One of the most basic fuzzy set systems use min and max for
and and or of pairs of membership grades which each must
be between 0 and 1. Other operators are used in other fuzzy
systems.

Three well-studied triangular norm (t_norm) systems are Kleene-Dienes,
Reichenbach, and Lukasiewicz.

The t_norm operator for each system is its "and" operator

NB. t_norms  Kleene-Dienes, Reichenbach, Lukasiewicz
KD    =: <.
reich =: *
lukas =: (0 >. 1 -~ +)"0

NB. sample and tables
NB. KD    /~ 0 0.25  0.5 0.75 1
NB. reich /~ 0 0.25  0.5 0.75 1
NB. lukas /~ 0 0.25  0.5 0.75 1


NB. With the deMorgan "not", the t_conorm for each system can be
defined as follows using the adverb S .

not =: -.
S =: 1 : 'n...@u& not'   NB. t_conorm
S =: & not(not@)        NB. t_conorm

NB. sample or tables
NB. KD    S/~ 0 0.25  0.5 0.75 1
NB. reich S/~ 0 0.25  0.5 0.75 1
NB. lukas S/~ 0 0.25  0.5 0.75 1

NB. next shows that t_conorm for KD is >.
NB. (>./-:KD S/)~ 0 0.25  0.5 0.75 1  NB. KD t_conorm is >.

NB. and next 2 show similar equivalences
NB. ((+- *)"0/-:reich S/)~ 0 0.25  0.5 0.75 1  NB. reich t_conorm is  (+-*)"0
NB. ((1<.+)"0/-:lukas S/)~ 0 0.25  0.5 0.75 1  NB. lukas t_conorm is (1<.+)"0

NB. material implication for these systems can be generated as follows

I =: 1 : 'n...@[ u S ]'   NB. I  is the implication operator
NB. KD    I"0/~0 0.25  0.5 0.75 1
NB. reich I"0/~0 0.25  0.5 0.75 1
NB. lukas I"0/~0 0.25  0.5 0.75 1

NB. next shows the equivalent implication operator for KD is -...@[>.]"0
NB. ((   -...@[>.])-:KD    I"0)/~0 0.25  0.5 0.75 1
NB. ((   -...@[+ *)-:reich I"0)/~0 0.25  0.5 0.75 1
NB. ((1<....@[+ ])-:lukas I"0)/~0 0.25  0.5 0.75 1






On Wed, Mar 4, 2009 at 4:28 PM, Raul Miller <[email protected]> wrote:
> On Wed, Mar 4, 2009 at 4:07 PM, Roger Hui <[email protected]> wrote:
>> It is by definition that 0=x*.y if x=0 or y=0.
>> Then the advantages are that *. is the same
>> as "and" for booleans and that the identity
>> (x*.y)=x(*%+.)y holds for x=0 or y=0 .
>
> Is it worth noting that min and max (<. and >.)
> could likewise serve as and and or, without
> needing any special case treatment?
>
> --
> Raul
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>



-- 
(B=) <-----my sig
Brian Schott
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to