Selecting puzzles of interest with a tridagon threat.

Post the puzzle or solving technique that's causing you trouble and someone will help

Selecting puzzles of interest with a tridagon threat.

Postby champagne » Fri Apr 25, 2025 6:03 pm

Several expressed the need to have a “filter” to extract from big files some puzzles of interest for manual players.
This is something I have done in the past for “exotic patterns”, but nothing is done for the “tridagon threat” where we have files with millions of puzzles.

To avoid dilution in other threads, I open this one to address the topic.

We are here interested by a “tridagon impossible pattern” coming in the PM of a puzzle.

We know that 3 digits cannot have in four boxes the following pattern
Code: Select all
..x x..
.x. .x.
x.. ..x

x.. x..
.x. .x.
..x ..x


At the end, the solution grid will have or not the digits at the right place in boxes 123 and 2 digits at the right place in box 4 this is commonly called here a “tridagon pattern in the solution grid”.

In the PM, the impossible pattern has the 3 digits plus extra candidates (at least one will be true).
We have a “tridagon threat” if, using a classical set of rules (for me Sudoku Explainer rating), the rating is very high with a small number of extra candidates.

With only one extra candidate, the solution grid will always have a “tridagon pattern in the solution grid”.
And if only one cell of the impossible pattern does not have one of the 3 digits, the puzzle can be morphed to have this in the box4

The corollary property is that if the solution grid does not have the tridagon pattern, then we have a minimum of 2 guardians.

In mith’s file of more than 4.3 million puzzles, ~870 000 puzzles have a solution grid without the tridagon pattern.
For all the puzzles of the file, we have in the PM a tridagon impossible pattern with 2 false extra candidates.
Here is the first in my file
Code: Select all
........1......234....2567...76.2....65.89...8..75.9....8.9...7.52..6...79...8... npuz= 1
623947851579861234184325679937612485265489713841753926318294567452176398796538142
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 2346      23478     3469     | 3489      346       347      | 58        589       1        |
| 5         178       169      | 189       16        17       | 2         3         4        |
| 134       1348      1349     | 13489     2         5        | 6         7         89       |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 9         134       7        | 6         134       2        | 13458     1458      358      |
| 1234      6         5        | 134       8         9        | 7         124       23       |
| 8         1234      134      | 7         5         134      | 9         1246      236      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 1346      134       8        | 25        9         134      | 1345      12456     7        |
| 134       5         2        | 134       7         6        | 1348      1489      389      |
| 7         9         1346     | 25        134       8        | 1345      12456     2356     |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|


The reader can see r5c1=2 and r9c3=6 the 2 extra digits.
I am not clear on what can be done here to select a puzzle of the file.
The number of guardians after “easy steps” is surely one possibility, but what else ???

For the 3.7M puzzles of mith’s file where the solution grid has the tridagon pattern, we have clearly 2 cases

One guardian: here the guardian can be assigned and rating the rest we have an idea of the interest of the puzzle,
More than one guardian: then, we are in the same situation as before.

Note ; with one cell having guardians, we can still have several extra candidates.


With 2 guardians, we have a classical simple ‘or’ (not exclusive) between the 2 extra candidates, something known for long in the AIC’s chains, so the cases with more than 2 guardians can be requested by the players.


After theses preliminary thoughts, what I intend to do is to clean files from “ easy after unique cell “ and to split files on the above criterias.
May be some have more precise ideas.

I’ll check the status of “2 TH” threats, but up to now, what I have seen is that only one of the 2 THs have a very small number of extra candidates.
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sat Apr 26, 2025 8:44 am

More thought about the selection of puzzles:

The “loki” pattern (TH impossible pattern with one extra candidate) is always here at the start in my search of new seeds “non degenerated tridagon”.
This is often not true.

Here below a puzzle morphed from mith’s file and the PM status after a first step in the brute force solver (a little more than singles)
Code: Select all
8......25......9..2...951.8..952..8......4..1..4.81.9..3......2..1......67....45.
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 8         9         367      | 13467     13467     367      | 367       2         5        |
| 13457     1456      3567     | 1234678   13467     23678    | 9         3467      367      |
| 2         46        367      | 3467      9         5        | 1         3467      8        |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 137       16        9        | 5         2         367      | 367       8         4        |
| 357       2568      235678   | 9         367       4        | 23567     367       1        |
| 357       256       4        | 367       8         1        | 23567     9         367      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 459       3         58       | 4678      4567      6789     | 678       1         2        |
| 459       2458      1        | 234678    34567     236789   | 3678      367       3679     |
| 6         7         28       | 1238      13        2389     | 4         5         39       |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|

As the rating is very high, we expect a TH. At that point, we can guess that it will be with the digits 357, but we have 2 possibilities, boxes 2356 or boxes 5689.

Applying easy solving steps (hidden and naked pairs), the “loki” pattern appears in the boxes 2356.
I followed the path in skfr, the second TH never reach the “loki” pattern, r9c9=9 is assigned with moves of a rating >10.

For a manual solver, the “loki” pattern could be rated at a level like a UR if the puzzle is morphed is the right way.
I see many examples with other morphs, but, in most cases, this is done with the help of a computer doing the “virtual morphing”.

Based on this, we could say that all Sudoku Explainer moves rating <=4.6 or similars should be done to show, if possible, the “loki” pattern.

If the solution grid does not have a type1 type2 pattern, the impossible pattern will have 2 or more extra candidates
Only very hard moves (we keep in mind that the final rating is very high) can solve it.

But we can have the chance to reach the minimal status (1 extra candidate for type1 type2, 2 extra candidates for the others) with moves of medium hardness (say in the range 6 to 8 with Sudoku Explainer for example).

With 2 extra candidates, we have the classical “or” that we can use in AICs; In Sudoku Explainer, from memory, such chains are rated around 9 (dynamic plus chains) and done with a limited set of patterns.

With more than 2 extra candidates, the game is open.

A player should not know what the status in the solution grid will be. Many eliminations leading to the “loki” status should be considered.
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sun Apr 27, 2025 8:56 am

I made a first test to extract puzzles where the TH is not "loki" after easy moves.

out of 50 000 puzzles having the TH pattern in the solution I got 964 with 2 or more guardians, but some with many guardians seem to be other kinds of 2 TH.

many with 2 guardians as these
Code: Select all
..1.......34...2..2.....6.51...5..687....15.28...2617.....87.....85.2..6...61.... 
951268437634975281287143695129754368763891542845326179516487923498532716372619854
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 569       56789     1        | 2         34679     34589    | 3479      3489      3479     |
| 569       3         4        | 19        679       589      | 2         189       179      |
| 2         789       79       | 1349      3479      3489     | 6         13489     5        |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 1         249       239      | 7         5         349      | 349       6         8        |
| 7         469       369      | 8         349       1        | 5         349       2        |
| 8         459       359      | 349       2         6        | 1         7         349      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 34569     12569     2569     | 349       8         7        | 349       25        1349     |
| 349       179       8        | 5         349       2        | 3479      1349      6        |
| 3459      2579      2579     | 6         1         349      | 8         25        3479     |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|



Code: Select all
..1......234.......6.52.......25..86.....72.9....9875..89.65.2..769....8...7..... 
951346872234879561768521943197254386845637219623198754389465127576912438412783695
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 5789      59        1        | 348       347       3469     | 34689     34679     2        |
| 2         3         4        | 18        17        169      | 15689     1679      15       |
| 789       6         78       | 5         2         1349     | 13489     13479     134      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 13479     149       37       | 2         5         134      | 134       8         6        |
| 134568    145       358      | 1346      134       7        | 2         134       9        |
| 1346      124       23       | 1346      9         8        | 7         5         134      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 134       8         9        | 134       6         5        | 134       2         7        |
| 1345      7         6        | 9         134       2        | 1345      134       8        |
| 1345      1245      235      | 7         8         134      | 69        69        1345     |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|



This one among a list with with 3 guardians
Code: Select all
..1.......34..2...67....5..9..86..57......2.67.....98.....78..2...2.6......59.67. 
281659734534712869679384521912863457358947216746125983165478392897236145423591678
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 258       2589      1        | 34679     3458      34579    | 3478      2346      3489     |
| 58        3         4        | 1679      158       2        | 178       16        189      |
| 6         7         289      | 1349      1348      1349     | 5         1234      13489    |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 9         124       23       | 8         6         134      | 134       5         7        |
| 13458     1458      358      | 79        1345      79       | 2         134       6        |
| 7         1456      356      | 134       2         1345     | 9         8         134      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 1345      14569     3569     | 134       7         8        | 134       1349      2        |
| 1348      1489      7        | 2         134       6        | 1348      1349      5        |
| 12348     1248      238      | 5         9         134      | 6         7         1348     |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|



The next easy step could be XYwing XYZwing or UR
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby totuan » Sun Apr 27, 2025 5:31 pm

champagne wrote:This one among a list with with 3 guardians
Code: Select all
..1.......34..2...67....5..9..86..57......2.67.....98.....78..2...2.6......59.67.

Code: Select all
 *--------------------------------------------------------------------*
 | 258    2589   1      | 34679  3458   34579  | 3478   2346   3489   |
 | 58     3      4      | 1679   158    2      | 178    16     189    |
 | 6      7     *28+9   | 1349   1348  #1349   | 5      1234   13489  |
 |----------------------+----------------------+----------------------|
 | 9      124   *23     | 8      6     #134    | 134    5      7      |
 | 13458  1458  *38+5   | 79     134-5  79     | 2      134    6      |
 | 7      1456   356    | 134    2     %1345   | 9      8      134    |
 |----------------------+----------------------+----------------------|
 | 1345   14569  3569   | 134    7      8      | 134    1349   2      |
 | 1348   1489   7      | 2      134    6      | 1348   1349   5      |
 | 12348  1248  *238    | 5      9     #134    | 6      7      1348   |
 *--------------------------------------------------------------------*

My path for this one:
DP(23589)r3459c3 *-marked cells => (5)r5c3=(9)r3c3
01: (5)r6c6=(134)r469c6-(134=9)r3c6-(9)r3c3==(5)r5c3 => r5c5<>5, r6c6=5
Code: Select all
 *--------------------------------------------------------------------*
 |%258    2589   1      | 34679  3458   3479   | 3478   2346   3489   |
 |%58     3      4      | 1679   158    2      | 178    16     189    |
 | 6      7      289    | 1349   1348   1349   | 5      1234   13489  |
 |----------------------+----------------------+----------------------|
 | 9      124    23     | 8      6     *134#   |*134#   5      7      |
 |*134+58 1458   358    | 79    *134#   79     | 2     *134#   6      |
 | 7      146    36     |*134#   2      5      | 9      8     *134#   |
 |----------------------+----------------------+----------------------|
 |%1345   14569  3569   |*134#   7      8      |*134#  *134+9  2      |
 | 1348   1489   7      | 2     *134#   6      | 1348  #134+9  5      |
 |*134+28 1248   238    | 5      9     *134#   | 6      7     *134+8# |
 *--------------------------------------------------------------------*

Tridagon(134) #-marked cells => (9)r8c8=(8)r9c9
Impossible pattern(134) *-marked cells => (9)r7c8=(258)r59c1=(8)r9c9

Combination Tridagon & Impossible pattern:
02: Present as diagram: => r9c9=8, then ER-6.6
Code: Select all
(8)r9c9*
 ||
(258)r1259c1-(5)r7c1=(134)r7c147-(134=9)r7c8-(9)r8c8==(8)r9c9*
 ||
(9)r7c8-(9)r8c8==(8)r9c9*

Impossible pattern(134) – not hard for proving:
Hidden Text: Show
Code: Select all
A=(1|3|4) => impossible
 *-----------------------------------------------------------*
 | .     .     .     | .     .     .     | .     .     .     |
 | .     .     .     | .     .     .     | .     .     .     |
 | .     .     .     | .     .     .     | .     .     .     |
 |-------------------+-------------------+-------------------|
 | .     .     .     | .     .     134   | 134   .     .     |
 | 134   .     .     | .     134   .     | .     134   .     |
 | .     .     .     | 134   .     .     | .     .     134   |
 |-------------------+-------------------+-------------------|
 | .     .     .     | 134   .     .     | 134   134   .     |
 | .     .     .     | .     134   .     | .     .     .     |
 |A134   .     .     | .     .     134   | .     .     134   |
 *-----------------------------------------------------------*

Some thing about tridagon puzzles – just my personel oppinion :D
01. I have checked many tridagon puzzles and it seems that ALL TRIDAGON PUZZLES can solve by manual players on using tridagon’s guardians/impossipbe patterns/RT (remote triple)/…
02. The difficuty of tridagon puzzles is not depend on numbers of guardians, I think about ER-8.9 to ER-9.1 based on SE rating.

Thanks for your puzzle!
totuan
totuan
 
Posts: 261
Joined: 25 May 2010
Location: vietnam

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sun Apr 27, 2025 6:09 pm

Hi "ttt"/"totuan",

I continue from time to time to fly over your paths.
Upon request of some players, I try to select puzzles of interest.

No idea of the interest, but my first test delivered these puzzles where the start leads quickly to 3 guardians.
Code: Select all
.12......3.4......67....51..9.81..56......7.1.6..7.98.....68.7......1..5...59.16.;;499;8
.12......3.4......67.5.1.....69....8.....71.5......96..8..75..6...8.97.1...16..8.;;500;8
.34...1..7.6......89.6.......526..13......4.6..3.4.52.....32.4.........1...15.63.;;501;8
.12......3.4......67....51..9.81..56......7.1.6..7.98.....68.7...6..1..5...59.1..;;502;8
.12......3.41.....67....5...8.9...56......7.1.6..7.89.....69.7...6..1..5...58.1..;;503;8
.57......23.4.5...8.9....5...21....6...2.35.4......1...6..34..2...6.13.....52..6.;;504;8
.12......3.4......67....51..9.87..56......1.7.6..1.98.....68..1..6..7......59.7..;;505;8
.31...52.7.2......89.........641..53......2.1..3.2.64.....34..2.....1......56.13.;;506;8
.425.6...7.6......89.....6....1...23...4.26.5......1....3.25..4..43.12.....64..3.;;508;8
.425.6...7.6......89..........1...23...4.26.5......1....3.25..4..43.12.6...64..3.;;509;8


I am curious to see what experts can do.
Last edited by champagne on Sun Apr 27, 2025 7:54 pm, edited 1 time in total.
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby totuan » Sun Apr 27, 2025 6:49 pm

champagne wrote:my first test delivered these puzzles where the start leads quickly to 3 guardians.
Code: Select all
12......3.4......67....51..9.81..56......7.1.6..7.98.....68.7......1..5...59.16.;;499;8
.12......3.4......67.5.1.....69....8.....71.5......96..8..75..6...8.97.1...16..8.;;500;8
.34...1..7.6......89.6.......526..13......4.6..3.4.52.....32.4.........1...15.63.;;501;8
.12......3.4......67....51..9.81..56......7.1.6..7.98.....68.7...6..1..5...59.1..;;502;8
.12......3.41.....67....5...8.9...56......7.1.6..7.89.....69.7...6..1..5...58.1..;;503;8
.57......23.4.5...8.9....5...21....6...2.35.4......1...6..34..2...6.13.....52..6.;;504;8
.12......3.4......67....51..9.87..56......1.7.6..1.98.....68..1..6..7......59.7..;;505;8
.31...52.7.2......89.........641..53......2.1..3.2.64.....34..2.....1......56.13.;;506;8
.425.6...7.6......89.....6....1...23...4.26.5......1....3.25..4..43.12.....64..3.;;508;8
.425.6...7.6......89..........1...23...4.26.5......1....3.25..4..43.12.6...64..3.;;509;8

I am curious to see what experts can do.

The first on error, the rest: after reduce to 2 guardians then same as my solution above.

Thanks for your puzzles!
totuan
totuan
 
Posts: 261
Joined: 25 May 2010
Location: vietnam

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sun Apr 27, 2025 7:59 pm

this is the output for the first 50 000 puzzles of mith's file.
It could be that all are coming from very close solution grids, explaining your findings.

I have to do more tests and run a wider test, this will come this week.
We will see if your remark remains true.
and I hope to come with more challenging puzzles
:D :D
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby marek stefanik » Sun Apr 27, 2025 11:45 pm

Note that even puzzles with just one guardian cell can be interesting.
There are a few situation which may arise after TH has been used:

If the only guardian is at the rectangle, the puzzles are generally not very difficult to solve using the RTs, even if they rate 10+ post-TH.

In the direct search thread, I noticed that puzzles like this one are very common:
3.4.....5.....5..7....1..6....12..79.2.7.95.6..9.5621.....61.9..9.5.27.....97.65.;11.4/11.0/2.6
10.9 skfr post-TH, the highest rating I've seen.
Each minicolumn in b9 contains one of 348, which unlocks some tricks with columns, ultimately giving you 8r1c6, 8r3c9 and reducing the puzzle to 4.2.
Hidden Text: Show
Code: Select all
,---------------------,-------------------,-------------------,
| 3      1678   4     | 268    89    78   | 189    28    5    |
| 12689  168    1268  | 23468  3489  5    | 13489  2348  7    |
| 25789  578    2578  | 2348   1     3478 | 3489   6     348  |
:---------------------+-------------------+-------------------:
| 4568   34568  3568  | 1      2     348  | 348    7     9    |
| 148    2      138   | 7      348   9    | 5      348   6    |
| 478    3478   9     | 348    5     6    | 2      1     348  |
:---------------------+-------------------+-------------------:
| 24578  34578  23578 |y348    6     1    |z348    9    x2348 |
| 1468   9      1368  | 5     z348   2    | 7     y348  x1348 |
| 1248   1348   1238  | 9      7    x348  | 6      5     12   |
'---------------------'-------------------'-------------------'
The 3|4|8 in b8c6 appears in b9c9 (otherwise we could shift the b9c9 digit into r9c9 and get a solution to TH).
The 3|4|8 in b8c4 must then appear in b9c8 and the one in b8c5 must appear in b9c7.
Each of c456789 has one of 348 in each minicolumn. Columns like this which lie in the same stack must have the same permutation parity.
If both stacks have the same parity, the 3|4|8 in b2c6 must appear in b3c6 (same last digit => same permutation), so 8r1c6, 8r3c9.
If they have different parities, the 3|4|8 in b2c5 must appear in b3c8 (different 2nd and 3rd digit => same 1st), so it must be 8.
This lets us eliminate several 8s (r1c2, b2p1479, b3p147).
We can also take each case a bit further and get 16r1c47:
same parity: 8r1c6 forces them via r1
opposite parity: 34r2 have r2c47 left, forcing 16 into r1
After that, we get 8r1c6 with singles, which forces the same stack parity and 8r3c9, 4.2 skfr.
Alternatively, we could have just broken the opposite parity case:
(8–3|4)b2c5&b3c8 = (34–1|6)r2c47 = (16–2|9)r1c47 = (29–8)r1c58 => same parity


Recently, I was happy to see this puzzle (10.3 post-TH):
.42.......7...8..35...6....7..65..31...3.18.7....8765...7.35.18.1...67........3..;11.2/2.0/2.0 # 5
Either the minirows or the minicolumns in b9 have this property, but it is not yet clear, which is the case (this reminds me of Frailty).
It is somewhat straightforward to eliminate 249r8c9 with column tricks, reducing the puzzle to 7.1.
Hidden Text: Show
After (3=9)r1c6 – (9=6)r1c9 – (6=5)r9c9 – 5r8c9 = (5–3)r8c3 = 3r8c1 => –3r1c1:
Code: Select all
,-------------------,------------------,-------------------,
| 1689   4     2    | 1579   179   3   | 159    78    69   |
| 169    7     169  | 12459  1249  8   | 12459  2469  3    |
| 5      389   1389 | 12479  6     249 | 1249   78    249  |
:-------------------+------------------+-------------------:
| 7      289   489  | 6      5     249 | 249    3     1    |
| 2469   2569  4569 | 3      249   1   | 8      249   7    |
| 12349  239   1349 | 249    8     7   | 6      5     249  |
:-------------------+------------------+-------------------:
| 2469   269   7    |y249    3     5   |z249    1     8    |
| 2349   1     3459 | 8     z249   6   | 7     y249  x249+5|
| 2489   2589  4589 | 17     17   x249 | 3      2469  56   |
'-------------------'------------------'-------------------'
Suppose 249r8c9. We get the same entanglement for b89 as in the other puzzle.
Look at 249c69. The first digit is different, but the last digit is the same => opposite parity.
The first 2|4|9 in c58 then must be the same, so 9, but that leaves 245r2 with just r2c47, i.e. contra.
So 5r8c9, 7.1 skfr.


Out of the puzzles I tried in the direct search thread, this was the only one which I couldn't find a decent path for:
5....1....3.8......4..6.....2.....984.....1.2..9...46...4.86.2181.2.49.6...19....;11.3/11.3/3.6
10.5 skfr post-TH, there are a couple more easy eliminations: 57r4c3, 5r9c2, but it's still 9.3 afterwards. 9.2 after –6r5c2, which requires a y-wing in the same parity case.
Assuming opposite band parity gives you 5r5c2, 5r8c3, btte, but it is unclear why same parity breaks.

Marek
marek stefanik
 
Posts: 385
Joined: 05 May 2021

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Mon Apr 28, 2025 6:19 am

Hi Marek,
nice examples of a pattern easy to extract that I could have missed.
It's true that in my direct search I have no constraint on the cell with extra candidates . This can explain why this comes so easily with the high ratings.
I'll see what happen in mith's file

EDIT : I made a quick test on mith's file (first 10000 items)
checking the status after locked sets and fishes, I got 306 examples with marek's pattern.
Likely most cases will vanish using harder moves
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sun May 04, 2025 5:47 am

I made a first test on mith's file with a code morphing the puzzles to get the
Code: Select all
..x x..
.x. .x.
x.. ..x

x.. x..
.x. .x.
..x ..x

pattern in the boxes 5689 and clear "easy puzzles" in the following conditions

the puzzle after solving steps (not t&E1) has only one extra candidate
Assigning this candidate, the puzzle is solve at the T&E1 level

We know that this left us the ~870000 puzzles where the solution grid has not the "loki pattern".
I got 906485 puzzles, so most of the puzzles with the "loki pattern" in the solution grid are discarded.

This has to be carefully checked.

I posted one puzzle appearing with many residual guardians in my run.

Next is to see what to do with this file, still very big

EDIT : small adjustment if only one cell has extra candidates, the relevant action is done (clear the digits of the triplet), so if the puzzle is easily solve with more than one extra candidate in the cell, the puzzle is discarded
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany

Re: Selecting puzzles of interest with a tridagon threat.

Postby champagne » Sun May 04, 2025 3:31 pm

After "marek stefanik"'s remark, i made another test on my current file of 202614 high skfr ratings (>=10.5 ) given by the direct search of non degenerated tridagons.

due to the design such puzzles have only one cell with extra candidates.

Clearing the impossible candidates and trying to solve the puzzle at the T&E(1) level, I got 35% of unsolved puzzles.
Keeping only puzzles with more than one extra candidate in the cell, I got 13% of unsolved puzzles.

And I found a significant number of puzzles with 4 extra candidates as these
Code: Select all
  ...21..6..42.......3...5.....612..79...7.96.57...5621.9...71.5...56.29.....59.... 
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 58        5789      789      | 2         1         3478     | 34578     6         3478     |
| 1568      4         2        | 389       368       378      | 15        89        1378     |
| 168       3         178      | 489       468       5        | 1478      289       12478    |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 3458      58        6        | 1         2         348      | 348       7         9        |
| 12348     128       1348     | 7         348       9        | 6         348       5        |
| 7         89        3489     | 348       5         6        | 2         1         348      |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|
| 9         26        348      | 348       7         1        | 348       5         26       |
| 1348      178       5        | 6         348       2        | 9         348       178      |
| 12348     12678     13478    | 5         9         348      | 13478     2348      1267     |
|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|++++++++++++++++++++++++++++++|

Code: Select all
2.73........5.....3.8..9....8.13..45...4.59.8..5.9831.8...41.5.51.9.38.....85....
9.....5....6.8.1..2....5.3.1.356..87...8.73.1.8..3165.....58.1...13.67.....71....
.7...1...9..5.....8.47...2..2.97..567.96.51.2....1279.....59.6.5..1.72.....26....
champagne
2017 Supporter
 
Posts: 7601
Joined: 02 August 2007
Location: France Brittany


Return to Help with puzzles and solving techniques