Blossom Loop

Advanced methods and approaches for solving Sudoku puzzles

Blossom Loop

Postby yzfwsf » Thu Dec 07, 2023 7:05 am

My solver has implemented a technique named Blossom Loop by Chinese Sudoku enthusiasts.
Image
Cell Type Blossom Loop: 168r7c9 => r2c8 <> 2, r3c1,r6c3 <> 4, r8c9 <> 6, r8c9 <> 8
1r7c9 - r7c3 = r9c1 - (1=4)r4c1 - r6c2 = r3c2 - r2c13 = (4-6)r2c8 = r2c9 - 6r7c9
8r7c9 - r78c8 = 8r2c8

Image
Region Type Blossom Loop: 1r6c1279 => b7p125 <> 1, r6c9 <> 2, r27c8 <> 3, r9c3 <> 7
1r6c1 - (1=7)r9c1 - (7=3)r9c8 - r46c8 = (3-1)r6c7
1r6c2 - r1c2 = 1r12c1
(1-7)r6c9 = 7r9c9
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby yzfwsf » Thu Dec 07, 2023 8:06 am

A cell has three or more candidate numbers. Let's assume that the cell only has two candidate numbers, pretend that the other candidate numbers don't exist, and then find a loop. Next, we will find a branch for each previously assumed non-existent number of candidates. If the endpoints of the branch are all located in a weak region of the initial loop or another weak region of the branch, and the truths do not overlap, then we will find a zero rank net.The implementation principle of region type is the same, but the RC space has become N/R/C/B space, and the search method is the same.
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby StrmCkr » Sat Dec 09, 2023 10:33 am

Disjointed Disjointed subsets using ahs degrees of freedom ,

instead of als degrees of freedom.

Can't tell, as the pictures are cut off.
Some do, some teach, the rest look it up.
stormdoku
User avatar
StrmCkr
 
Posts: 1427
Joined: 05 September 2006

Re: Blossom Loop

Postby yzfwsf » Sat Dec 09, 2023 11:22 am

StrmCkr wrote:Disjointed Disjointed subsets using ahs degrees of freedom , instead of als degrees of freedom.

Maybe not, in fact, this technique belongs to the composite of chains.
StrmCkr wrote:Can't tell, as the pictures are cut off.


Can't you open the image in the new tab?The embedded image seems to only display thumbnails.
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 4:29 am

I'm trying to implement this in my solver.

Why are you eliminating 1r78c2 in the second example but not 8r7c456 in the first example ?
MAGREMENT
 
Posts: 62
Joined: 20 October 2023

Re: Blossom Loop

Postby yzfwsf » Mon Dec 11, 2023 4:46 am

MAGREMENT wrote:I'm trying to implement this in my solver.

Why are you eliminating 1r78c2 in the second example but not 8r7c456 in the first example ?

Because 8r7 is not a weak link, and 8b9 is that weak link.
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 6:26 am

yzfwsf wrote:
MAGREMENT wrote:I'm trying to implement this in my solver.

Why are you eliminating 1r78c2 in the second example but not 8r7c456 in the first example ?

Because 8r7 is not a weak link, and 8b9 is that weak link.


Oh right it is a pointing column my bad
MAGREMENT
 
Posts: 62
Joined: 20 October 2023

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 6:47 am

I'm gonna need some more help lmao

On this grid :

Code: Select all
+--------------+--------------+--------------+
|<6>  <2>  <3> |<4>  59   189 |15   <7>  158 |
|49   <1>  49  |25   <7>  28  |<3>  <6>  258 |
|<8>  <7>  <5> |123  <6>  123 |14   124  <9> |
+--------------+--------------+--------------+
|<1>  <6>  <2> |<9>  34   34  |<8>  <5>  <7> |
|79   <3>  789 |158  258  17  |<6>  129  <4> |
|<5>  <4>  789 |178  28   <6> |19   <3>  12  |
+--------------+--------------+--------------+
|<3>  <5>  <1> |78   489  79  |<2>  49   <6> |
|247  <9>  47  |<6>  <1>  234 |457  <8>  35  |
|247  <8>  <6> |23   349  <5> |1479 149  13  |
+--------------+--------------+--------------+


My solver finds :

Loop : 8r1c6 - 8r1c9 = 8r2c9 - 8r2c6 = 2r2c6 - 2r2c4 = 5r2c4 - 5r2c9 = PR : 5r1c79 - 5r1c5 = 9r1c5 - 9r1c6 =
Branches :
1r1c6 - 1r5c6 = 7r5c6 - 7r7c6 = 7r7c4 - 8r7c4 = 8r7c5 - 8r6c5 = 2r6c5 - 2r6c9 = 2r2c9

=> r3c4 <> 2, r3c6 <> 2

Which are faulty eliminations and I can't understand why.

2r2c9 is in the weak region 2r2c6 - 2r2c4 of the loop and the loop and the branch dont seem to overlap

Could you give me some help ?

Image

Image
MAGREMENT
 
Posts: 62
Joined: 20 October 2023

Re: Blossom Loop

Postby yzfwsf » Mon Dec 11, 2023 7:07 am

Your loop can only eliminate 1r3c6 and 8r5c5.
The end of branch(2r2c9 ) must be included in the weak region.

ALS Continuous Nice Loop: (1=235)r239c4 - (5=17892)r5c13468 - (2=41)r3c78 => r3c6<>1,r5c5<>8
No need for Blossom Loop.
Last edited by yzfwsf on Mon Dec 11, 2023 7:19 am, edited 1 time in total.
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 7:18 am

yzfwsf wrote:The end of branch(2r2c9 ) must be included in the weak region.


Oh okay I didn't understand that. Thank you !

yzfwsf wrote:ALS Continuous Nice Loop: (1=235)r239c4 - (5=17892)r5c13468 - (2=41)r3c78 => r3c6<>1,r5c5<>8
No need for Blossom Loop.


I know. I'm testing the strategy so I'm disabling others
MAGREMENT
 
Posts: 62
Joined: 20 October 2023

Re: Blossom Loop

Postby yzfwsf » Mon Dec 11, 2023 8:38 am

The difficulty of this technique is comparable to Forcing Chain, but in most cases, it has more deletions than regular Forcing Chain, but also has more limitations, so it is more appropriate to call it between AIC and Forcing Chain.If you call it after the Forcing Chain, there will be no result!
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 9:39 am

yzfwsf wrote:it is more appropriate to call it between AIC and Forcing Chain


That's the conclusion I also came to
MAGREMENT
 
Posts: 62
Joined: 20 October 2023

Re: Blossom Loop

Postby yzfwsf » Mon Dec 11, 2023 10:34 am

MAGREMENT wrote:
yzfwsf wrote:it is more appropriate to call it between AIC and Forcing Chain


That's the conclusion I also came to

Your implementation cannot find the Blossom Loop in the two Rank 0 challenge puzzles.
yzfwsf
 
Posts: 854
Joined: 16 April 2019

Re: Blossom Loop

Postby MAGREMENT » Mon Dec 11, 2023 12:08 pm

yzfwsf wrote:Your implementation cannot find the Blossom Loop in the two Rank 0 challenge puzzles.


Do you mean the two at the start of this thread ? If not, can you give me the grids so I can correct that ?
MAGREMENT
 
Posts: 62
Joined: 20 October 2023


Next

Return to Advanced solving techniques

cron