190 likes | 300 Views
Formal Methods of Systems Specification Logical Specification of Hard- and Software. Prof. Dr. Holger Schlingloff Institut für Informatik der Humboldt Universität and Fraunhofer Institut für Rechnerarchitektur und Softwaretechnik. Recap: Propositional Logic.
E N D
Formal Methods of Systems SpecificationLogical Specification of Hard- and Software Prof. Dr. Holger Schlingloff Institut für Informatik der Humboldt Universität and Fraunhofer Institut für Rechnerarchitektur und Softwaretechnik
Recap: Propositional Logic • Syntax PL ::= Ρ | | (PL PL) • Semantics • Propositional Model M: (U,I); Interpretation I: Ρ↦ U • Validation relation M ⊨ • M ⊨ p if I(p)=true • M ⊭ • M ⊨ () if M ⊨ implies M ⊨ • Validity (⊨), Satifiability(SAT()) • Calculus • axiom schemes: weakening, distribution, excluded middle • rule: modus ponens
Q.: exponential blowup? (rs) = ((rs) (sr)) (q(rs)) = ((q(rs)) ((rs)q)) = ((q ((rs) (sr))) (((rs) (sr)) q)) (p(q(rs))) = (p(q(rs))) ((q(rs))p) = (p((q((rs) (sr))) (((rs) (sr))q))) (((q((rs) (sr))) (((rs) (sr))q))p) p (q r) (¬q s) q (t u) p ((t u) r) (¬ (t u) s) logarithmic reduction by introduction of abbreviations?
Boolean Quantification QPL ::= Ρ | | (QPL QPL) | Ρ QPL • intuitively, p (p) means p is „hidden“ • M ⊨ p if there is an M‘= (U,I‘) such that I‘(q)=I(q) for all qp and M‘ ⊨ • ⊨ p ((p:=) (p:=Τ)) • ∀p = ¬p ¬; ⊨∀p ((p:=) (p:=Τ))
Conciseness and Complexity • Quantified formulae can be logarithmically more concise • = [… ((rs)t) … ((rs)t) … ((rs)t) …] • (p(p ((rs)t)) [… p … p … p …]) • Quantified formulae can be exponentially harder to analyze • SAT(PL) is NP-hard, SAT(QPL) is PSPACE-hard • To check whether p holds one has to check both alternatives
Boolean Normal Forms • DNF (disjunctive normal form) • each formula is equivalent to a disjunction of conjunction of literals e.g. ((p¬ qr) (¬pq¬r) (p¬q¬r)) • obtained by truth table • CNF (conjunctive normal form) • de Morgan dual of DNF • used in PLAs (programmable logic arrays) • NAND-, NOR-normal form • (p|q)=(p¬q); ¬p=(p|p); (pq)=(p|¬q) • used for gate arrays • Algebraic normal form • XOR of conjunction of (positive) propositions • used in linear feedback shift registers
next week: tree normal form • (ordering of variables)
Boolean Modelling of Reactive Systems • Many modelling formalisms are being used • Transition systems • Parallel and hierarchical transition systems, statecharts • Shared variables programs • UML diagrams • Abstract state machines • ...
Transition Systems • Transition system TS=(,S, , S0), where • is a nonempty finite alphabet • S is a nonempty finite set of states • S S is the transition relation, and • S0 S is the set of initial states • similar to a nondeterministic finite automaton, with many initial states but without finite states • transition system generates a (finite or infinite) word w0w1w2... iff there are states s0s1s2s3... such that s0 S0 and each (si,wi,si+1) Δ
off tape play dn up up up dn dn memory dn Example • ={up, dn} • S={off, tape, memory, play} • ={(off,dn,tape), (tape,up,off), (tape,dn,memory), (memory,up,off), (memory,dn,play), (play,dn,tape), (play,up,off)} • S0={off}
Parallel Transition Systems • Parallel transition system T=(T1,…,Tn) • each Ti is a transition system • SiSj= • interleaving semantics • on its private alphabet, each Ti can make an independent move • synchronization is via common events • example: power switch and camcorder mode
play tape camera switch on but_hi but_lo off dn memory dn,pwr_res dn up up,pwr_fail dn dn on dn Example • T=(switch, camera) • {pwr_fail, pwr_res} are private to camera • synchronization alphabet {up,dn} • how big is the state space?
The global transition system T associated with a parallel transition system (T1,…,Tn) is defined as T=(, S, , S0), where • = i • S= S1 … Sn • S0 = S1,0 … Sn,0, and • ((s1,…,sn),a,(s1‘,…,sn‘)) iff for all Ti • if ai, then ((s i),a,(s i‘))i, and • if ai, then s i=s i‘.
Shared Variables • A shared variables program is given by a tuple (V,D,T,s0), where • V=(v1,…,vn) is a set of program variables • D=(D1,…,Dn) is a tuple of corresponding finite domains Di={di1,…,dim} • TDD is a transition relation, and • s0 = (d11,…,dn1) is the initial state
Example (1) • A request granting algorithm • V={request,state} • D=({true, false}, {ready, busy}) • T=(((true, ready), (true, busy)), ((false, ready), (false, ready)), ((true, busy), (true, busy)), ((true, busy), (true, ready)), …)
Example (2) • Euclidean algorithm gcd(a, b) if a = 0 return b while b ≠ 0 if a > b then a := a − b else b := b − a return a • Shared variables program • V=(a,b) • D=(Nat, Nat) (finite?) • T={((0,0),(0,0)), ((7,4),(3,4)), ((3,4),(3,1)), …} • s0=D
Transition Systems and Programs • For every (parallel) transition system there is an equivalent shared variables program of the same order of size. • The translation in the other direction may cause an exponential blowup. • Exercise: describe the translations!In which sense are the translations equivalent?
Transition Relation as Boolean Formula • State=(d1,…,dn) (diDi) • Transition relation T can be defined by a propositional formula T • atomic propositions: let V‘={v1‘,…,vn‘}P={(x=y) | x,y (VV‘Di)} • Any propositional formula T in this alphabet defines a transition relation via the following convention If s=(d1,…,dn) and s‘=(d1‘,…dn‘), then (s,s‘) T iff M⊨ T, where I(vi)=di and I(vi‘)=di‘.
Example • request granting algorithm • V={request,state} • D=({true, false}, {ready, busy}) • T=((request=true)(state=ready)(state‘=busy)) • „Propositional logic as a programming language“ • Used in model checkers such as nuSMV