The standard answer to "Does J have a dictionary data type?" seems to be
"Use symbols."
(Two other answers today recommend locales.) To which we ask, "How do
we use symbols?" Given that I may have overlooked a lab, I'll show my
effort. My goal is to maintain a database of probabilities written in a
standard long form. "P(a)" is an invalid j pronoun.
I solve this with string keys mapped to symbols. J implements symbols
with red-black trees. They're quickly accessed. The symbol values are
indexes into a boxed list of values. This is my dict.ijs file. The
note at the end gives example use. I've fancifully called the wikipedia
"Encyclopedia Galactica", so the example refers to that of
http://en.wikipedia.org/wiki/Posterior_probability .
coclass 'dict'
NB. implement associative array
VALUE =: 0$a:
Monad =: :[:
symbol =: s:^:('literal'-:datatype) Monad
NB. index'`literal'
index =: 6&(s: symbol) Monad
NB. may fetch multiple items, returns boxed items
get =: (index { ".@:('VALUE'"_)) Monad
NB. may fetch multiple items, opens the boxes!
openGet =: >@:get Monad
put =: dyad define
KEY =. symbol y
if. 1 (= #) KEY do.
V =. < x
else.
assert KEY =&# x
V =. ;/ x
end.
I =. index KEY
L =. >: >./ I
if. L (> #) VALUE do. VALUE =: (+:L){.VALUE end.
VALUE =: V I} VALUE
KEY
)
cutSTRING =: #@:[ }.&.> , <;.1~ [ E. ,
assert ((,0);4 5 6 7) -: 1 2 3 cutSTRING i.8
embed =: >@:{.@:[ , ] , >@:{:@:[
assert 'bats' -: 'bs'embed'at'
Filter =: (#~`)(`:6)
assert 0 1 2 -: <&3 Filter i.6
assert 'SUPPER'-:<&(a.i.'a')Filter&.:(a.&i.)'SelectUPPERcase'
getAllSymbols =: _6 s: [: i. s:~@:0:
mp =: ($: |:@:+) :(+/ .*)
NB. probability of pants, from law of total probability
NB. y is the value for which to compute total probability
NB. totalProbability =: openGet :: (verb define)
NB. y is the value for which to compute total probability
totalProbability =: verb define
NB. returns P(y|CONDITIONi)P(CONDITIONi)
assert 'P()' -: 1 2 _1 { y
CONDITIONAL_GOAL =. '|'_1}y
SYMBOLS =. getAllSymbols''
LITERALS =. ": SYMBOLS
A =. ('`'~:{.&>) Filter CONDITIONAL_GOAL cutSTRING LITERALS
B =. ({.~i.&')')&.>A NB. B are boxes of conditionals.
PYI =. openGet;('`P(';')')&embed&.>B
PXYI =. openGet;(CONDITIONAL_GOAL;')')&embed&.>B
assert PYI -:&# PXYI
y put~ PYI mp PXYI
openGet y
)
Note 'Encyclopedia Galactica: posterior probability'
NB. wikipedia example on Tue Oct 22 15:18:33 EDT 2013
load'dict.ijs'
D=:conew'dict'
NB. probability student is a boy, complementary girl
((,-.)0.6)put__D'`P(B)`P(G)'
NB. probability wearing pants given sex
1 0.5 put__D'`P(P|B)`P(P|G)'
NB. compute and store probability to observe pants.
EMPTY [ totalProbability__D'`P(P)'
NB. Finally, Bayes Law. Find P(G|P).
%/ ; openGet__D'`P(P|G)`P(P)`P(G)'
0.25
)
Date: Mon, 3 Mar 2014 09:06:00 -0600
From: Yike Lu<yikelu.h...@gmail.com>
To:programm...@jsoftware.com
Subject: Re: [Jprogramming] Sorry state of slow parsing of JSON in J
Message-ID:
<CAO=ap+cujrjnshftlhipnyzrk2mfrav_wak+-qzbl5vigee...@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
A tangential but related question: does J have a dictionary data type?
Where can I read about it?
Thanks.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
