rate differently afterward. I would think it could never cause the rating to go up on its own - that is, if this tie due to rounding occurs at a 11.6 step, whichever step is applied should leave the alternative(s) at 11.6 or less - other than considerations like uniqueness techniques or somehow affecting the non-exhaustive nature of the search. But I definitely don't have a clear picture of what goes into the "length" calculation for chains.
6.......2.9.4...5...1...7...5..84.......2.......3.5.4.2.....6...3...9.8...7.....1;11.80;11.80;11.10;tax;coloin-04-10;13;21;
mith wrote:For the 70 puzzles, had an average of 91.56, max 235.05, min 35.08
The one that took the longest was
- Code: Select all
6.......2.9.4...5...1...7...5..84.......2.......3.5.4.2.....6...3...9.8...7.....1;11.80;11.80;11.10;tax;coloin-04-10;13;21;
I'll try rating some later with the pattern game serate to compare.