Regularity Problems of Process Rewrite Systems Fei Yang 1 (Based on - - PowerPoint PPT Presentation

regularity problems of process rewrite systems
SMART_READER_LITE
LIVE PREVIEW

Regularity Problems of Process Rewrite Systems Fei Yang 1 (Based on - - PowerPoint PPT Presentation

Introduction REG and REG for tnPA Remarks Regularity Problems of Process Rewrite Systems Fei Yang 1 (Based on joint work with Yuxi Fu 2 ) 1 Department of Mathematics and Computer Science Eindhoven University of Technology 2 BASICS,


slide-1
SLIDE 1

Introduction ≈REG and ≃REG for tnPA Remarks

Regularity Problems of Process Rewrite Systems

Fei Yang1 (Based on joint work with Yuxi Fu2)

1Department of Mathematics and Computer Science

Eindhoven University of Technology

2BASICS, Department of Computer Science and Engineering

Shanghai Jiao Tong University

FSA Colloquium November 11, 2014

Fei Yang Regularity Problems of Process Rewrite Systems 1/33

slide-2
SLIDE 2

Introduction ≈REG and ≃REG for tnPA Remarks

Outline

1

Introduction Overview Process Rewrite System

2

≈REG and ≃REG for tnPA Semi-Decidability Witness of Infinity Finite Witness

3

Remarks Related Work Open Problems

Fei Yang Regularity Problems of Process Rewrite Systems 2/33

slide-3
SLIDE 3

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Outline

1

Introduction Overview Process Rewrite System

2

≈REG and ≃REG for tnPA Semi-Decidability Witness of Infinity Finite Witness

3

Remarks Related Work Open Problems

Fei Yang Regularity Problems of Process Rewrite Systems 3/33

slide-4
SLIDE 4

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

A hardware system is always a finite state system. A software system has potentially infinite states, no matter how simple its functionality may be.

Fei Yang Regularity Problems of Process Rewrite Systems 4/33

slide-5
SLIDE 5

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Three Verification Problems

Equivalence Input: processes α, β, equivalence relation ∼ =. Problem: α ∼ = β ? Finiteness Input: process α, FS γ, equivalence relation ∼ =. Problem: α ∼ = γ ? Regularity Input: process α, equivalence relation ∼ =. Problem: ∃ FS γ. α ∼ = γ ?

Fei Yang Regularity Problems of Process Rewrite Systems 5/33

slide-6
SLIDE 6

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Equivalence Checking

Is an implementation correct w.r.t. a specification?

Finiteness Checking

Is a hardware design correct w.r.t. a specification?

Regularity Checking

Is a specification implementable by hardware?

Fei Yang Regularity Problems of Process Rewrite Systems 6/33

slide-7
SLIDE 7

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Several Bisimulation Relations

1

∼ Strong Bisimulation [Par81]

2

≈ Weak Bisimulation [Mil89]

3

≃ Branching Bisimulation [vGW96]

Fei Yang Regularity Problems of Process Rewrite Systems 7/33

slide-8
SLIDE 8

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Several Bisimulation Relations

1

∼ Strong Bisimulation [Par81]

2

≈ Weak Bisimulation [Mil89]

3

≃ Branching Bisimulation [vGW96]

Fei Yang Regularity Problems of Process Rewrite Systems 7/33

slide-9
SLIDE 9

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Several Bisimulation Relations

1

∼ Strong Bisimulation [Par81]

2

≈ Weak Bisimulation [Mil89]

3

≃ Branching Bisimulation [vGW96]

Fei Yang Regularity Problems of Process Rewrite Systems 7/33

slide-10
SLIDE 10

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Syntax

Definition Act = {a, b, . . .} is a set of atomic actions; Const = {ǫ} ∪ {X, Y, Z, . . .} is a set of process constants. S = {α1, α2, . . .} is the set of process terms, which describe the states

  • f the system, are generated from the following BNF:

α ::= ǫ | X | α1.α2 | α1 α2 where ǫ is the empty process; α1.α2 is a sequential process; α1 α2 is a parallel process. We shall write α, β, γ, . . . for process terms.

Fei Yang Regularity Problems of Process Rewrite Systems 8/33

slide-11
SLIDE 11

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Operational Semantics

The transition relation → is generated from a finite set ∆ of transition rules of the form α

a

− → β. For every a ∈ Act, the transition relation

a

− → is the smallest relation constructed from the following inference rules: α

a

− → β ∈ ∆ α

a

− → β α

a

− → α′ α.β

a

− → α′.β α

a

− → α′ α β

a

− → α′ β β

a

− → β′ α β

a

− → α β′ The parallel composition α β is different from concurrent

  • composition. No communication between α and β is admitted.

Fei Yang Regularity Problems of Process Rewrite Systems 9/33

slide-12
SLIDE 12

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Classification of Process Term

Definition We distinguish four classes of process term: 1: Terms consisting of a single process constant like X. S: Terms consisting of a single constant or a sequential composition of process constants like X.Y.Z. P: Terms consisting of a single constant or a parallel composition of process constants like X Y Z. G: Terms with arbitrary sequential and parallel composition like (X.(Y Z)) W.

Fei Yang Regularity Problems of Process Rewrite Systems 10/33

slide-13
SLIDE 13

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Classification of PRS

Definition Let Ξ, Π ∈ {1, S, P, G}. A (Ξ, Π)-PRS is a finite set of rules ∆ satisfying the following: for every rewrite rule α

a

− → β ∈ ∆ Ξ ⊆ Π, α ∈ Ξ \ {ǫ}, β ∈ Π, the initial state is given as a term α0 ∈ Ξ. A (G, G)-PRS is simply called a PRS. W.l.o.g. it can be assumed that the initial state α0 of a PRS is a single constant.

Fei Yang Regularity Problems of Process Rewrite Systems 11/33

slide-14
SLIDE 14

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Example of BPA

Recursive specification: X def = a.X.B + c.ǫ B def = b.ǫ The following is a (1, S)-PRS transition system with initial state X: X

a

− → X.B X

c

− → ǫ B

b

− → ǫ This process recognises the language {ancbn : n ≥ 0}.

Fei Yang Regularity Problems of Process Rewrite Systems 12/33

slide-15
SLIDE 15

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

PRS Hierarchy

PRS (G, G) PAD (S, G) PAN (P, G) PDA (S, S) PA (1, G) PN (P, P) BPA (1, S) BPP (1, P) FS (1, 1)

Fei Yang Regularity Problems of Process Rewrite Systems 13/33

slide-16
SLIDE 16

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

Normed Process

Definition A process is normed if it can reach ǫ after a finite number of transition steps. We say a process is totally normed if it is normed and can reach ǫ after at least one non-τ transition step. We write for example nBPA for normed BPA and tnBPA for totally normed BPA.

Fei Yang Regularity Problems of Process Rewrite Systems 14/33

slide-17
SLIDE 17

Introduction ≈REG and ≃REG for tnPA Remarks Overview Process Rewrite System

PA

PA nPA tnPA ∼REG ? NL [Kuˇ c96] NL [Kuˇ c96] PSPACE-H [Srb02a] NL-H [Srb02a] NL-H [Srb02a] ≃REG ? ? P EXPT-H [May03] PSPACE-H [Srb03] NL-H [Srb02a] ≈REG ? ? P EXPT-H [May03] PSPACE-H [Srb03] NL-H [Srb02a]

Table: PA Regularity

Fei Yang Regularity Problems of Process Rewrite Systems 15/33

slide-18
SLIDE 18

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Outline

1

Introduction Overview Process Rewrite System

2

≈REG and ≃REG for tnPA Semi-Decidability Witness of Infinity Finite Witness

3

Remarks Related Work Open Problems

Fei Yang Regularity Problems of Process Rewrite Systems 16/33

slide-19
SLIDE 19

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

A PA Process

X a − → X X X b − → X.X X a − → ǫ X τ − → ǫ X− → ∗ X.(X (X.X)) X− → ∗ X X (X.(X X))

Fei Yang Regularity Problems of Process Rewrite Systems 17/33

slide-20
SLIDE 20

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

A PA Process

X a − → X X X b − → X.X X a − → ǫ X τ − → ǫ X− → ∗ X.(X (X.X)) X− → ∗ X X (X.(X X))

Fei Yang Regularity Problems of Process Rewrite Systems 17/33

slide-21
SLIDE 21

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

∼REG for nPA is in polynomial time [Kuˇ c96]. How about ≈REG and ≃REG? Why not adding totally normed constraint?

Fei Yang Regularity Problems of Process Rewrite Systems 18/33

slide-22
SLIDE 22

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

∼REG for nPA is in polynomial time [Kuˇ c96]. How about ≈REG and ≃REG? Why not adding totally normed constraint?

Fei Yang Regularity Problems of Process Rewrite Systems 18/33

slide-23
SLIDE 23

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Lemma (Semi-Decidability) Given a process (α, ∆), if ∼ = or ∼ =FS is decidable, then ∼ =REG is semi-decidable.

Fei Yang Regularity Problems of Process Rewrite Systems 19/33

slide-24
SLIDE 24

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Infinite Transition Path

Lemma (Infinite ∼ = Path) Given a process (α, ∆), it is non-regular w.r.t. ∼ =, iff there exists an infinite transition path α

a0

− → α1

a1

− → α2

a2

− → . . . with αi ∼ = αj, for any i = j.

Fei Yang Regularity Problems of Process Rewrite Systems 20/33

slide-25
SLIDE 25

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Lemma (Infinite ≡ Path) Given a process (α, ∆), it is non-regular w.r.t. ∼ =, then there exists an infinite transition path α

a0

− → α1

a1

− → α2

a2

− → . . . with αi ≡ αj, for any i = j. Lemma (Infinite Norm Increasing Path) Given a process (α, ∆), it is non-regular w.r.t. ∼ =, if there exists an infinite transition path α

a0

− → α1

a1

− → α2

a2

− → . . . s.t., for any i, there exists j, with norm(αi) < norm(αj).

Fei Yang Regularity Problems of Process Rewrite Systems 21/33

slide-26
SLIDE 26

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Lemma (Infinite ≡ Path) Given a process (α, ∆), it is non-regular w.r.t. ∼ =, then there exists an infinite transition path α

a0

− → α1

a1

− → α2

a2

− → . . . with αi ≡ αj, for any i = j. Lemma (Infinite Norm Increasing Path) Given a process (α, ∆), it is non-regular w.r.t. ∼ =, if there exists an infinite transition path α

a0

− → α1

a1

− → α2

a2

− → . . . s.t., for any i, there exists j, with norm(αi) < norm(αj).

Fei Yang Regularity Problems of Process Rewrite Systems 21/33

slide-27
SLIDE 27

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Finite Witness

Fire(α) =        ∅ if α = ǫ {X} if α = X Fire(β1) if α = β1.β2 Fire(β1) ∪ Fire(β2) if α = β1 β2

Fei Yang Regularity Problems of Process Rewrite Systems 22/33

slide-28
SLIDE 28

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Finite Witness

Definition Given a tnPA process (α, ∆), a process constant X ∈ Var(∆) is a growing constant if there exists β s.t., X− → ∗β, with X ∈ Fire(β) and length(β) ≥ 2. Lemma Given a tnPA process (α, ∆), it is ≃reg and ≈reg, iff it cannot activate any growing constant.

Fei Yang Regularity Problems of Process Rewrite Systems 23/33

slide-29
SLIDE 29

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Finite Witness

Definition Given a tnPA process (α, ∆), a process constant X ∈ Var(∆) is a growing constant if there exists β s.t., X− → ∗β, with X ∈ Fire(β) and length(β) ≥ 2. Lemma Given a tnPA process (α, ∆), it is ≃reg and ≈reg, iff it cannot activate any growing constant.

Fei Yang Regularity Problems of Process Rewrite Systems 23/33

slide-30
SLIDE 30

Introduction ≈REG and ≃REG for tnPA Remarks Semi-Decidability Witness of Infinity Finite Witness

Polynomial Time Decidability

Theorem There is a polynomial time algorithm deciding whether there is a growing constant. The time complexity is O(n3 + mn), where n is the number of transition rules, and m is the maximum length of the rules. Theorem ≈REG and ≃REG for tnPA can be decided in polynomial time.

Fei Yang Regularity Problems of Process Rewrite Systems 24/33

slide-31
SLIDE 31

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Outline

1

Introduction Overview Process Rewrite System

2

≈REG and ≃REG for tnPA Semi-Decidability Witness of Infinity Finite Witness

3

Remarks Related Work Open Problems

Fei Yang Regularity Problems of Process Rewrite Systems 25/33

slide-32
SLIDE 32

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

BPA

BPA nBPA ∼REG Decidable [BCS95, BCS96] PSPACE-H [Srb02b] NL-C [Kuˇ c96, Srb02b] ≃REG ? EXPT-H [May03] Decidable [Fu13] NP-H [Srb03, Stˇ r98] ≈REG ? EXPT-H [May03] ? NP-H [Srb03, Stˇ r98]

Table: BPA Regularity

Fei Yang Regularity Problems of Process Rewrite Systems 26/33

slide-33
SLIDE 33

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

BPP

BPP nBPP ∼REG PSPACE-C [Kot05, JE96, Srb02a] NL [Kuˇ c96] NL-H [Srb02a] ≃REG ? ? ≈REG ? PSPACE-H [Srb03] ? PSPACE-H [Srb03]

Table: BPP Regularity

Fei Yang Regularity Problems of Process Rewrite Systems 27/33

slide-34
SLIDE 34

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

PDA

PDA nPDA ∼REG ? EXPT-H [KM02, Srb02b] P [EHRS00] NL-H [Srb02b] ≃REG ? ? ≈REG ? EXPT-H [KM02, Srb02b] ? EXPT-H [KM02, Srb02b]

Table: PDA Regularity

Fei Yang Regularity Problems of Process Rewrite Systems 28/33

slide-35
SLIDE 35

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

PN

PN nPN ∼REG Decidable [JE96] PSPACE-H [Srb02a] EXPSAPCE [Rac78] EXPSPACE-H [CLM76] ≃REG Undecidable ? EXPSPACE-H [CLM76] ≈REG Undecidable [JE96] EXPSPACE-H [CLM76] ? EXPSPACE-H [CLM76]

Table: PN Regularity

Fei Yang Regularity Problems of Process Rewrite Systems 29/33

slide-36
SLIDE 36

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Open Problems

Decidability for of ≃ for nBPP is proved in [CHL11]. Thus we have Corollary ≃REG for nBPP is semi-decidable. Is ≃REG decidable for nBPP?

Fei Yang Regularity Problems of Process Rewrite Systems 30/33

slide-37
SLIDE 37

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Open Problems

Decidability for of ≃ for nBPP is proved in [CHL11]. Thus we have Corollary ≃REG for nBPP is semi-decidable. Is ≃REG decidable for nBPP?

Fei Yang Regularity Problems of Process Rewrite Systems 30/33

slide-38
SLIDE 38

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Equivalence Checking of a Model with its Sub-model

For example: PN ∼ BPP

Fei Yang Regularity Problems of Process Rewrite Systems 31/33

slide-39
SLIDE 39

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Complexity Lower Bound

Is ∼REG for BPA EXPT-hard? Is the proof for ≈REG valid for ≃REG?

Fei Yang Regularity Problems of Process Rewrite Systems 32/33

slide-40
SLIDE 40

Introduction ≈REG and ≃REG for tnPA Remarks Related Work Open Problems

Complexity Lower Bound

Is ∼REG for BPA EXPT-hard? Is the proof for ≈REG valid for ≃REG?

Fei Yang Regularity Problems of Process Rewrite Systems 32/33

slide-41
SLIDE 41

Introduction ≈REG and ≃REG for tnPA Remarks

Thank You!

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-42
SLIDE 42

Introduction ≈REG and ≃REG for tnPA Remarks

  • O. Burkart, D. Caucal, and B. Steffen.

An elementary bisimulation decision procedure for arbitrary context-free processes. Mathematical Foundations of Computer Science, 969:423–433, 1995.

  • O. Burkart, D. Caucal, and B. Steffen.

Bisimulation collapse and the process taxonomy. In CONCUR’96: Concurrency Theory, pages 247–262. Springer, 1996.

  • W. Czerwi´

nski, P. Hofman, and S. Lasota. Decidability of branching bisimulation on normed commutative context-free processes. CONCUR 2011–Concurrency Theory, pages 528–542, 2011.

  • E. Cardoza, R. Lipton, and Albert R. Meyer.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-43
SLIDE 43

Introduction ≈REG and ≃REG for tnPA Remarks

Exponential space complete problems for petri nets and commutative semigroups (preliminary report). In Proceedings of the eighth annual ACM symposium on Theory

  • f computing, pages 50–54. ACM, 1976.
  • J. Esparza, D. Hansel, P. Rossmanith, and S. Schwoon.

Efficient algorithms for model checking pushdown systems. In Computer Aided Verification, pages 232–247. Springer, 2000.

  • Y. Fu.

Checking equality and regularity for normed BPA with silent moves. 2013.

  • P. Janˇ

car and J. Esparza. Deciding finiteness of petri nets up to bisimulation. Automata, Languages and Programming, pages 478–489, 1996.

  • A. Kuˇ

cera and R. Mayr.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-44
SLIDE 44

Introduction ≈REG and ≃REG for tnPA Remarks

On the complexity of semantic equivalences for pushdown automata and bpa. In Mathematical Foundations of Computer Science, pages 433–445. Springer, 2002.

  • M. Kot.

Regularity of BPP is PSPACE-complete. In Proceedings of the 3rd Ph. D. Workshop of Faculty of Electrical Engineering and Computer Science (WOFEX’05), pages 393–398, 2005.

  • A. Kuˇ

cera. Regularity is decidable for normed PA processes in polynomial time. In Foundations of Software Technology and Theoretical Computer Science, pages 111–122. Springer, 1996.

  • R. Mayr.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-45
SLIDE 45

Introduction ≈REG and ≃REG for tnPA Remarks

Weak bisimilarity and regularity of BPA is EXPTIME-hard. In Proccedings of the 10th International Workshop on Expressiveness in Concurrency (EXPRESS¡¯03), pages 160–143, 2003.

  • R. Milner.

Comunication and Concurrency. Prentice Hall, 1989.

  • D. Park.

Concurrency and automata on infinite sequences. Theoretical computer science, pages 167–183, 1981.

  • C. Rackoff.

The covering and boundedness problems for vector addition systems. Theoretical Computer Science, 6(2):223–231, 1978.

  • J. Srba.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-46
SLIDE 46

Introduction ≈REG and ≃REG for tnPA Remarks

Strong bisimilarity and regularity of basic parallel processes is PSPACE-hard. Lecture Notes in Computer Science, (2285):535–546, 2002.

  • J. Srba.

Strong bisimilarity and regularity of basic process algebra is pspace-hard. In Automata, Languages and Programming, pages 716–727. Springer, 2002.

  • J. Srba.

Complexity of weak bisimilarity and regularity for BPA and BPP. Electronic Notes in Theoretical Computer Science, 39(1):79–93, 2003.

  • J. Stˇ

ríAbrná. Hardness results for weak bisimilarity of simple process algebras.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33

slide-47
SLIDE 47

Introduction ≈REG and ≃REG for tnPA Remarks

Electronic Notes in Theoretical Computer Science, 18:179–190, 1998. R.J. van Glabbeek and W.P. Weijland. Branching time and abstraction in bisimulation semantics. Journal of the ACM (JACM), 43(3):555–600, 1996.

Fei Yang Regularity Problems of Process Rewrite Systems 33/33