I think you're going to need to dive some more into the math of the
permutations here.

I have not done that myself. However, I suspect that you would need
four kinds of "constraint rectangles" here, to fully constrain the
possibilities using Hui's approach. But Sudoku only provides three
(rows, columns, blocks).

Specifically, I think R needs to have shape 64 16, instead of 48 16.

That said, this is just a hunch - I haven't analyzed this in any depth.

Good luck,

-- 
Raul

On Sat, Jun 18, 2022 at 11:14 AM Brian Schott <schott.br...@gmail.com> wrote:
>
> I have been trying to adapt Roger Hui's sudoku solver
> https://code.jsoftware.com/wiki/Essays/Sudoku to a 4x4 grid from his 3x3
> grid. I actually do not want such an automatic solution, and prefer a
> solution which helps a human with some of the tedium, but in the meantime,
> I think I would like an automatic solution as well.
>
> My problem seems to be that my adaptation does not screen adequately for
> multiple identical "candidates". By that I mean in the following example,
> for example, the "candidate" 3 appears three times, (by coincidence), as a
> "solution" produced by the repeated application of the verb 'f', and in
> particular, here, in the second update of the grid. To see what I mean,
> execute the following commands on the attached script.
>
>    load'~user/sudoku16forum.txt'
>    9!:7'+++++++++|-'
>    f=: + (ac >. ar)@free          NB. one step of assign
>    $,.see t=: f^:a: x                NB. forced moves leading from grid x
>    3{.see diff t
>
> In particular, look at the following result which shows 3 boxes (which I
> hope are not wrapped by your email client, but if it is wrapped, you might
> just look at 1 2&{ instead of 3&{. ).
> In the third box, focus on the bottom righthand 4x4 region which contains
> the characters E330 and '.'s . In that region itself there are two 3's,
> which should not happen. But in addition each of those 3's collide with
> external 3's in their own row or column, but outside the 4x4 region we are
> focusing on. [I have attempted to produce the 3's below in *bold* type.]
> I can't tell if this problem is a result of something I have done wrong in
> my adaptation or if there is more coding required for 4x4 relative to 3x3,
> or what.
>
>
>
>    3{.see diff t
> +---------------------+---------------------+---------------------+
> |+----+----+----+----+|+----+----+----+----+|+----+----+----+----+|
> ||5E..|2.1.|.F6.|039.|||....|....|....|...C|||....|....|....|....||
> ||....|5.D.|0E..|....|||....|....|....|.2..|||....|....|....|...1||
> ||.28E|6F3.|4...|B...|||....|...0|....|.D..|||....|....|....|....||
> ||.F.0|....|D...|.48.|||....|....|....|E...|||....|..A.|..2.|....||
> |+----+----+----+----+|+----+----+----+----+|+----+----+----+----+|
> ||.5C.|.1.3|2...|A9BD|||....|....|..0E|....|||....|....|....|....||
> ||..A.|4...|1...|50..|||....|....|....|...2|||....|...6|....|..E.||
> ||1...|...E|..F2|...F|||....|....|9...|4C..|||....|....|....|..*3*.||
> ||.9.6|A.2.|..D.|7816|||..E.|....|....|....|||0...|....|...4|....||
> |+----+----+----+----+|+----+----+----+----+|+----+----+----+----+|
> ||D1.8|...C|.49.|3.2.|||....|....|....|...0|||....|.E..|....|.6..||
> ||..09|..C.|....|.5..|||.6..|.A..|...D|....|||....|....|E...|....||
> ||73..|.2..|.A.6|..DE|||....|....|....|....|||....|..8.|....|....||
> ||AC..|.6B.|5..0|9..A|||..E.|....|....|....|||....|....|....|....||
> |+----+----+----+----+|+----+----+----+----+|+----+----+----+----+|
> ||3...|.9..|..7.|D...|||.0..|...2|....|....|||....|....|....|.E..||
> ||.4..|.0.1|6DE9|2...|||....|....|....|....|||8.5.|*3*...|....|...*3*||
> ||....|E7.8|.0..|...7|||....|....|....|....|||..2.|....|....|..*3*.||
> ||EB.F|8..A|.2..|C1..|||..7.|....|3...|...9|||....|....|....|..0.||
> |+----+----+----+----+|+----+----+----+----+|+----+----+----+----+|
> +---------------------+---------------------+---------------------+
>
>
> --
> (B=) <-----my sig
> Brian Schott
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to