=Paper=
{{Paper
|id=None
|storemode=property
|title=Solving Hypertree Structured CSP: Sequential and Parallel Approaches
|pdfUrl=https://ceur-ws.org/Vol-589/paper11.pdf
|volume=Vol-589
|dblpUrl=https://dblp.org/rec/conf/aiia/LalouHA09
}}
==Solving Hypertree Structured CSP: Sequential and Parallel Approaches==
https://ceur-ws.org/Vol-589/paper11.pdf
Solving hypertree structured CSP : Sequential and
parallel approaches
Mohammed Lalou 1 , Zineb Habbas 2 , and Kamal Amroun 3
1
University of Bejaia, e-mail : mohammed.lalou@gmail.com
2
LITA, University of Metz, e-mail : zineb@univ-metz.fr
3
University of Bejaia, e-mail : k− amroun25@yahoo.fr
Abstract
Solving CSP is in general N P-Complete. However, there are vari-
ous subsets of CSPs that can be solved in polynomial time. Some of
them can be identified by analyzing their structure. Unfortunately the
proposed methods for exploiting these structural proprieties are not ef-
ficient in practice. So exploiting these structural properties for solving
CSP∫ is a crucial challenge. In this paper, we propose efficient algo-
rithms which exploit these structural proprieties, for both sequential
and parallel resolutions. Some experiments done on academic bench-
marks show the efficiciency of our approach.
1 Introduction
A CSP consists of a set V of variables, the domains D of these variables
and a set C of constraints over these variables. The objective is to assign
values in D to the variables in such a way that all constraints are satisfied.
CSP∫ are known to be N P-Complete. Considerable efforts have been made
to identify tractable classes. One approach is based on exploiting structural
properties of the constraints network. If the CSP is tree structured then
it can be solved in polynomial time. Many techniques have been devel-
oped to transform a CSP into a tree or an equivalent hypertree structured
CSP, we can cite [4, 6, 1]. The Generalized Hypertree Decomposition
method is the most general method. However the algorithm proposed to
solve the hypertree structured CSP is not efficient in practice. In this work,
we propose to improve this algorithm. Mainly we propose two algorithms:
an improved sequential algorithm and a parallel algorithm. The sequential
algorithm is based on the hashing technique, while the parallel algorithm
explores the pipeline technique and the parallel tree contraction algorithm
between the nodes to achieve the semi-join operation. In order to validate
our proposition we first compare the sequential algorithm proposed in this
paper with the basic algorithm proposed in [5] . Based on some experiments
done on benchmarks from the literature, we observed a promising gain in
terms of cpu time with our sequential approach. More particularly, good
results are observed when large size relations are considered. To validate
our parallel algorithm, we have proposed a simulation model using logical
Proceedings of the 16th International RCRA workshop (RCRA 2009):
Experimental Evaluation of Algorithms for Solving Problems with Combinatorial Explosion
Reggio Emilia, Italy, 12 December 2009
time assumptions. The experimental results outline the practical efficiency
of this algorithm and its performances in term of space memory.
This paper is organized as follows : section 2 gives preliminaries of con-
straint satisfaction problems and well known CSP decomposition methods
by developing more particulary the (generalized) hypertree decomposition
method which is the most general one. In section 3, we present our sequen-
tial algorithm called S HBR (for Sequential Hash Based Resolution) which
solves the hypertree structured CSP and we give some experiments of our
proposition with respect to the algorithm proposed in [5]. In section 4, we
present our parallel algorithm and in section 5, we give some experimental
results of this algorithm. Finally, in section 6, we give our conclusion and
some perspectives to this work.
2 Preliminary notions
The notion of Constraint Satisfaction Problem CSP was introduced by [7].
Definition 1. Constraint Satisfaction Problem: A constraint satisfac-
tion problem is defined as a 3-tuple P = hX, D, Ci where :
X = {x1 , x2 , ..., xn } is a set of n variables.
D = {d1 , d2 , ..., dn } is a set of finite domains; a variable xi takes its values
in its domain di .
C = {C1 , C2 , ..., Cm } is a set of m constraints. Each constraint Ci is a pair
(S(Ci ), R(Ci )) where
Q S(Ci ) ⊆ X, is a subset of variables, called scope of
Ci and R(Ci ) ⊂ xk ∈S(C)i ) dk is the constraint relation, which specifies the
allowed values combinations.
A solution of a CSP is an assignment of values to variables which sat-
isfies all constraints.
Definition 2. Hypergraph: The constraint hypergraph [2] of a CSP
P =< X, D, C > is given by H =< V, E > where E is a set of hyper-
edges corresponding to the scopes of the constraints in C, V is the set of
variables of P.
In this paper, hyperedges(H) is the set of the hyperedges of the hyper-
graph H. If h is a hyperedge, then var(h) is the set of variables of h.
Definition 3. A join tree for a hypergraph H is a tree T whose nodes are
the hyperedges of H, such that, when a vertex v of H occurs in two hyper-
edges e1 and e2 then v occurs in each node of the unique path connecting e1
and e2 in the tree T.
2
Definition 4. Hypertree: A hypertree for a hypergraph H is a triple
< T, χ, λ > where T = (N, E) is a rooted tree, and χ and λ are labelling
functions which associate each vertex p ∈ N with two sets χ(p) ⊆ var(H)
and λ(p) S⊆ hyperedges(H). If T 0 = (N 0 , E 0 ) is a subtree of T, we define
χ(T 0 ) = v∈N 0 χ(v). We denote the set of vertices N of T by vertices(T)
and the root of T by root(T). Tp denotes the subtree of T rooted at the node p.
Proposition 1. A CSP whose structure is acyclic can be solved in polyno-
mial time [6].
The goal of all structural decomposition methods is to transform a CSP
into an equivalent acyclic CSP which can be solved in polynomial time.
A decomposition method D associates to each hypergraph H a parameter
D-width called the width of H. The method D ensures that for a fixed k,
each CSP instance with D-width ≤ k is tractable and then can be solved in
polynomial time. Among these methods, the generalized hypertree decom-
position (GHD) dominates all the other structural decomposition methods.
In the next paragraph, we present the (generalized) hypertree decomposition
method.
Definition 5. A generalized hypertree decomposition [10] of a hypergraph
H =< V, E >, is a hypertree HD =< T, χ, λ > which satisfies the following
conditions :
1. For each edge h ∈ E, there exists p ∈ vertices(T ) such that var(h) ⊆
χ(p). We say that p covers h.
2. For each variable v ∈ V , the set {p ∈ vertices(T )|v ∈ χ(p)} induces a
connected subtree of T.
3. For each vertex p ∈ vertices(T ), χ(p) ⊆ var(λ(p)).
A hypertree decomposition of a hypergraph H =< V, E >, is a general-
ized hypertree decomposition HD =< T, χ, λ > which additionally satisfies
the following special condition : T
For each vertex p ∈ vertices(T ), var(λ(p)) χ(Tp ) ⊆ χ(p).
The width of a (generalized) hypertree decomposition < T, χ, λ > is
maxp∈vertices(T ) |λ(p)|. The (generalized) hypertree width (g)hw(H) of a hy-
pergraph H is the minimum width over all its (generalized) hypertree decom-
positions.
A hyperedge h of a hypergraph H = hV, Ei is strongly covered in
HD = hT , χ, λi if there exists p ∈ vertices(T ) such that the vertices in h
are contained in χ(p) and h ∈ λ(p). A (generalized) hypertree decomposition
HD = hT , χ, λi of H = hV, Ei is called complete if every hyperedge h of H
is strongly covered in HD.
3
3 A sequential algorithm
3.1 The basic algorithm
The sequential resolution of a CSP represented by its hypertree decomposi-
tion is given by the following algorithm [5] (see algorithm 1), called in this
paper B A algorithm (for Basic algorithm).
Algorithm 1 B A algorithm
1: Input : HD = hT , χ, λi
2: step1: complete the hypertree decomposition HD
3: step2: Solve each λ-term using multi-join operation.
4: step3: Solve the resulting CSP using semi-join operation.
5: Output: A solution of the given problem.
The second step consists in Solving each sub-problem represented as
the λ-term in each node of the hypertree. As a result, we have a new
hypertree whose the λ-terms are replaced by a unique constraint relation
Rp corresponding to the projection on the variables in χ(p) of the join of
the constraint relations in λ(p). More formally Rp = (./t∈λ(p) t)[χ(p)]. This
operation is expensive. The third step consists in semi-join operation which
is an other expensive operation.
3.2 The S HBR algorithm
Some notations : ϑi represents the intersection variables of the con-
straint Ci with the set of all variables in the union of Cj where j < i in a
given order. HT i is the hash table associated to Ri .
The S HBR for Sequential Hash Based Resolution algorithm, is for-
mally presented by Algorithm 2. It proceeds in two steps : The SP HBR
for Sub Problem Hash Based Resolution algorithm (lines 3 to 6 ) is the
optimization of the join operation. The A HBR for (Acyclic Hash Based
Resolution) algorithm (line7) is an optimization of the classical Acyclic solv-
ing algorithm [3].
We now give more details about SP HBR and A HBR algorithms.
a) SP HBR algorithm: The result of the SP HBR execution is a
join-tree whose every node n contains only one table Rn composed of the
tuples satisfying all sub-problems constraints. Nodes of this join-tree are
represented as follows: Rn = hχ(p), Hrel(p)i where χ(p) is the set of vari-
ables of the node p, and Hrel(p) is the constraint relation generated by
the join operation. Additionally, for the leaves nodes, Hrel(p)’s tuples are
hashed on intersection variables with the parent node of p. This algorithm
decomposes each node of the join tree into connected components. Then it
applies the join operation algorithm for each component using the hash join
4
Algorithm 2 Sequential hash Based Resolution(S HBR)
1: Input: a hypertree < T , χ, λ > of a hypergraph < H = V, E > where
T is a tree, χ associates to each t of T a set of nodes χ(t) ⊂ V , and λ,
a set of arcs λ(t) ⊂ E
2: Output : A solution.
3: Step1:
4: for each node n of T do
5: Apply the SP HBR algorithm on the node n.
6: end for
7: Step2: Apply A HBR algorithm on T .
principle. For doing that, we establish an order among relations in the same
component. Then, we hash each relation Ri of the constraint Ci , except the
first one, on intersection variables with the relations of the constraints that
are before Ci in this order. The join operation is applied using the Join pro-
cedure. These procedures cannot be described in this paper for restricted
space reason. We illustrate SP HBR by the following example.
Example 1. We consider a node p of a hypertree HD labelled as follows:
λ(p) = {C1 , C2 , C3 , C4 , C5 , C6 }, with: S(C1 ) = {a, b, c}, S(C2 ) = {i, j},
S(C3 ) = {a, d}, S(C4 ) = {d, b, f }, S(C5 ) = {f, g}, S(C6 ) = {i, k}.
There are two connected components R1 = {C1 , C3 , C4 , C5 } and R2 =
{C2 , C6 }. According to the order of the constraints in each component, we
have:
ϑ3 = S(C3 ) ∩ S(C1 ) = {a}, ϑ4 = S(C4 ) ∩ {S(C1 ) ∪ S(C3 )} = {b, d},
ϑ5 = S(C5 ) ∩ {S(C1 ) ∪ S(C3 ) ∪ S(C4 )} = {f }, ϑ6 = S(C6 ) ∩ S(C2 ) = {i}.(if
C is a hyperedge, then S(C) denotes the scope of C).
Therefore, for the first component, the R3 ’s tuples will be hashed on the
set of intersection variables ϑ3 = {a}, those of R4 and R5 on, respectively,
ϑ4 = {b, d} and ϑ5 = {f }. For the second component, the tuples of R6
will be hashed on ϑ6 = {i}. After computing a cartesian product of the
components, the resulting tuples t are hashed according to the intersection
variables with the parent node of p, if p is a leaf. So, t will be ready for the
semi-join operation in A-HBR algorithm. We do not add t to its related
partition P unless if P = φ. Thus, we will have one and only one tuple
in each partition by eliminating the duplicated tuples. This elimination has
no impact on the resolution because, for the A-HBR algorithm, each parent
node is filtered on its sons 1 . Moreover, we are interested to get only one
solution for the CSP.
b) The A HBR algorithm: The A HBR algorithms tests the consis-
tency of a constraint acyclic network and generates a solution if it exists.
1
It is sufficient to have only one tuple by partition in the son node relations.
5
This algorithm takes the hypertree resulting from the SP HBR algorithm
as its input. The A HBR algorithm (see. Algorithm 3 ) proceeds in two
steps: the CONTRACT step and the S SEARCH step. The first step con-
tracts the deep-rooted hypertree HT . A semi-join operation is associated
to each contraction operation. The result is a directional from root toward
leaves arc consistent hypertree HT 0 (line 3). The second step is the solution
search operation, which develops the solution from the HT 0 ’s root to leaf
nodes (line 4).
Algorithm 3 Acyclic Hash Based Resolution algorithm (A HBR)
1: Input: A hypertree HT = hT , χ, λi
2: Output : Determine a directional arc consistent and generate a solution.
3: HT 0 = CONTRACT (HT )
4: S SEARCH (HT 0 ),
The CONTRACT algorithm strengths the directional arc consistency of
the input hypertree. Given a hypertree, a contraction operation is realized
between each leaf node and its parent. For each contraction operation we
hash each tuple t of a parent node (p) relation on intersection variables with
its sons leaves nodes ai . We denote hi the hashed values. If ai ’s hash tables
partitions related to hi are not empty , we hash t on intersection variables
of p with its parent node and save it. Otherwise, we eliminate t. The leaves
nodes ai are marked and considered as pruned. We make the same with
the obtained hypertree, and so on, until all the nodes, excepted the root,
will be marked. The final result of the contraction operation of a parent
node p with its sons leaves nodes, is a node p0 = hχ(p0 ), Hrel(p0 )i, whose
relation is arc consistent with those of the son nodes relations, and hashed
on the intersection variables of p with its parent node. If the hashed relation
Hrel(p0 ) is empty, the problem has no solution.
The S SEARCH algorithm is a classical Backtrack free algorithm applied
to the directional arc consistent hypertree resulting from the CONTRACT
algorithm execution.
3.3 Complexity of S HBR
The complexity of the SP HBR algorithm is
h−1
X
(r ∗ P i ) (1)
i=1
Where r is the the maximum relation size, h is the hypertree width, P is the
maximum number of tuples in the partition of hashing on the variables of ϑ
(ϑ represents the intersection variables of Ci with all variables in Cj where
j < i ). For the A HBR algorithm, its complexity is O(nP 2 ). Where n is
6
the number of constraints in the obtained CSP. But in our case, we have
only one tuple by partition, so the complexity of A HBR algorithm is only
O(n).
3.4 Experiments
We have implemented the S HBR algorithm and the basic algorithm [5]
(noted here B A). We have tested the two solvers on a set of CSP bench-
marks. This benchmarks collection is represented using the new XCSP 2.1
format proposed by The organizational committee of the 3th international
competition of the CSPs solvers 2 [8]. The solvers inputs are CSPs instances
and the corresponding hypertrees in GML format. For this, we have ex-
ploited a GML Parser proposed by Raitner and Himsol 3 and another one,
proposed by Roussel 4 , for XCSP file. The experiments are made on Linux
using an Intel Pentium IV, with 2.4 GHz of CPU and 600 Mb of RAM. In
our results, |V |, |E| and |R| represent respectively the variables number, the
constraints number and maximum number of tuples by relation associated
to a given CSP. Nd and HT W are respectively the nodes number of the
hypertree and the hypertree width. S-HBR(sec) illustrate the S-HBR com-
putational time in seconds, and B A(sec) the time of the B A algorithm.
The table 1 summarizes our experiments. The hash function used in this
paper is
Xn
f (t) = ((xi + 1) ∗ 10ln(f (t))+1 ) (2)
i=1
We observe that S-HBR clearly improves the B A algorithm in terms of
CPU time for all the instances.
The S HBR algorithm gets more better results for the instances which
have large relations, because it browses only a sub- part of the relation rather
than the totality one as in B A.
4 The parallel Algorithm
In this section, we introduce a parallel version of our previous sequential
algorithm S HBR. We called it P HT R for Parallel Hypertree Resolution
algorithm and it is described by the algorithm 4. This algorithm uses both
pipeline technique and the parallel tree contraction technique. The
pipeline avoids the storage of intermediate results so it reduces the mem-
ory space explosion. The parallel tree contraction is the most well known
adapted approach for the semi-join operation. However the known parallel
2
http://www.cril.univ-artois.fr/ lecoutre/research/benchmarks/benchmarks.html
3
http://www.cs.rpi.edu/ puninj/XGMML/GML-XGMML/gml-parser.html
4
http://www.cril.univ-artois.fr/ roussel/CSP-XML-parser/
7
CSP
Name |V| |E| Nd HT W |R| B A (sec.) S HBR (sec.)
3-insertions-3-3 56 110 86 7 3 >1000 127
domino-100-100 100 100 50 2 100 27 7
domino-100-200 100 100 50 2 200 103 28
domino-100-300 100 100 50 2 300 241 61
series-7 13 42 39 4 42 >1000 106
hanoi-7 126 125 125 1 6558 71 9
haystacks-06 36 95 88 3 8 9 4
haystacks-07 49 153 144 4 9 514 66
langford-2-4 8 32 31 4 8 11 3
Renault 101 134 81 2 48721 780 20
bf-1355-e-63 532 339 223 6 3124 96 11
bf-1355-g-63 532 339 223 6 7775 561 36
bf-2670-b 1244 1354 733 8 31 >1000 77
bf-2670-c 1244 1354 641 13 31 >1000 125
Table 1: Experimental results of S-HBR algorithm
tree contraction considers at each iteration, for each parent node with only
one son node leaf. To increase the parallelism degree in this operation, we
developed a new alternative of this technique, based on partitioning the par-
ent node relation (the critical resource) between all processors, which allows
an asynchronous updating.
The P HT R(algorithm 4) procedure proceeds in two steps. The first one
concerns the parallel resolution of sub-problems using the P SBR algorithm
(line 4). The second step is the resolution of the whole problem. It is
performed using the P Solving algorithm (line 6).
Algorithm 4 P HT R (Parallel HyperT ree Resolution) algorithm
1: Input: a hypertree hT , χ, λi of a hypergraph H = (V(H), E(H))
2: Output : The solution of the problem
3: for each node n of T in parallel do
4: P SBR (n) // Apply the P SBR algorithm on the node n.
5: end for
6: P Solving(T’) // Apply P Solving algorithm on T 0 (T 0 is the obtained
hypertree in the step 3).
4.1 The P SBR algorithm
The P SBR for Parallel Sub Problem Resolution algorithm uses the hash
technique. The P SBR algorithm proceeds as the SP HBR algorithm.
After establishing an order on the constraints, it builds a pipeline line com-
posed of join operators whose the first relation is the probe relation and the
remaining ones are build relations. The join operation is performed accord-
ing to the hash pipelined join principle. Thus, each build relation is hashed
on its intersection variables (ϑi ) with the union of the previous relations.
8
The parallelism in this algorithm is provided by the Pipeline. At each level
of the pipeline corresponds a join operation with two operands. The first
one is R1 (the probe relation) for the first operator, Rtmp [i − 1] for the ith
operator. The second one is the build relation Ri for the ith operator. We
make a join operation between each tuple of the first operand and all the
tuples of the second operand. The obtained tuples from the ith operator are
saved in the temporary relation of (Rtmp [i + 1]) of the following operator.
4.2 P Solving algorithm
P Solving algorithm is the parallel version of A HBR algorithm. It con-
sists in two operations, P CON T RACT (for Parallel CONTRACT), and
P S SEARCH (for Parallel SEARCH ). The first operation concerns the
parallelization of the CONTRACT operation, while the second one corre-
sponds to the parallel version of the S SEARCH operation.
Algorithm 5 P Solving algorithm
1: Input: A join tree T .
2: Output : The solution for the problem.
3: T 0 = P CON T RACT (T ).
4: P S SEARCH(T 0 ).
4.2.1 P CON T RACT algorithm
The P CON T RACT operation strengths the directional arc consistency in
the input hypertree. It is based on two operations, P RAKE (for Parallel
RAKE), and P COM P RESS (for Parallel COMPRESS ), where RAKE
and COM P RESS are the basic operations of a parallel tree contraction [9].
We explain these two operations.
P RAKE operation: is based on the fact that the existence of only
one resource is the reason of the sequential execution. This resource is the
parent node relation. This resource must be manipulated simultaneously
by all processors. Therefore, it must be shared between them. Thus, this
technique consists in partitioning the parent relation between all processors
which compute the semi-join operation of this last relation with its son nodes
relations. These partitions are manipulated periodically between processors.
P COM P RESS operation : is a pipelined execution of COMPRESS
operation. It is applied on a sequence of nodes or a chain. Let {n1 , n2 , ..., nj }
be a chain of a tree T where ni+1 is the ni ’s only son, the application of
P COM P RESS to this chain results in a new tree contracted T 0 in which
the node n1 is transformed thus:
• χT 0 (n1 ) = χT (n1 )
9
• HrelT 0 (n1 ) = Πχ(n1 ) (HrelT (n2 ) ./ HrelT (n3 ) ./ · · · ./ HrelT (nj )).
It makes a pipeline of join operations between the chain’s nodes, and at
each time when it takes a tuple from the probe relation, it joins it with all
tuples of the corresponding partition in the hash table of the build relation.
Then it communicates it to the probe relation of the next operator. After the
execution of the last operator, it projects the tuples results on the variable set
χ of each node of the chain, it hashes it on the intersection variables of each
node with its parent node, and puts the tuple result in the corresponding
node relation.
4.2.2 P S SEARCH algorithm
P S SEARCH algorithm is the parallel version of S SEARCH algorithm.
It works in the same way, except that, each time, after taking a tuple from a
parent node, the research of corresponding tuples in the son nodes is made
in parallel.
4.3 Complexity of P HT R
For a binary hypertree, the complexity of the P SBR algorithm is
O(r ∗ P h−1 ) (3)
using O(h) operations in a PRAM of type EREW. r is the the maximum
relation size, h is the hypertree width, P is the maximum number of tuples
in the partition of hashing on the variables of ϑ (ϑ represents the intersection
variables of Ci with all variables in Cj where j < i ). For the P Solving algo-
rithm, its complexity is O(log n2 ) using O(n) operations. n is the hypertree
size (number of nodes).
5 Experiments of parallel algorithm
We have developed a simulation model for P SBR and P Solving algo-
rithms based on logical time assumptions.
5.1 Simulation model
Our system is composed of simple modules described as classical sequential
programs, which are collected and executed in a parallel way by the simula-
tor. A module is the set of operations performed by a processor we called a
process. For the internal working of a process, we split operations that can
be performed in three main types : Reading a tuple, Searching in hash tables
including the tuples join operation and Writing a result tuple. These op-
erations are executed sequentially between them and in parallel with those
10
of the other processes. Each process explores the maximum of potential
parallelism if its environment (its input variables) do not conflict during the
other process execution. For a pipeline execution, if the processes are not
adjacent, there is no conflict. Otherwise, we use the semaphore mechanism
to synchronize them.
5.2 Simulation process
Each operation or each event is performed in an interval of time and the
simulator must therefore keep a logical time counter. The asked question
is : When we increment this counter, and what is the relation between this
counter and the events execution time (physical counter)? We make a step-
by-step simulation so that for each step (logical time unit) all operations
which can be performed at the same time start as follows :
1. The simulator executes all processes and it increments the logical time
counter at each iteration.
2. For all processes that are not in conflict, the simulator increments the
physical time counter.
3. After a logical time step, the simulator updates all the shared variables.
4. It restarts a new step by considering the new variable states.
5.3 Experimental protocol
We developed a simulation of P HT R algorithm in order to check: first, if
it gives good temporal and spatial complexities and second, if the new ap-
proach of parallel tree contraction proposed in this paper can be considered
as an interesting alternative of the one considered in [9]. In order to compare
our results, we simulate PT AC algorithm. In this section, experimentations
are divided in three categories :
1. Simulation of P SBR algorithm in order to estimate its interest in
term of space memory, since the sub-problems resolution is the source
of a bad spatial complexity in general.
2. Simulation of P RAKE in order to evaluate its performances.
3. Simulation of P Solving algorithm in order to estimate its temporal
optimization, and to measure the contribution of the new parallel tree
contraction technique P Rake by leading a comparison with the one
proposed in (PT AC algorithm).
The simulation has been accomplished on a PC under Linux 6.0.52 ver-
sion with CPU Intel Pentium IV 2.4 GHz and 512 Mb of RAM. We use a
11
set of benchmarks CSP which exist in the literature. We take the average
of the data obtained after several executions. In our results, |V |, |E| et
|r| represent respectively the variables number, the constraints number and
the maximum relations cardinality. Nd , Nf and HT W are respectively, the
nodes number, the leaf nodes number and the hypertree width. Np rep-
resents the processors number and N bS represents the number of pipeline
stages.
In order to be hardware independent, measure units are Tuple for space
and Operation (Reading, Searching or Writing) for time.
5.4 P SBR algorithm simulation
The table 2 presents the space memory required for a pipeline (R P SBR)
and no pipeline (R join) execution of the join operations for different CSP
instances (consistent and inconsistent).
CSP
Name |V| |E| Nd HTW |R| NNW > 2 |R join| |R P SBR|
geom-30a-4 30 81 70 4 4 20 12168 36
haystacks-07 49 153 144 4 9 27 109368 48
Renault 101 134 81 2 48721 3 979 3
pret-60-25 60 40 28 5 4 19 3288 40
pret-150-60 150 100 50 5 4 50 6136 114
pret-150-75 150 100 54 5 4 54 6552 127
queen-5-5-3 25 160 7 10 2 7 1582 38
haystacks-06 36 95 88 3 8 18 7870 18
langford-2-4 8 32 31 4 8 5 3520 9
pigeons-7 7 21 19 3 6 4 6600 7
series-6 11 30 27 3 30 5 2920 5
queen-12 12 66 61 6 12 6 36180 24
mug-100-25-4 100 166 133 3 31 32 26762 32
Table 2: P SBR algorithm simulation of CSP instances
N N W > 2 denotes the number of nodes in the hypertree with width
more than 2. We observed that the gain in space memory is very impor-
tant in instances for which the pipeline executions number (NNW>2) is
important (haystacks-07 ). We also observed that the gain depends on the
number of nodes in hypertree (mug-100-25-4 ). It is also proportional to
the number of tuples by relation. As the two factors that influence on CSP
problems complexity are the constraints relations size, and the structural
decomposition width, and in order to estimate the contribution of P SBR,
we evaluate in figure 1, in (a) and (b), the P SBR algorithm behavior w.r.t
respectively the constraints relations size and the stages number in pipeline
for two classes of CSP problems: aim-100-6 and full-Insertion.
N.B: P SBR algorithm forms a pipeline in each node of the hypertree.
So, the stages number in the pipeline is the number of constraints in the
12
node minus one.
Spatial gain (a) (b)
Spatial gain
Relations size Pipeline stages number
Figure 1: spatial performances of P SBR according to (a) relations size (b)
stages number in pipeline
Figure 1 shows that more the relations size is important more the mem-
ory space gain is considerable. A similar interpretation is illustrated for the
stages number in pipeline.
5.5 P RAKE algorithm simulation
P RAKE tries to optimize the CPU time of the whole problem. We present
here a simulation of this operation by computing its efficiency Ef f =
Ts
NP ∗Tp where NP is the processors number and Ts (reps. Tp ) the time of
sequential (reps. parallel) resolution of the sub- hypertree.
Sub-hypertree
n Np |R| Ts Tp Eff
1 2 196 1034 550 0.94
2 4 4500 230720 59974 0.96
3 3 27000 31401 10911 0.96
4 4 12000 95095 24875 0.95
5 5 18000 156884 33061 0.95
6 6 680 9812 1659 0.98
7 6 87808 1055270 176943 0.99
8 8 7776 90125 12252 0.92
9 9 18144 297846 34410 0.96
10 10 46656 820947 84798 0.97
Table 3: P RAKE algorithm efficiency
Table 3 presents the obtained results for some sub-hypertrees of different
CSP instances. We observed a good efficiency of P RAKE operation from 1
to ten processors. This algorithms allows a load balancing by sharing tuples
to be filtered between all processors.
N.B: The number of processors used to contract a sub-hypertree is equal
to the number of semi-join operations which is equal to the stages number
13
in pipeline.
Speed up
Theoritical
P-RAKE
Nbr of processors
Figure 2: P RAKE algorithm speedup
The P RAKE algorithm speedup is presented in Figure 2. It is obvious
that it is close to theoretical speedup (linear).
5.6 P Solving algorithm simulation
Before comparing P Solving with A HBR, it seems natural to compare
P CON T RACT algorithm with an algorithm applying the tree contraction.
CSP A − HBR PTAC P − Solving
Name |V | |E| Nd HT W |R| Nf T Nc T Nc T
hayst06 36 95 88 3 8 23 16736 13 253 10 158
dom200 100 100 50 2 200 1 3632 48 3632 48 3632
hanoi-6 62 61 59 1 2148 2 260 37 169 37 169
mug25-4 100 166 133 3 31 26 119 12 47 9 38
myc-3-4 11 20 15 4 12 4 267 6 164 3 42
myc-4-4 23 71 48 6 6 12 124 8 315 4 299
Renault 101 134 81 2 48721 12 7758 26 38167 9 1456
series-6 11 30 27 3 30 5 622 8 256 4 155
hole-6 42 133 132 2 8 127 1970 85 1201 1 29
Table 4: P-Solving algorithm simulation
Nc represents the number of contraction operations performed and T the
computational time. We have considered 1000 operations as a measure unit.
Table 4 gives the CPU time of both A HBR algorithm and P T AC algo-
rithm (CONTRACT operation) and the one of P Solving (P CON T RACT
operation). It results that P CON T RACT generates a gain in temporal
complexity inversely proportional to the number of contractions. More the
number of the leaf nodes contracted is big more the number of contraction is
less and more the gain in time is important (myc-3-4, Renault and hole-6 ),
and vice versa, which is the case for hayst06 and mug25-4. We can also
remark for the problem hole-6, the CPU time gain (29 instead of 1201) is
considerable. This is due to the fact the hypertree is contracted on only one
step, instead of taking 81 steps.
14
6 Conclusion
In this paper, we have presented sequential and parallel algorithms to solve
CSPs by exploiting their structural properties. The algorithms exploit the
hash technique. For the sequential algorithm, we have done some experi-
ments on benchmarks from the literature and the results we have obtained
are promising. Good results are observed when the number of tuples of rela-
tions is important. The parallel algorithm is based on the notion of pipeline
and parallel tree contraction which is a parallel version of the well known
tree contraction. We performed simulations on some benchmarks and the
results are very good. Future works include the execution of our parallel
algorithm on a parallel machine.
References
[1] Gyssens, M., Cohen D., Jeavons, P.: A unified theory of structural
tractability for constraint satisfaction problems. J. Comp. and Sys. Sci.
74, 721–743 (2007)
[2] Dechter, R.: Constraint Processing. Morgan Kaufmann, (2003)
[3] Dechter, R.: Constraint Networks. Encyclopedia of Artificial Intelli-
gence, pages 276-285, (1992)
[4] Dechter, R., Pearl, J.: Tree clustering for constraint networks. Art. Int.
38, 353–366, (1989)
[5] Gottlob, G., Leone, N., Scarcello, F.: A comparison of structural csp
decomposition methods. Art. Int. 124, 243–282 (2000)
[6] Gyssens, M., Jeavons, P.G., Cohen, D.A.: Decomposing constraint sat-
isfaction problems using database techniques. Art. Int. 66, 57–89 (1994)
[7] Montanari, U.: Networks of constraints: Fundamental properties and
applications to pictures processing. Inf. Sci. 7, 95–132 (1974)
[8] XML Representation of Constraint Networks Format XCSP 2.1,
http://www.cril.univ-artois.fr/CPAI08
[9] Karp, R. M. and Zhang, Y.: Randomized Parallel Algorithms for Back-
track Search and Branch-and-Bound Computation, Journal of the Asso-
ciation for Computing Machinery, 40, 3, 765–789 (1993)
[10] Gottlob, G. and Leone, N. and Scarcello, F.: Hypertree Decomposi-
tions: A survey, Proceedings of MFCS ’01,(2001),
15