- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Wolfgang Lux <lux_at_wi.uni-muenster.de>

Date: Mon, 14 Jan 2002 13:32:49 +0100

Hello Michael,

following up my previous mail, I should also note that I don't think option

2 is that problematic as you say. However, one has to be careful about the

definition of search strategies.

*> Option 2 has the disadvantage that we cannot really encapsulate
*

*> non-deterministic computations if the goals are defined outside
*

*> and passed by pattern variables into a search goal. For instance,
*

*> consider the following definition of "negation as failure":
*

*>
*

*> naf c = (findall \_->c) =:= []
*

*>
*

*> Since the constraint c is passed into the findall via the
*

*> binding of the non-local variable c, its evaluation will always
*

*> cause a global non-deterministic splitting of the computation space.
*

It is true, that *this* definition will not work under option 2. But then I

think it is bold to assume that negation-as-failure could be implemented in

that way! What really can be done with the encapsulated search in Curry is to

ask whether there exists no solution for a predicate p, i.e. \not\exists x.p(x).

But this means, p has to have an argument. Therefore, the only way

negation-as-failure can be implemented in Curry is by defining a function

like:

noSolution g = findall g =:= []

And if you want to see whether g is not satisfied for some value x,

you can use the following function:

notSatisfiedFor g x = findall (\y -> x=:=y & g y) =:= []

*> This means that most search operators are no longer applicable.
*

I'm not convinced. Do you have any other examples? The only problem I see

at the moment is to implement committed choice (at least in the way it is

specified in the appendix of the Curry report).

Regards

Wolfgang

Date: Mon, 14 Jan 2002 13:32:49 +0100

Hello Michael,

following up my previous mail, I should also note that I don't think option

2 is that problematic as you say. However, one has to be careful about the

definition of search strategies.

It is true, that *this* definition will not work under option 2. But then I

think it is bold to assume that negation-as-failure could be implemented in

that way! What really can be done with the encapsulated search in Curry is to

ask whether there exists no solution for a predicate p, i.e. \not\exists x.p(x).

But this means, p has to have an argument. Therefore, the only way

negation-as-failure can be implemented in Curry is by defining a function

like:

noSolution g = findall g =:= []

And if you want to see whether g is not satisfied for some value x,

you can use the following function:

notSatisfiedFor g x = findall (\y -> x=:=y & g y) =:= []

I'm not convinced. Do you have any other examples? The only problem I see

at the moment is to implement committed choice (at least in the way it is

specified in the appendix of the Curry report).

Regards

Wolfgang

-- Wolfgang Lux Phone: +49-251-83-38263 Institut fuer Wirtschaftinformatik FAX: +49-251-83-38259 Universitaet Muenster Email: wlux_at_uni-muenster.deReceived on Di Jan 15 2002 - 06:46:15 CET

*
This archive was generated by hypermail 2.3.0
: Do Jan 21 2021 - 07:15:03 CET
*