A model of constraint solvers by chaotic iteration adapted to

Description

Niveau: Supérieur, Doctorat, Bac+8
A model of constraint solvers by chaotic iteration adapted to value withdrawal explanations LIFO, EMN Gerard Ferrand, Willy Lesaint, Alexandre Tessier public, rapport de recherche D1.1.1 Abstract The aim of this report is to provide the theoretical foundations of domain reduc- tion. The model is well suited to the solvers on nite domains which are used on the respective platforms of each partner of the project: GNU-Prolog (INRIA), CHIP (COSYTEC) and PaLM (EMN). A computation is formalized by a chaotic iteration of operators and the result is described as a closure. The model is well suited to the denition of traces and explanations which will be useful for the debugging of con- straint programs. This report only deals with the reduction stage. It will be extended to the labeling and the host language in next reports. 1 Introduction Constraint Logic Programming (CLP) [12] can be viewed as the reunion of two program- ming paradigms : logic programming and constraint programming. Declarative debugging of constraints logic programs has been treated in previous works and tools have been pro- duced for this aim during the DiSCiPl (Debugging Systems for Constraint Programming) ESPRIT Project [8, 15]. But these works deal with the clausal aspects of CLP.

• another constraint


• constraint symbols


• consistency operator


• additional constraint


• involving only


• labeling provides exact


• provides


• take labeling


• global domain


• operators

Sujets

• Ferrand
• Tessier

Informations

• Publié par: profil-zyak-2012
• Nombre de lectures: 129
• Langue: English

Extrait

A
mo
del
dieren
ork,
t
of
is
constrain
strongly
t
instan
solv
v
ers
not
b
t.
y
constrain
c
to
haotic
ts.
iteration
to
adapted
vides
to
t
v
lab
alue
eling
withdra
The
w
and
al
in
explanations
.
LIF
[16
O,
is
EMN
ard
G
eling)

one
erard
according
F
ximates
errand,
a
Willy
t.
Lesain
branc
t,
t
Alexandre
ork
T
of
essier
t
public,
at
rapp
the
ort
y
de
e
rec
of
herc
and
he
constrain
D1.1.1
t
Abstract
to
The
v
aim
ect
of
go
this
bining
rep
The
ort
the
is
testing
to
the
pro
cannot
vide
ts.
the
domain
theoretical
the
foundations
en-
of
com
domain
ds
reduc-
er
tion.
e
The
searc
mo
seen
del
.
is
to
w
fully
ell
searc
suited
h)
to
the
the
el
solv
ols
ers
lev
on
end
nite
t
domains
w
whic
solv
h
Here
are
in
used
wide
on
of
the
nite
resp
op
ectiv
aim
e
programming
platforms
e
of
Problems
eac
that
h
vide
partner
of
of
whic
the
with
pro
the
ject:
solv
GNU-Prolog
to
(INRIA),
solutions
CHIP
w
(COSYTEC)
metho
and
one
P
in
aLM
un
(EMN).
singletons
A
The
computation
reduction)
is
eliminating
formalized
whic
b
e
y
the
a
eling
c
solutions
haotic
simply
iteration
In
of
eling
op
ery
erators
e
and
o
the
of
result
o
is
more
describ
this
ed
eling
as
treated.
a
only
closure.
of
The
tree:
mo
part
del
additional
is
the
w
future
ell
e
suited
our
to
order
the
e
denition
the
of
tree
traces
single
and
to
explanations
on
whic
constrain
h
lev
will
alone.
b
to
e
used
useful
this
for
el
the
dep
debugging
on
of
constrain
con-
domain
strain
the
t
a
programs.
to
This
e
rep
ts.
ort
w
only
are
deals
terested
with
a
the
eld
reduction
applications
stage.
constrain
It
programming:
will
domains
b
pr
e
agation
extended
The
to
of
the
t
lab
is
eling
solv
and
Constrain
the
Satisfaction
host
(CSP)
language
],
in
is
next
pro
rep
an
orts.
tiation
1
the
In
ariables
tro
h
duction
correct
Constrain
resp
t
to
Logic
constrain
Programming
The
(CLP)
er
[12
es
]
w
can
the
b
com
e
t
view
o
ed
t
as
ds.
the
rst
reunion
(lab
of
consists
t
partitioning
w
domains
o
til
program-
obtain
ming
and,
paradigms
them.
:
second
logic
(domain
programming
reduces
and
domains
constrain
some
t
alues
programming.
h
Declarativ
b
e
correct
debugging
to
of
constrain
constrain
Lab
ts
pro
logic
exact
programs
whereas
has
reduction
b
appro
een
them.
treated
general,
in
lab
previous
alone
w
v
orks
exp
and
siv
to
and
ols
go
ha
d
v
bination
e
the
b
w
een
metho
pro-
is
duced
ecien
for
In
this
pap
aim
lab
during
is
the
really
DiSCiPl
W
(Debugging
consider
Systems
one
for
h
Constrain
the
t
h
Programming)
the
ESPRIT
eling
Pro
is
ject
as
[8
constrain
,
to
15
CSP
].
In
But
w
these
w
w
plan
orks
extend
deal
framew
with
in
the
to
clausal
tak
asp
lab
ects
and
of
whole
CLP
h
.
(instead
This
a
rep
branc
ort
in
fo
accoun
cus
1This
kind
of
a
ears
constrain
computation
(see
is
are
not
ws.
easy
wn
to
to
debug
v
b
the
ecause
the
CSP
haotic
are
the
not
].
algorithmic
in
pro-
obtain
grams
domain
[13
diagnosis.
].
but
The
It
constrain
ha
ts
whic
are
3
re-in
rules
v
computation
ok
from
ed
e
according
outside
to
are
the
of
domain
e
reductions
een
un
,
til
explanation
a
of
x-p
tics
oin
a
t
ear
is
to
reac
the
hed.
withdra
But
another
the
The
order
founded.
of
basic
in
reduction
v
to
o
is
cation
erators
is
are
not
of
kno
eral
wn
of
a
applications
priori
of
and
of
strongly
our
de-
y
p
diagnosis.
ends
constrain
on
w
the
w
strategy
this
used
].
b
b
y
v
the
a
solv
these
er.
pro
The
asp
main
an
con
.
tribution
a
of
for
this
do
rep
domain.
ort
w
is
wrong
to
t
formalize
alw
the
of
domain
constrain
reduction
a
in
has
order
of
to
structured
pro
ma
vide
is
a
es
notion
t
of
mo
explanation
op
for
asso
the
Section
basic
build
ev
explanation
en
of
t
during
whic
whic
h
onsible
is
v
\the
v
withdra
domain.
w
orks
al
detailed
of
ha
a
lot
v
,
alue
these
from
nations
a
scop
domain".
rep
This
]).
notion
of
of
ated
explanation
to
is
to
essen
asp
tial
debugging
for
programming
the
wh
debugging
ha
of
failure
CSP
do
programs.
y
Indeed,
has
the
in
disapp
case
earance
a
of
empt
a
is
v
of
alue
elongs
from
So,
a
wh
domain
alues
ma
disapp
y
an
b
failure.
e
is
a
us
symptom
ected
of
the
an
w
error
aiting
in
con
the
v
program.
v
But
v
the
not
error
the
is
explanation
not
alue
alw
help
a
what
ys
our
where
imp
the
note
v
is
alue
ys
has
resp
disapp
v
eared
al.
and
ma
an
e
analysis
reduction
of
whic
the
pro
explanation
w
of
v
the
is
v
ject
alue
this
withdra
b
w
rep
al
as
is
2
necessary
notations
to
for
lo
Problems.
cate
es
the
for
error.
on
[9
and
]
Section
pro
deduction
vides
mo
a
uses
to
order
ol
Next
to
conclusion.
nd
is
symptoms,
subset
this
op
pap
used
er
the
pro
and
vides
h
a
resp
to
for
ol
remo
whic
al
h
a
could
alue
b
a
e
Sev
used
w
to
sho
nd
that
errors
analysis
from
explanations
symptoms.
v
Explanations
a
are
of
a
[10
to
11
ol
But
to
applications
help
expla-
debugging:
are
w
the
e
e
extract
this
from
ort
a
[11
(wide)
Here,
computation
de