Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.java.programmer > #53751

Make the Sudoku solver faster [GigaLIPS via NPU] (Re: Lean CLP for Dogelog Player)

Show key headers only | View raw

Hi,

Instead of plumbing an OpenCL to image recognize(*)
a Sudoku game. Make the Sudoku solver faster,
by tapping into Local AI. Can SWI-Prolog do that?

I doubt so. You could tap into GigaLIPS I speculate.

Bye

(*) The only use of local AI by deepclause. Thats
quite sad to see, that the Prolog community has
become sleepy joe concerning infererencing and Local AI.

Mild Shock schrieb:
 > Hi,
 >
 > Icelandish Idiot payed Putins Troll,
 > you got a micro penis. You are probably
 > the author of ths nonsense:
 >
 > https://github.com/deepclause/deepclause-desktop
 >
 > Go fuck yourself.
 >
 > Bye
 >
 > Gilmeh Serda schrieb:
 >> On Fri, 28 Nov 2025 14:20:59 +0100, Mild Shock wrote:
 >>
 >>> SWI-Prolog and Trealla Prolog.
 >>>
 >>> Bye
 >>
 >> True.
 >>
 >> *PLONK*
 >>

Mild Shock schrieb:
> 
> Hi,
> 
> Many existing and evolving constraint logic
> programming projects resemble some ancient
> invention of gunpowder. For example SWI-Prologs
> 9.3.35 corouting for delayed goals is mainly
> based on unify hooks. We show how verify hooks,
> already used in formerly Jekejeke Prolog, can
> be braught to Dogelog Player in a 100% Prolog fashion.
> 
> Since the recent version of Dogelog Player
> supports cyclic terms, we could let the Jini
> out of the bottle, and provide the experimental
> library(edge/railgun) to model delayed goals with
> nothing else than Alain Colmerauers rational trees.
> The result is a Lean CLP of ca. 100 lines of code,
> that already provides a simple constraint (#\=)/2
> and a global constraint all_different/1.
> 
> The results are encouraging. For problems that
> are not over constrained, Dogelog Player leaves
> existing Prolog systems clearly behind, showing
> a 2-3x times speed-up against SWI-Prolog and a
> 20-30x times speed-up against Trealla Prolog.
> For more constrained problems we suggest ommiting
> forward checking in favor of a form of ahead of
> time (AOT) variable ordering. With this approach
> and for Sudoku problems we are then in the midfield
> between SWI-Prolog and Trealla Prolog.
> 
> Bye
> 
> See also:
> 
> Lean CLP for Dogelog Player
> https://qiita.com/j4n_bur53/items/addf1fc86856dd682dcb
> 
> Mild Shock schrieb:
>> Dear All,
>>
>> We are happy to announce a new edition
>> of the Dogelog Player:
>>
>> - Dot Notation:
>> We borrowed from formerly Jekejeke Prolog
>> optionable custom binary forms. An operator
>> declaration such as op(100, yfx, '.') does
>> now parse and unparse the dot notation A.B
>> as '$DOT'(A,B). Avoiding a functor clash
>> with the consing notation [A|B]. The
>> implementation is 100% Prolog, and thus
>> uniformly available for the JavaScript,
>> Python and Java target.
>>
>> - Arrow Notation:
>> We borrowed from formerly Jekejeke Prolog
>> optionable custom variadic forms. An operator
>> declaration such as op(700, xfx, '=>') does
>> now parse and unparse the arrow notation
>> (P => G) as '$ANON'(H,K,B,...). Compared to
>> library(lambda), it leads to more efficient
>> single step reductions inside the new
>> library(arrow). Again uniformly available
>> for the JavaScript, Python and Java target.
>>
>> - DCG phrase/3:
>> This was a gap that slipped our attention.
>> It turns out that Dogelog Player didn't have
>> DCG phrase/3, although it had already
>> DCG (-->)/2. Luckily the step from (-->)/2
>> to phrase/3 is relatively small and this
>> release features this predicate. Turns out it
>> can be used for Strudel style music coding.
>>
>> Have Fun!
>>
>> Jan Burse, https://www.herbrand.ai/ , 24.11.2025
> 

Back to comp.lang.java.programmer | Previous | Next | Find similar

Thread

Make the Sudoku solver faster [GigaLIPS via NPU] (Re: Lean CLP for Dogelog Player) Mild Shock <janburse@fastmail.fm> - 2025-12-01 11:38 +0100

csiph-web