Grounding Issues in Parallel and Multi-Engine ASP Solving Francesco - - PowerPoint PPT Presentation
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion Grounding Issues in Parallel and Multi-Engine ASP Solving Francesco Ricca Dipartimento di Matematica e Informatica, Universit della Calabria 87036 Rende,
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Dipartimento di Matematica e Informatica, Università della Calabria 87036 Rende, Italy
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
2
3
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
DGH09, CHO+09, RDG+10, RGA+12, GNA13, MRT15, LR15]
Do not miss LPNMR Session 16 (08:45-10:40)
→ The 6th ASP Competition Report and winners announcement!
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
2
2 /ΠP 2 [EGM97, DEGV01]
→ even problems not (polynomially) translatable to SAT or CSP
3
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
2
2 /ΠP 2 [EGM97, DEGV01]
→ even problems not (polynomially) translatable to SAT or CSP
3
4
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Delete rules with false body w.r.t I I is an Answer Set if I is a minimal model of P′
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
Receives a program P that contains variables and outputs a program P′ semantically equivalent to P but variable-free
2
Model Generation
→ Produces candidate models → Similar to a SAT Solver: DPLL/CDCL Search
Stable Model Checking
Checks if candidates are Answer Sets Co-NP check
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
Develop new evaluation techniques Exploit the features of the hardware, e.g., CPU/GPU cores
2
Algorithm selection techniques [Ric76] Select (ideally) the “best combination” of solving tools per each instance
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
Develop new evaluation techniques Exploit the features of the hardware, e.g., CPU/GPU cores
2
Algorithm selection techniques [Ric76] Select (ideally) the “best combination” of solving tools per each instance
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Instantiation has impact on the overall efficiency of a system → Intelligent Instantiators solve problems in P
→ Information Integration [LGI+05], industrial applications [LR15] → Analysis of alternative encodings of the 5th ASP Comp. [GMR15]
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Instantiation has impact on the overall efficiency of a system → Intelligent Instantiators solve problems in P
→ Information Integration [LGI+05], industrial applications [LR15] → Analysis of alternative encodings of the 5th ASP Comp. [GMR15]
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Instantiation has impact on the overall efficiency of a system → Intelligent Instantiators solve problems in P
→ Information Integration [LGI+05], industrial applications [LR15] → Analysis of alternative encodings of the 5th ASP Comp. [GMR15]
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Instantiation has impact on the overall efficiency of a system → Intelligent Instantiators solve problems in P
→ Information Integration [LGI+05], industrial applications [LR15] → Analysis of alternative encodings of the 5th ASP Comp. [GMR15]
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Distributed evaluation (beowulf cluster)
Parallel evaluation on SMP machines DLV Input language
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Avoid thread starvation Reduce the usage of locks
Backtracking algorithm Recursion
→ Parallel semi naive
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Output equivalent but smaller than theoretical instantiation
→ produce relevant ground information
→ divide program in modules + semi naïve → At iteration n use only information derived at iteration n − 1
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
i.e.,(b, h) ∈ E ⇐ ⇒ ∃ r ∈ P | h ∈ Head(r) and b ∈ Body +(r)
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
(≺s denotes the transitive closure of ≺)
1
2
First process exit rules Then process recursive ones
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = { g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive } Istance: {a(1), a(2), t(1), t(2)} Instantiation: q(1) :- a(1), not b(1). q(2) :- a(2), not b(2). p(1) :- q(1). p(2) :- q(2). g(1) :- t(1). g(2) :- t(2). r(1) :- q(1), g(1). r(2) :- q(2), g(2). g(1) :- q(1), r(1). g(2) :- q(2), r(2).
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = { g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive } Istance: {a(1), a(2), t(1), t(2)} Instantiation: q(1) :- a(1), not b(1). q(2) :- a(2), not b(2). p(1) :- q(1). p(2) :- q(2). g(1) :- t(1). g(2) :- t(2). r(1) :- q(1), g(1). r(2) :- q(2), g(2). g(1) :- q(1), r(1). g(2) :- q(2), r(2).
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = { g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive } Istance: {a(1), a(2), t(1), t(2)} Instantiation: q(1) :- a(1), not b(1). q(2) :- a(2), not b(2). p(1) :- q(1). p(2) :- q(2). g(1) :- t(1). g(2) :- t(2). r(1) :- q(1), g(1). r(2) :- q(2), g(2). g(1) :- q(1), r(1). g(2) :- q(2), r(2).
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = { g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive } Istance: {a(1), a(2), t(1), t(2)} Instantiation: q(1) :- a(1), not b(1). q(2) :- a(2), not b(2). p(1) :- q(1). p(2) :- q(2). g(1) :- t(1). g(2) :- t(2). r(1) :- q(1), g(1). r(2) :- q(2), g(2). g(1) :- q(1), r(1). g(2) :- q(2), r(2).
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = { g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive } Istance: {a(1), a(2), t(1), t(2)} Instantiation: q(1) :- a(1), not b(1). q(2) :- a(2), not b(2). p(1) :- q(1). p(2) :- q(2). g(1) :- t(1). g(2) :- t(2). r(1) :- q(1), g(1). r(2) :- q(2), g(2). g(1) :- q(1), r(1). g(2) :- q(2), r(2).
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
SCCs of the Dependency Graph DG subdivide the input program in modules Evaluate modules in parallel by respecting dependencies
2
Rules in modules are evaluated in parallel First exit rules, then recursive ones
(parallel semi-naïve technique)
3
Parallel matching of variables dynamically “split” the extension of the first body predicate
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
SCCs of the Dependency Graph DG subdivide the input program in modules Evaluate modules in parallel by respecting dependencies
2
Rules in modules are evaluated in parallel First exit rules, then recursive ones
(parallel semi-naïve technique)
3
Parallel matching of variables dynamically “split” the extension of the first body predicate
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
SCCs of the Dependency Graph DG subdivide the input program in modules Evaluate modules in parallel by respecting dependencies
2
Rules in modules are evaluated in parallel First exit rules, then recursive ones
(parallel semi-naïve technique)
3
Parallel matching of variables dynamically “split” the extension of the first body predicate
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
SCCs of the Dependency Graph DG subdivide the input program in modules Evaluate modules in parallel by respecting dependencies
2
Rules in modules are evaluated in parallel First exit rules, then recursive ones
(parallel semi-naïve technique)
3
Parallel matching of variables dynamically “split” the extension of the first body predicate
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Input Program: q(X) :- a(X), not b(X). p(X) :- q(X). r(X) :- q(X), g(X). g(X) :- q(X), r(X). g(X) :- t(X). Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = {g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive }
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Input Program: q(X) :- a(X), not b(X). p(X) :- q(X). r(X) :- q(X), g(X). g(X) :- q(X), r(X). g(X) :- t(X). Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = {g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive }
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Input Program: q(X) :- a(X), not b(X). p(X) :- q(X). r(X) :- q(X), g(X). g(X) :- q(X), r(X). g(X) :- t(X). Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = {g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive }
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Input Program: q(X) :- a(X), not b(X). p(X) :- q(X). r(X) :- q(X), g(X). g(X) :- q(X), r(X). g(X) :- t(X). Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = {g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive }
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Input Program: q(X) :- a(X), not b(X). p(X) :- q(X). r(X) :- q(X), g(X). g(X) :- q(X), r(X). g(X) :- t(X). Program Modules:
1
Mq= { q(X) :- a(X), not b(X).}
2
Mp = { p(X) :- q(X).}
3
Mr,g = {g(X) :- t(X). % exit g(X) :- q(X), r(X). % recursive r(X) :- q(X), g(X). % recursive }
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Given a program with few rules (r) col(X, red) | col(X, yellow) | col(X, green) :- node(X). (c) :- edge(X, Y), col(X, C), col(Y, C).
Instance: node(1). node(2). node(3). node(4). node(5). node(6). . . . edge(1, 2). edge(2, 3). edge(1, 3). edge(3, 4). . . . Instantiation: col(1, red) | col(1, yellow) | col(1, green) :- node(1). col(2, red) | col(2, yellow) | col(2, green) :- node(2). col(3, red) | col(3, yellow) | col(3, green) :- node(3). col(3, red) | col(3, yellow) | col(3, green) :- node(4). col(3, red) | col(3, yellow) | col(3, green) :- node(5). col(3, red) | col(3, yellow) | col(3, green) :- node(6). . . .
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Given a program with few rules (r) col(X, red) | col(X, yellow) | col(X, green) :- node(X). (c) :- edge(X, Y), col(X, C), col(Y, C).
Instance: node(1). node(2). node(3). node(4). node(5). node(6). Instantiation: col(1, red) | col(1, yellow) | col(1, green) :- node(1). col(2, red) | col(2, yellow) | col(2, green) :- node(2). col(3, red) | col(3, yellow) | col(3, green) :- node(3). col(3, red) | col(3, yellow) | col(3, green) :- node(4). col(3, red) | col(3, yellow) | col(3, green) :- node(5). col(3, red) | col(3, yellow) | col(3, green) :- node(6). . . .
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
→ Selection heuristic
2
→ Load balancing/granularity control
3
→ Resort to the thread scheduler → i.e., number of splits > available CPUs → Exploit OS resources, minimize locks
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
C(k) = if k < 2 T(L1) · T(L2) if k = 2 C(k − 1) + T(L1 ⋊ ⋉ · · · ⋊ ⋉ Lk−1) · T(k)
(1) Ci ≤ Cj, 0 ≤ i < j < n (2)
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
C(k) = if k < 2 T(L1) · T(L2) if k = 2 C(k − 1) + T(L1 ⋊ ⋉ · · · ⋊ ⋉ Lk−1) · T(k) if k > 2 (1) Ci = C(n) − C(i − 1) si + C(i − 1), 1 ≤ i ≤ n (2) Ci ≤ Cj, 0 ≤ i < j < n (3) where si is the maximum number of splits allowed by the i-th literal Cj denotes the cost of splitting on the i-th literal
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
C(k) = if k < 2 T(L1) · T(L2) if k = 2 C(k − 1) + T(L1 ⋊ ⋉ · · · ⋊ ⋉ Lk−1) · T(k) if k > 2 (1) Ci = C(n) − C(i − 1) si + C(i − 1), 1 ≤ i ≤ n (2) Ci ≤ Cj, 0 ≤ i < j < n (3) where si is the maximum number of splits allowed by the i-th literal Cj denotes the cost of splitting on the i-th literal
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
→ the same as in DLV
2
→ Cheap to compute!
3
→ The first step applies in most cases
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
(if split instantiation requires less than starting threads!)
(granularity control)
(load balancing)
(load balancing)
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
(if split instantiation requires less than starting threads!)
(granularity control)
(load balancing)
(load balancing)
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
W(r) < wseq → do not split (serial) ”very easy rule”
”even easier rules”
W(r) < whard → equally sized splits “regular rules” W(r) > whard → equally sized splits + final redistribution “very hard rules”
unary split size for the last total_split_number(r) − ncpu splits intuition: hard rule hard unbalanced splits → → further subdivide the last part of the computation
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
thread creation: time&space waste lock contention (shared resources, output stream): starvation
Master thread + a thread pool for each level of parallelism A buffer between each level of parallelism An output buffer to reduce lock contention
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
n-Queens (n ∈ {37, 39, 41, 43, 45})
2
Ramsey Numbers (ramsey(7,7) and n ∈ 31, 32, 33, 34, 35)
3
Clique (Random generated graphs)
4
Sudoku (ASP Competition 2009)
5
Golomb Ruler (ASP Competition 2009)
6
Reachability (Random trees)
7
Hamiltonian Path (Simplex graphs - P . Simons)
8
3-Colorability (Simplex graphs - Stanford GraphBase)
9
Real World Applications: Timetabling, Food
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
(a) Hamiltonian Path (b) 3-Colorability
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Food: super-linear speedup, efficiency > 1070
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
“for any algorithm, any elevated performance over one class of problems is offset by performance over another class” [WM97]
Automatically select the “best” algorithm on a per-instance basis [Ric76]
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Problem space P , a feature space F Set A of considered algorithms, and the performance space Y Find the selection mapping S(f(x)) maximizing the performance mapping y(a(x)) → Usually done learning S(.) from a training set
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Based on regression and dynamic features White box: Combines variants of clasp Winner of several ASP competitions
Based on classification Cheap to compute static features Black Box: Combines several solvers
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Based on regression and dynamic features White box: Combines variants of clasp Winner of several ASP competitions
Based on classification Cheap to compute static features Black Box: Combines several solvers
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Based on regression and dynamic features White box: Combines variants of clasp Winner of several ASP competitions
Based on classification Cheap to compute static features Black Box: Combines several solvers
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
1
Use GRINGO and DLV Apply ME-ASP approach to grounders
2
Improve solver “prediction” using non-ground features Classify programs w.r.t. non-ground features Apply algorithm selection on each sub-class Can be combined with selection of grounder
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Features Problem size, balance and proximity to Horn features Presence of queries Strongly Connected Components Number of Head-Cycle Free (HCF) and non-HCF components Features indicating if the program is recursive, tight, stratified Number of builtins ... PART algorithm for classification Supervised classification algorithm Patterns are the feature vectors Classes (labels) are the grounders
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Solver Grounder P NP Total # Time # Time # Time DLV-g 48 128.26 93 62.65 141 84.99 Clasp GRINGO 35 97.21 72 32.69 107 53.80
SELECTOR
48 70.94 95 59.64 143 63.43 DLV-g 46 130.47 86 82.42 132 99.16 Cmodels3 GRINGO 32 116.29 67 58.44 99 77.14
SELECTOR
46 70.60 87 80.72 133 77.22 DLV-g 41 129.17 59 71.39 100 95.08 DLV GRINGO 31 107.89 37 28.53 68 64.71
SELECTOR
41 71.26 59 69.50 100 70.22 DLV-g 43 136.54 92 71.13 135 91.96
IDP
GRINGO 32 140.57 72 46.97 104 75.77
SELECTOR
43 74.16 94 70.19 137 71.43
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ One has to deal with queries → Encoding contains rules having large bodies
(e.g., ≥ 4 literals)
→ and the program has a coarse structure
(few components)
→ Program is non-disjunctive and recursive with many
components
→ Most of the rules have a short body
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Orthogonal to solver selection
→ Makes difficult to freely exchange solvers and grounders → It can be done on a restricted language → Conversions reduce efficiency
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Extend feature set to “cover” ASP Core 2.0 → Improve solver “prediction” using non-ground features
1
Classify programs w.r.t. non-ground features
2
Apply solver selection on each sub-class
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Solver C1 C2 C3 C4 CQ
CLASP
LP2BV2+BOOLECTOR
– –
LP2GRAPH
– –
LP2MAXSAT+CLASP
–
LP2NORMAL2+CLASP
LP2SAT3+GLUCOSE
– – –
LP2SAT3+LINGELING
– – –
WASP1
–
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ Training & Testing: Randomly split in half the 8572 instances
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
→ ASP On-Site Modeling Competition
→ The 6th ASP Competition Report and winners announcement!
→ For more details on ME-ASP ML
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Mario Alviano, Carmine Dodaro, Nicola Leone, and Francesco Ricca. Advances in WASP. In Logic Programming and Nonmonotonic Reasoning - 13th International Conference, LPNMR 2015, Lexington, KY, USA, September 27-30, 2015. Proceedings, volume 9345 of Lecture Notes in Computer Science, pages 40–54. Springer, 2015. Marcello Balduccini, Enrico Pontelli, Omar El-Khatib, and Hung Le. Issues in parallel execution of non-monotonic reasoning systems. Parallel Computing, 31(6):608–647, 2005. Ozan Caldiran, Kadir Haspalamutgil, Abdullah Ok, Can Palaz, Esra Erdem, and Volkan Patoglu. Bridging the gap between high-level reasoning and low-level control. In LPNMR, pages 342–354, 2009.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Francesco Calimeri, Simona Perri, and Francesco Ricca. Experimenting with parallelism for the instantiation of ASP programs.
Evgeny Dantsin, Thomas Eiter, Georg Gottlob, and Andrei Voronkov. Complexity and Expressive Power of Logic Programming. ACM Computing Surveys, 33(3):374–425, 2001. Agostino Dovier, Andrea Formisano, Enrico Pontelli, and Flavio Vella. Parallel execution of the ASP computation - an investigation on gpus. In Vos et al. [VELT15]. James P . Delgrande, Torsten Grote, and Aaron Hunter. A general approach to the verification of cryptographic protocols using answer set programming. In LPNMR, pages 355–367, 2009.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Halit Erdogan, Esra Erdem, and Olivier Bodenreider. Exploiting umls semantics for checking semantic consistency among umls concepts. In In Proc. of MedInfo, 2010. Thomas Eiter, Wolfgang Faber, Nicola Leone, and Gerald Pfeifer. The diagnosis frontend of the dlv system. AI Commun., 12(1-2):99–111, 1999.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Enrico Ellguth, Martin Gebser, Markus Gusowski, Benjamin Kaufmann, Roland Kaminski, Stefan Liske, Torsten Schaub, Lars Schneidenbach, and Bettina Schnor. A simple distributed conflict-driven answer set solver. In Esra Erdem, Fangzhen Lin, and Torsten Schaub, editors, Logic Programming and Nonmonotonic Reasoning, 10th International Conference, LPNMR 2009, Potsdam, Germany, September 14-18, 2009. Proceedings, volume 5753 of Lecture Notes in Computer Science, pages 490–495. Springer, 2009. Thomas Eiter, Georg Gottlob, and Heikki Mannila. Disjunctive datalog. ACM Trans. Database Syst., 22(3):364–418, 1997.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Thomas Eiter, Giovambattista Ianni, Roman Schindlauer, and Hans Tompits. Effective integration of declarative rules with external evaluations for semantic-web reasoning. In ESWC, pages 273–287, 2006. Jean Gressmann, Tomi Janhunen, Robert E. Mercer, Torsten Schaub, Sven Thiele, and Richard Tichy. On probing and multi-threading in platypus. In ECAI 2006, 17th European Conference on Artificial Intelligence, August 29 - September 1, 2006, Riva del Garda, Italy, Including Prestigious Applications of Intelligent Systems (PAIS 2006), Proceedings, pages 392–396, 2006.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Martin Gebser, Benjamin Kaufmann, André Neumann, and Torsten Schaub. Conflict-driven answer set solving. In Twentieth International Joint Conference on Artificial Intelligence (IJCAI-07), pages 386–392. Morgan Kaufmann Publishers, January 2007. Martin Gebser, Benjamin Kaufmann, and Torsten Schaub. Multi-threaded ASP solving with clasp. TPLP, 12(4-5):525–545, 2012. Michael Gelfond and Vladimir Lifschitz. Classical negation in logic programs and disjunctive databases. New Generation Comput., 9(3/4):365–386, 1991.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Martin Gebser, Marco Maratea, and Francesco Ricca. The design of the sixth answer set programming competition. In Logic Programming and Nonmonotonic Reasoning - 13th International Conference, LPNMR 2015, Lexington, KY, USA, September 27-30, 2015. Proceedings, pages 531–544, 2015. Terracina Giorgio, Leone Nicola, and Martello Alessandra. Logic-based techniques for data cleaning: an application to the italian national healthcare system. LPNMR, 2013. Holger H. Hoos, Roland Kaminski, Marius Thomas Lindauer, and Torsten Schaub. aspeed: Solver scheduling via answer set programming. TPLP, 15(1):117–142, 2015.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Holger Hoos, Marius Thomas Lindauer, and Torsten Schaub. claspfolio 2: Advances in algorithm selection for answer set programming. TPLP, 14(4-5):569–585, 2014. Tomi Janhunen. Some (in)translatability results for normal logic programs and propositional theories. Journal of Applied Non-Classical Logics, 16(1-2):35–86, 2006.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Nicola Leone, Gianluigi Greco, Giovambattista Ianni, Vincenzino Lio, Giorgio Terracina, Thomas Eiter, Wolfgang Faber, Michael Fink, Georg Gottlob, Riccardo Rosati, Domenico Lembo, Maurizio Lenzerini, Marco Ruzzi, Edyta Kalka, Bartosz Nowicki, and Witold Staniszkis. The INFOMIX system for advanced integration of incomplete and inconsistent data. In Proceedings of the ACM SIGMOD International Conference on Management of Data, Baltimore, Maryland, USA, June 14-16, 2005, pages 915–917. ACM, 2005. Vladimir Lifschitz. Answer set programming and plan generation.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Yuliya Lierler and Marco Maratea. Cmodels-2: SAT-based Answer Set Solver Enhanced to Non-tight Programs. In Vladimir Lifschitz and Ilkka Niemelä, editors, Proceedings of the 7th International Conference on Logic Programming and Non-Monotonic Reasoning (LPNMR-7), volume 2923 of LNAI, pages 346–350. Springer, January 2004. Nicola Leone, Gerald Pfeifer, Wolfgang Faber, Thomas Eiter, Georg Gottlob, Simona Perri, and Francesco Scarcello. The DLV System for Knowledge Representation and Reasoning. TOCL, 7(3):499–562, July 2006.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Nicola Leone and Francesco Ricca. Answer set programming: A tour from the basics to advanced development tools and industrial applications. In Reasoning Web. Web Logic Rules - 11th International Summer School 2015, Berlin, Germany, July 31 - August 4, 2015, Tutorial Lectures, volume 9203 of Lecture Notes in Computer Science, pages 308–326. Springer, 2015. Marco Maratea, Luca Pulina, and Francesco Ricca. Automated selection of grounding algorithm in answer set programming. In AI*IA 2013: Advances in Artificial Intelligence - XIIIth International Conference of the Italian Association for Artificial Intelligence, Turin, Italy, December 4-6, 2013. Proceedings, volume 8249 of Lecture Notes in Computer Science, pages 73–84. Springer, 2013.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Marco Maratea, Luca Pulina, and Francesco Ricca. A multi-engine approach to answer-set programming. TPLP, 14(6):841–868, 2014. Marco Maratea, Luca Pulina, and Francesco Ricca. Advances in multi-engine ASP solving. In AI*IA 2015, Advances in Artificial Intelligence - XIVth International Conference of the Italian Association for Artificial Intelligence, Ferrara, Italy, September 23-25, 2015, Proceedings, volume 9336 of Lecture Notes in Computer Science, pages 179–190. Springer, 2015.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Marco Maratea, Luca Pulina, and Francesco Ricca. Multi-level algorithm selection for ASP. In Logic Programming and Nonmonotonic Reasoning - 13th International Conference, LPNMR 2015, Lexington, KY, USA, September 27-30, 2015. Proceedings, volume 9345 of Lecture Notes in Computer Science, pages 439–445. Springer, 2015. Marco Manna, Francesco Ricca, and Giorgio Terracina. Taming primary key violations to query large inconsistent data via ASP. TPLP, 15(4-5):696–710, 2015. Enrico Pontelli, Hung Viet Le, and Tran Cao Son. An investigation in parallel execution of answer set programs on distributed memory platforms: Task sharing and dynamic scheduling. Computer Languages, Systems & Structures, 36(2):158–202, 2010.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Simona Perri, Francesco Ricca, and Marco Sirianni. Parallel instantiation of ASP programs: techniques and experiments. TPLP, 13(2):253–278, 2013. Francesco Ricca, Antonella Dimasi, Giovanni Grasso, Salvatore Maria Ielpa, Salvatore Iiritano, Marco Manna, and Nicola Leone. A logic-based system for e-tourism.
Francesco Ricca, Giovanni Grasso, Mario Alviano, Marco Manna, Vincenzino Lio, Salvatore Iiritano, and Nicola Leone. Team-building with answer set programming in the gioia-tauro seaport. TPLP, 12(3):361–381, 2012.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
John R. Rice. The algorithm selection problem. Advances in Computers, 15:65–118, 1976. Chiaki Sakama. Dishonest reasoning by abduction. In IJCAI, pages 1063–1064, 2011. Timo Soininen and Ilkka Niemelä. Developing a declarative rule language for applications in product configuration. In PADL, pages 305–319, 1999. Ilias Tachmazidis, Grigoris Antoniou, and Wolfgang Faber. Efficient computation of the well-founded semantics over big data. TPLP, 14(4-5):445–459, 2014.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Marina De Vos, Thomas Eiter, Yuliya Lierler, and Francesca Toni, editors. Proceedings of the Technical Communications of the 31st International Conference on Logic Programming (ICLP 2015), Cork, Ireland, August 31 - September 4, 2015, volume 1433 of CEUR Workshop Proceedings. CEUR-WS.org, 2015. David Wolpert and William G. Macready. No free lunch theorems for optimization. IEEE Trans. Evolutionary Computation, 1(1):67–82, 1997. Johan Wittocx, Maarten Mariën, and Marc Denecker. The IDP system: a model expansion system for an extension of classical logic. In Marc Denecker, editor, Logic and Search, Computation of Structures from Declarative Descriptions (LaSh 2008), pages 153–165, Leuven, Belgium, November 2008.
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving
Introduction Parallel Instantiation Grounding in Multi-Engine ASP-System Conclusion
Mohan Yang, Alexander Shkapsky, and Carlo Zaniolo. Parallel bottom-up evaluation of logic programs: Deals on shared-memory multicore machines. In Vos et al. [VELT15].
Francesco Ricca Grounding Issues in Parallel and Multi-Engine ASP Solving