Anti-knight Ferz-non-consecutive

For fans of Killer Sudoku, Samurai Sudoku and other variants

Anti-knight Ferz-non-consecutive

Postby primeprover » Wed Jan 08, 2020 11:54 am

I was fiddling around with JSudoku and ended up finding this variant. Unfortunately, I am having no luck in solving it at all. Is anyone skilled enough to solve this puzzle without branching? Could you show what I have missed?

Puzzle rules:

1. Normal sudoku rules apply
2. Anti-knight - no number can be a knights move away from itself.
3. Ferz-non-consecutive - no consecutive numbers can be diagonally adjacent.

Code: Select all
..9|...|7..
...|...|...
7..|...|..9
---+---+---
...|.8.|...
...|6.2|...
...|.4.|...
---+---+---
3..|...|..4
...|...|...
..5|...|8..


As a side note: Is 12 clues the minimum number of clues for this variant?
primeprover
 
Posts: 4
Joined: 21 December 2015

Re: Anti-knight Ferz-non-consecutive

Postby tarek » Wed Jan 08, 2020 12:52 pm

I will be able to give you a comprehensive answer in 48h

Watch this space!

Tarek
User avatar
tarek
 
Posts: 3362
Joined: 05 January 2006

Re: Anti-knight Ferz-non-consecutive

Postby creint » Wed Jan 08, 2020 6:32 pm

JSudoku won't give me a solution, it said multiple solutions.

But my solver gave single solution:
Hidden Text: Show
Code: Select all
2 8 9 1 3 4 7 6 5
1 5 4 9 6 7 2 3 8
7 6 3 2 5 8 1 4 9
9 7 6 3 8 1 4 5 2
5 4 1 6 7 2 9 8 3
8 3 2 5 4 9 6 7 1
3 2 7 8 9 6 5 1 4
4 9 8 7 1 5 3 2 6
6 1 5 4 2 3 8 9 7


Required 4 levels of chains/nets. Hard but not that hard. Rating should be around SE 9.0 or lower.
creint
 
Posts: 147
Joined: 20 January 2018

Re: Anti-knight Ferz-non-consecutive

Postby tarek » Thu Jan 09, 2020 2:38 am

primeprover wrote:I was fiddling around with JSudoku and ended up finding this variant. Unfortunately, I am having no luck in solving it at all. Is anyone skilled enough to solve this puzzle without branching? Could you show what I have missed?

This is an extremely tough puzzle. I modified Sukaku Explainer to support Ferz NC and with the set of techniques available it couldn't solve it (the solver still needs some tweaking but overall I think still a difficult puzzle). I confirm the unique solution that creint has also mentioned.

Solvable is the following:
Code: Select all
.........
...5.....
.....65..
.9....1..
.........
..5....9.
..12.....
.....1...
.........


primeprover wrote:As a side note: Is 12 clues the minimum number of clues for this variant?
No.
Code: Select all
.........
.........
.........
......3..
.........
..6......
...9.....
.......7.
.........
User avatar
tarek
 
Posts: 3362
Joined: 05 January 2006

Re: Anti-knight Ferz-non-consecutive

Postby primeprover » Fri Jan 10, 2020 11:08 am

Thanks for having a look. It looks like this puzzle is hard enough that an extreme amount of effort is needed to find a solution. This is beyond a lot of solvers(without recursion) including humans.

JSudoku gave me the unique solution after recursively solving.
primeprover
 
Posts: 4
Joined: 21 December 2015

Re: Anti-knight Ferz-non-consecutive

Postby tarek » Fri Jan 10, 2020 7:49 pm

I generated a few & rated them with an un-released version of sukaku explainer:
Here is one that is solvable & SE rates it at ER=3.2 (X-Wing)
Code: Select all
...3.....
.....9...
.....6...
4.....67.
.........
.58.....1
...6.....
...9.....
.....7...
User avatar
tarek
 
Posts: 3362
Joined: 05 January 2006

Re: Anti-knight Ferz-non-consecutive

Postby HATMAN » Wed Jan 15, 2020 9:23 pm

Nice puzzle

Early on I used a couple of FNC/AN interactions that I thought neat and hoped to avoid the x-wing, but as I plodded on there it was.

I remember making quite a few of these early on (probably as killers) as I found NC/AN too restrained.
HATMAN
 
Posts: 254
Joined: 25 February 2006
Location: Nigeria

Re: Anti-knight Ferz-non-consecutive

Postby tarek » Thu Jan 16, 2020 11:28 am

HATMAN wrote:Nice puzzle

Early on I used a couple of FNC/AN interactions that I thought neat and hoped to avoid the x-wing, but as I plodded on there it was.

I remember making quite a few of these early on (probably as killers) as I found NC/AN too restrained.

thanks for the feedback, I'll post a few of these on the forbidden pairs thread in case there is an interest

tarek
User avatar
tarek
 
Posts: 3362
Joined: 05 January 2006

Re: AN FNC+ ER=3.6

Postby tarek » Sat Jan 18, 2020 7:50 pm

Another fun to solve puzzle
This is an AN FNC+ (Ferz Non consecutive including 1,9 and Anti-kNight)
Code: Select all
.........
...9.5...
.........
.5.....1.
.........
.2.....5.
.........
...2.7...
.........
User avatar
tarek
 
Posts: 3362
Joined: 05 January 2006


Return to Sudoku variants