Abstract
We provide a framework for compositional and iterative design and verification of systems with quantitative information, such as rewards, time or energy. It is based on disjunctive modal transition systems where we allow actions to bear various types of quantitative information. Throughout the design process, the actions can be further refined and the information made more precise. We show how to compute the results of standard operations on the systems, including the quotient (residual), which has not been previously considered for quantitative non-deterministic systems. Our quantitative framework has close connections to the modal nu-calculus and is compositional with respect to general notions of distances between systems and the standard operations.
Similar content being viewed by others
References
Aceto L, Ingólfsdóttir A, Larsen KG, Srba J (2007) Reactive systems. Cambridge University Press, Cambridge
Aceto L, Fábregas I, de Frutos-Escrig D, Ingólfsdóttir A, Palomino M (2013) On the specification of modal systems: a comparison of three frameworks. Sci Comput Program 78(12):2468–2487
Aliprantis CD, Border KC (2007) Infinite dimensional analysis: a hitchhiker’s guide. Springer, Berlin
Bauer SS, Fahrenberg U, Juhl L, Larsen KG, Legay A, Thrane C (2011) Quantitative refinement for weighted modal transition systems. In: Murlak F, Sankowski P (eds) MFCS, volume 6907 of Lecture Notes Computer Science. Springer, Berlin, pp 60–71
Bauer SS, David A, Hennicker R, Larsen KG, Legay A, Nyman U, Wąsowski A (2012a) Moving from specifications to contracts in component-based design. In: de Lara J, Zisman A (eds) FASE, volume 7212 of Lecture Notes in Computer Science. Springer, Berlin, pp 43–58
Bauer SS, Fahrenberg U, Legay A, Thrane C (2012b) General quantitative specification theories with modalities. In: Hirsch EA, Karhumäki J, Lepistö A, Prilutskii M (eds) CSR, volume 7353 of Lecture Notes in Computer Science. Springer, Berlin, pp 18–30
Bauer SS, Juhl L, Larsen KG, Legay A, Srba J (2012c) Extending modal transition systems with structured labels. Math Struct Comput Sci 22(4):581–617
Bauer SS, Fahrenberg U, Juhl L, Larsen KG, Legay A, Thrane C (2013) Weighted modal transition systems. Form Methods Syst Des 42(2):193–220
Ben-David S, Chechik M, Uchitel S (2013) Merging partial behaviour models with different vocabularies. In: D’Argenio PR, Melgratti HC (eds) CONCUR, volume 8052 of Lecture Notes in Computer Science. Springer, Berlin, pp 91–105
Beneš N, Křetínský J, Larsen KG, Srba J (2009) On determinism in modal transition systems. Theor Comput Sci 410(41):4026–4043
Beneš N, Černá I, Křetínský J (2011) Modal transition systems: composition and LTL model checking. In: Bultan T, Hsiung P-A (eds) ATVA, volume 6996 of Lecture Notes in Computer Science. Springer, Berlin, pp 228–242
Beneš N, Delahaye B, Fahrenberg U, Křetínský J, Legay A (2013) Hennessy-Milner logic with greatest fixed points. In: D’Argenio PR, Melgratti HC (eds) CONCUR, volume 8052 of Lecture Notes in Computer Science. Springer, Berlin, pp 76–90
Bertrand N, Legay A, Pinchinat S, Raclet J-B (2012) Modal event-clock specifications for timed component-based design. Sci Comput Program 77(12):1212–1234
Boudol G, Larsen KG (1992) Graphical versus logical specifications. Theor Comput Sci 106(1):3–20
Caillaud B, Delahaye B, Larsen KG, Legay A, Pedersen ML, Wąsowski A (2011) Constraint markov chains. Theor Comput Sci 412(34):4373–4404
Caires L, Cardelli L (2003) A spatial logic for concurrency (part I). Inf Comput 186(2):194–235
Cardelli L, Larsen KG, Mardare R (2011) Modular markovian logic. In: Aceto L, Henzinger M, Sgall J (eds) ICALP(2), volume 6756 of Lecture Notes in Computer Science. Springer, Berlin, pp 380–391
Černý P, Henzinger TA, Radhakrishna A (2012) Simulation distances. Theor Comput Sci 413(1):21–35
Continuity, modulus of. Encyclopedia of Mathematics. http://www.encyclopediaofmath.org/index.php?title=Continuity,_modulus_of&oldid=30705
David A, Larsen KG, Legay A, Nyman U, Traonouez L-M, Wąsowski A (2015) Real-time specifications. Int J Softw Tools Technol Transf 17(1):17–45
de Alfaro L, Henzinger TA, Stoelinga M (2002) Timed interfaces. In: Sangiovanni-Vincentelli AL, Sifakis J (eds) EMSOFT, volume 2491 of Lecture Notes in Computer Science, vol 2491. Springer, Berlin, pp 108–122
de Alfaro L (2003) Quantitative verification and control via the mu-calculus. In: Amadio RM, Lugiez D (eds) CONCUR, volume 2761 of Lecture Notes in Computer Science. Springer, Berlin, pp 102–126
de Alfaro L, Faella M, Henzinger TA, Majumdar R, Stoelinga M (2005) Model checking discounted temporal properties. Theor Comput Sci 345(1):139–170
de Alfaro L, Faella M, Stoelinga M (2009) Linear and branching system metrics. IEEE Trans Softw Eng 35(2):258–273
de Alfaro L, Henzinger TA (2001) Interface automata. In: ESEC/SIGSOFT FSE. ACM, pp 109–120
Delahaye B, Larsen KG, Legay A, Pedersen ML, Wąsowski A (2012) Consistency and refinement for interval Markov chains. J Log Algebr Program 81(3):209–226
Delahaye B, Fahrenberg U, Larsen KG, Legay A (2014) Refinement and difference for probabilistic automata. Log Methods Comput Sci 10(3). doi:10.2168/LMCS-10(3:11)2014
Desharnais J, Gupta V, Jagadeesan R, Panangaden P (2004) Metrics for labelled Markov processes. Theor Comput Sci 318(3):323–354
Fahrenberg U, Acher M, Legay A, Wąsowski A (2014a) Sound merging and differencing for class diagrams. In: Gnesi S, Rensink A (eds) FASE, volume 8411 of Lecture Notes in Computer Science. Springer, Berlin, pp 63–78
Fahrenberg U, Legay A, Traonouez L-M (2014b) Structural refinement for the modal nu-calculus. In: Ciobanu G, Méry D (eds) ICTAC, volume 8687 of Lecture Notes in Computer Science. Springer, Berlin, pp 169–187
Fahrenberg U, Křetínský J, Legay A, Traonouez L-M (2014c) Compositionality for quantitative specifications. In: Lanese I, Madelaine E (eds) FACS, volume 8997 of Lecture Notes in Computer Science. Springer, Berlin, pp 306–324
Fahrenberg U, Legay A (2012) A robust specification theory for modal event-clock automata. In: Bauer SS, Raclet J-B (eds) FIT, volume 87 of EPTCS. pp 5–16
Fahrenberg U, Legay A (2013) Generalized quantitative analysis of metric transition systems. In: Shan C-C (ed) APLAS, voume 8301 of Lecture Notes in Computer Science. Springer, Berlin, pp 192–208
Fahrenberg U, Legay A (2014a) General quantitative specification theories with modal transition systems. Acta Inform 51(5):261–295
Fahrenberg U, Legay A (2014b) The quantitative linear-time-branching-time spectrum. Theor Comput Sci 538:54–69
Fahrenberg U, Legay A, Thrane C (2011) The quantitative linear-time–branching-time spectrum. In: Chakraborty S, Kumar A (eds) FSTTCS, volume 13 of LIPIcs. pp 103–114
Feuillade G, Pinchinat S (2007) Modal specifications for the control theory of discrete event systems. Discrete Event Dyn Syst 17(2):211–232
Girard J-Y (1987) Linear logic. Theor Comput Sci 50:1–102
Hennessy M (1985) Acceptance trees. J ACM 32(4):896–928
Henzinger TA, Majumdar R, Prabhu VS (2005) Quantifying similarities between timed systems. In: Pettersson P, Yi W (eds) FORMATS, volume 3829 of Lecture Notes in Computer Science. Springer, Berlin, pp 226–241
Henzinger TA, Sifakis J (2006) The embedded systems design challenge. In: Misra J, Nipkow T, Sekerinski E (eds) FM, volume 4085 of Lecture Notes in Computer Science. Springer, Berlin, pp 1–15
Huth M, Kwiatkowska MZ (1997) Quantitative analysis and model checking. In: LICS. IEEE Computer Society, pp 111–122
Jacobs B, Poll E (2001) A logic for the Java modeling language JML. In: Hußmann H (ed) FASE, volume 2029 of Lecturte Notes in Computer Science. Springer, Berlin, pp 284–299
Jonsson B, Larsen KG (1991) Specification and refinement of probabilistic processes. In: LICS. IEEE Computer Society, pp 266–277
Klin B, Sassone V (2013) Structural operational semantics for stochastic and weighted transition systems. Inf Comput 227:58–83
Kozen D (1983) Results on the propositional \(\mu \)-calculus. Theor Comput Sci 27:333–354
Křetínský J, Sickert S (2013) MoTraS: a tool for modal transition systems and their extensions. In: Van Hung D, Ogawa M (eds) ATVA, volume 8172 of Lecture Notes in Computer Science. Springer, Berlin, pp 487–491
Larsen KG, Thomsen B (1998) A modal process logic. In: LICS. IEEE Computer Society, pp 203–210
Larsen KG, Xinxin L (1990) Equation solving using modal transition systems. In: LICS. IEEE Computer Society, pp 108–117
Larsen KG (1990) Proof systems for satisfiability in Hennessy–Milner logic with recursion. Theor Comput Sci 72(2&3):265–288
Larsen KG, Legay A, Traonouez L-M, Wąsowski A (2011) Robust specification of real time components. In: Fahrenberg U, Tripakis S (eds) FORMATS, volume 6919 of Lecture Notes in Computer Science. Springer, Berlin, pp 129–144
Larsen KG, Mardare R, Panangaden P (2012) Taking it to the limit: approximate reasoning for Markov processes. In: Rovan B, Sassone V, Widmayer P (eds) MFCS, volume 7464 of Lecture Notes in Computer Science. Springer, Berlin, pp 681–692
Larsen KG, Legay A, Traonouez L-M, Wąsowski A (2014) Robust synthesis for real-time systems. Theor Comput Sci 515:96–122
Liskov B, Wing JM (1994) A behavioral notion of subtyping. ACM Trans Program Lang Syst 16(6):1811–1841
Mio M (2011) Probabilistic modal mu-calculus with independent product. In: Hofmann M (ed) FOSSACS, volume 6604 of Lecture Notes in Computer Science. Springer, Berlin, pp 290–304
Morgan C, McIver A (1997) A probabilistic temporal calculus based on expectations. In: Groves L, Reeves S (eds) Formal methods. Springer, Singapore
Raclet J-B (2007) Residual for component specifications. In: Publication interne 1843. IRISA, Rennes
Romero-Hernández D, de Frutos-Escrig D (2012a) Defining distances for all process semantics. In: Giese H, Rosu G (eds) FMOODS/FORTE, volume 7273 of Lecture Notes in Computer Science. Springer, Berlin, pp 169–185
Romero-Hernández D, de Frutos-Escrig D (2012b) Distances between processes: a pure algebraic approach. In: Martí-Oliet N, Palomino M (eds) WADT, volume 7841 of Lecture Notes in Computer Science. Springer, Berlin, pp 265–282
Sifakis J (2011) A vision for computer science-the system perspective. Cent Eur J Comput Sci 1(1):108–116
Traonouez L-M (2012) A parametric counterexample refinement approach for robust timed specifications. In Bauer SS, Raclet J-B (eds) FIT, volume 87 of EPTCS. pp 17–33
Uchitel S, Chechik M (2004) Merging partial behavioural models. In: Taylor RN, Dwyer MB (eds) SIGSOFT FSE. ACM, New York, pp 43–52
van Breugel F, Worrell J (2005) A behavioural pseudometric for probabilistic transition systems. Theor Comput Sci 331(1):115–142
Acknowledgements
The J. Křetínský acknowledges support by the European Research Council (ERC) under Grant 267989 (QUAREM), by the Austrian Science Fund (FWF) under Grants S11402-N23 (RiSE) and Z211-N23 (Wittgenstein Award), the Czech Science Foundation Grant No. P202/12/G061, and the People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme (FP7/2007-2013) REA Grant No. 291734. The U. Fahrenberg, A. Legay, L.-M. Traonouez acknowledge support by ANR MALTHY, Grant No. ANR-13-INSE-0003 from the French National Research Foundation, and by the EU FP7 SENSATION Project, Grant No. 318490 (FP7-ICT-2011-8).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Communicated by M. Droste, Z. Esik and K. Larsen.
This paper is based on the conference contribution (Fahrenberg et al. 2014c), which has been presented at the 11th International Symposium on Formal Aspects of Component Software in Bertinoro, Italy.
Appendix: Proof of Theorem 7
Appendix: Proof of Theorem 7
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \textit{da}( \mathcal D_1), \textit{da}( \mathcal D_2))\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2):}\)
Let and be DMTS. There exists a DMTS refinement family \(R=\{ R_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\}\) such that for all \(s^0_1\in S^0_1\), there is \(s^0_2\in S^0_2\) with \(( s^0_1, s^0_2)\in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2)}\). We show that R is an NAA refinement family.
Let \(\alpha \in \mathbb {L}\) and \(( s_1, s_2)\in R_\alpha \). Let \(M_1\in \text {Tran} _1( s_1)\) and define
The condition
is satisfied by construction. For the inverse condition, let \(( a_1, t_1)\in M_1\), then , and as R is a DMTS refinement family, this implies that there is and \(\beta \in \mathbb {L}\) for which \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), so that \(( a_2, t_2)\in M_2\) by construction.
We are left with showing that \(M_2\in \text {Tran} _2( s_2)\). First we notice that by construction, indeed for all \(( a_2, t_2)\in M_2\). Now let ; we need to show that \(N_2\cap M_2\ne \emptyset \).
We have such that \(\forall ( a_1, t_1)\in N_1: \exists ( a_2, t_2)\in N_2, \beta \in \mathbb {L}:( t_1, t_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). We know that \(N_1\cap M_1\ne \emptyset \), so let \(( a_1, t_1)\in N_1\cap M_1\). Then there is \(( a_2, t_2)\in N_2\) and \(\beta \in \mathbb {L}\) such that \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). But \(( a_2, t_2)\in N_2\) implies , hence \(( a_2, t_2)\in M_2\).
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2)\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \textit{da}( \mathcal D_1), \textit{da}( \mathcal D_2)):}\)
Let and be DMTS. There exists an NAA refinement family \(R=\{ R_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\}\) such that for all \(s^0_1\in S^0_1\), there is \(s^0_2\in S^0_2\) for which \(\left( s^0_1, s^0_2\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \textit{da}( \mathcal D_1), \textit{da}( \mathcal D_2))}\). We show that R is a DMTS refinement family. Let \(\alpha \in \mathbb {L}\) and \(( s_1, s_2)\in R_\alpha \).
Let , then we cannot have . Let , then \(M_1\in \text {Tran} _1( s_1)\) by construction. This implies that there is \(M_2\in \text {Tran} _2( s_2)\), \(( a_2, t_2)\in M_2\) and \(\beta \in \mathbb {L}\) such that \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also as was to be shown.
Let and assume, for the sake of contradiction, that there is no for which \(\forall ( a_1, t_1)\in N_1: \exists ( a_2, t_2)\in N_2, \beta \in \mathbb {L}:( t_1, t_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \) holds. Then for each , there is an element \(( a_{ N_1}, t_{ N_1})\in N_1\) such that \(\exists ( a_2, t_2)\in N_2, \beta \in \mathbb {L}:( t_{ N_1}, t_2)\in R_\beta , F( a_{ N_1}, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \) does not hold.
Let , then \(M_1\in \text {Tran} _1( s_1)\) by construction. Hence, we have \(M_2\in \text {Tran} _2( s_2)\) such that \(\forall ( a_2, t_2)\in M_2: \exists ( a_1, t_2)\in M_1, \beta \in \mathbb {L}:( t_1, t_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq \alpha \). Now \(N_2\cap M_2\ne \emptyset \), so let \(( a_2, t_2)\in N_2\cap M_2\), then there is \(( a_1, t_1)\in M_1\) and \(\beta \in \mathbb {L}\) such that \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), in contradiction to how \(M_1\) was constructed.
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2):}\)
Let \(\mathcal A_1=\left( S_1, S^0_1, \text {Tran} _1\right) \), \(\mathcal A_2=\left( S_2, S^0_2, \text {Tran} _2\right) \) be NAA, with DMTS translations , . There is an NAA refinement family \(R=\{ R_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\}\) such that for all \(s^0_1\in S^0_1\), there is \(s^0_2\in S^0_2\) with \(\left( s^0_1,s^0_2\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)}\).
Define a relation family \(R'=\left\{ R'_\alpha \subseteq D_1\times D_2\mid \alpha \in \mathbb {L}\right\} \) by
We show that \(R'\) is a witness for \(\smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)\). Let \(\alpha \in \mathbb {L}\) and \(( M_1, M_2)\in R'_\alpha \).
Let . By construction of , there is \(( a_2, t_2)\in M_2\) such that \(N_2=\left\{ \left( a_2, M_2'\right) \mid M_2'\in \text {Tran} _2( t_2)\right\} \). Then \(( M_1, M_2)\in R'_\alpha \) implies that there must be \(( a_1, t_1)\in M_1\) and \(\beta \in \mathbb {L}\) such that \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). Let \(N_1=\left\{ ( a_1, M_1')\mid M_1'\in \text {Tran} _1( t_1)\right\} \), then .
We show that \(\forall \left( a_1,M_1'\right) \in N_1: \exists \left( a_2,M_2'\right) \in N_2:\left( M_1', M_2'\right) \in R'_\beta \): Let \(\left( a_1, M_1'\right) \in N_1\), then \(M_1'\in \text {Tran} _1( t_1)\). From \(( t_1, t_2)\in R_\beta \) we get \(M_2'\in \text {Tran} _2( t_2)\) such that
hence \(\left( M_1', M_2'\right) \in R'_\beta \); also, \(\left( a_2, M_2'\right) \in N_2\) by construction of \(N_2\).
Let , then we have for which \(\left( a_1, M_1'\right) \in N_1\) by construction of . This in turn implies that there must be \(( a_1, t_1)\in M_1\) such that \(N_1=\left\{ \left( a_1, M_1''\right) \mid M_1''\in \text {Tran} _1( t_1)\right\} \). By \(( M_1, M_2)\in R'_\alpha \), we get \(( a_2, t_2)\in M_2\) and \(\beta \in \mathbb {L}\) such that \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). Let \(N_2=\{( a_2, M_2')\mid M_2'\in \text {Tran} _2( t_2)\}\), then and hence for all \(( a_2, M_2')\in N_2\). By the same arguments as above, there is \(( a_2, M_2')\in N_2\) for which \(( M_1', M_2')\in R'_\beta \).
We miss to show that \(R'\) is initialized. Let \(M_1^0\in D_1^0\), then we have \(s_1^0\in S_1^0\) with \(M_1^0\in \text {Tran} _1( s_1^0)\). As R is initialized, this entails that there is \(s_2^0\in S_2^0\) with \(( s_1^0, s_2^0)\in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)}\), which gives us \(M_2^0\in \text {Tran} _2( s_2^0)\) which satisfies the conditions in the definition of \(R'_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)}\), whence \(( M_1^0, M_2^0)\in R'_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)}\).
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2)):}\)
Let \(\mathcal A_1=\left( S_1, S^0_1, \text {Tran} _1\right) \), \(\mathcal A_2=\left( S_2, S^0_2, \text {Tran} _2\right) \) be NAA, with DMTS translations , . There is a DMTS refinement family \(R=\{ R_\alpha \subseteq D_1\times D_2\mid \alpha \in \mathbb {L}\}\) such that for all \(M_1^0\in D_1^0\), there exists \(M_2^0\in D_2^0\) with \(\left( M_1^0, M_2^0\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))}\).
Define a relation family \(R'=\{ R'_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\}\) by
we will show that \(R'\) is a witness for \(\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal A_1, \mathcal A_2)\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))\).
Let \(\alpha \in \mathbb {L}\), \(( s_1, s_2)\in R'_\alpha \) and \(M_1\in \text {Tran} _1( s_1)\), then by construction of \(R'\), we have \(M_2\in \text {Tran} _2( s_2)\) with \(( M_1, M_2)\in R_\alpha \).
Let \(( a_2, t_2)\in M_2\) and define \(N_2=\{( a_2, M_2')\mid M_2'\in \text {Tran} _2( t_2)\}\), then . Now \(( M_1, M_2)\in R_\alpha \) implies that there must be satisfying \(\forall ( a_1, M_1')\in N_1: \exists ( a_2, M_2')\in N_2, \beta \in \mathbb {L}:( M_1', M_2')\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). We have \(( a_1, t_1)\in M_1\) such that \(N_1=\{( a_1, M_1')\mid M_1'\in \text {Tran} _1( t_1)\}\); we only miss to show that \(( t_1, t_2)\in R'_\beta \) for some \(\beta \in \mathbb {L}\) for which \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). Let \(M_1'\in \text {Tran} _1( t_1)\), then \(( a_1, M_1')\in N_1\); hence, there is \(( a_2, M_2')\in N_2\) and \(\beta \in \mathbb {L}\) such that \(( M_1', M_2')\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq \alpha \), but \(( a_2, M_2')\in N_2\) also entails \(M_2'\in \text {Tran} _2( t_2)\).
Let \(( a_1, t_1)\in M_1\) and define \(N_1=\{( a_1, M_1')\mid M_1'\in \text {Tran} _1( t_1)\}\), then . Now let \(( a_1, M_1')\in N_1\), then ; hence, we have and \(\beta \in \mathbb {L}\) such that \(( M_1', M_2')\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \). By construction of , this implies that there is with \(( a_2, M_2')\in N_2\), and we have \(( a_2, t_2)\in M_2\) for which \(N_2=\{( a_2, M_2'')\mid M_2''\in \text {Tran} _2( t_2)\}\). Now if \(M_1''\in \text {Tran} _1( t_1)\), then \(( a_1, M_1'')\in N_1\); hence, there is \(( a_2, M_2'')\in N_2\) with \(( M_1'', M_2'')\in R_\beta \), but \(( a, M_2'')\in N_2\) also gives \(M_2''\in \text {Tran} _2( t_2)\).
We miss to show that \(R'\) is initialized. Let \(s^0_1\in S^0_1\) and \(M^0_1\in \text {Tran} _1( s^0_1)\). As R is initialized, this gets us \(M^0_2\in D_2\) with \(( M^0_1, M^0_2)\in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))}\), but \(M^0_2\in \text {Tran} _2( s^0_2)\) for some \(s^0_2\in S^0_2\), and then \(( s^0_1, s^0_2)\in R'_{ \smash {d_\textsf {m} ^\mathbb {L}}( \textit{ad}( \mathcal A_1), \textit{ad}( \mathcal A_2))}\).
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \textit{dn}( \mathcal D_1), \textit{dn}( \mathcal D_2))\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2):}\)
Let and be DMTS, with \(\nu \)-calculus translations \(\textit{dn}( \mathcal D_1)=\left( S_1, S_1^0, \varDelta _1\right) \) and \(\textit{dn}( \mathcal D_2)=( S_2, S_2^0, \varDelta _2)\). There is a DMTS refinement family \(R=\{ R_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\}\) such that for all \(s_1^0\in S_1^0\), there exists \(s_2^0\in S_2^0\) for which \(( s_1^0, s_2^0)\in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2)}\).
Let \(\alpha \in \mathbb {L}\), \(( s_1, s_2)\in R_\alpha \), \(a_1\in \varSigma \), and \(t_1\in \Box ^{ a_1}_1( s_1)\). Then ; hence, we have and \(\beta \in \mathbb {L}\) with \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also \(t_2\in \Box ^{ a_2}_2( s_2)\).
Let \(N_2\in \Diamond _2( s_2)\), then also , so that there must be such that \(\forall ( a_1, t_1)\in N_1: \exists ( a_2, t_2)\in N_2, \beta \in \mathbb {L}:( t_1, t_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also \(N_1\in \Diamond _1( s_1)\).
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2)\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \textit{dn}( \mathcal D_1), \textit{dn}( \mathcal D_2)):}\)
Let and be DMTS, with \(\nu \)-calculus translations
\(\textit{dn}( \mathcal D_1)=\left( S_1, S_1^0, \varDelta _1\right) \) and \(\textit{dn}( \mathcal D_2)=\left( S_2, S_2^0, \varDelta _2\right) \). There is a \(\nu \)-calculus refinement family \(R=\left\{ R_\alpha \subseteq S_1\times S_2\mid \alpha \in \mathbb {L}\right\} \) such that for all \(s_1^0\in S_1^0\), there exists \(s_2^0\in S_2^0\) for which \(\left( s_1^0, s_2^0\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal D_1, \mathcal D_2)}\).
Let \(\alpha \in \mathbb {L}\) and \(( s_1, s_2)\in R_\alpha \) and assume that . Then \(t_1\in \Box ^{ a_1}_1( s_1)\), so that there is \(a_2\in \varSigma \), \(t_2\in \Box ^{ a_2}_2( s_2)\) and \(\beta \in \mathbb {L}\) for which \(( t_1, t_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also .
Assume that , then \(N_2\in \Diamond _2( s_2)\). Hence, there is \(N_1\in \Diamond _1( s_1)\) so that \(\forall ( a_1, t_1)\in N_1: \exists ( a_2, t_2)\in N_2, \beta \in \mathbb {L}:( t_1, t_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also .
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \textit{nd}( \mathcal N_1), \textit{nd}( \mathcal N_2))\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal N_1, \mathcal N_2):}\)
Let \(\mathcal N_1=\left( X_1, X_1^0, \varDelta _1\right) \), \(\mathcal N_2=\left( X_2, X_2^0, \varDelta _2\right) \) be \(\nu \)-calculus expressions in normal form, with DMTS translations and . There is a \(\nu \)-calculus refinement family \(R=\{ R_\alpha \subseteq X_1\times X_2\mid \alpha \in \mathbb {L}\}\) such that for all \(x_1^0\in X_1^0\), there is \(x_2^0\in X_2^0\) for which \(\left( x_1^0, x_2^0\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal N_1, \mathcal N_2)}\).
Let \(\alpha \in \mathbb {L}\) and \(( x_1, x_2)\in R_\alpha \) and assume that . Then \(y_1\in \Box _1^{ a_1}( x_1)\); hence, there are \(a_2\in \varSigma \), \(y_2\in \Box _2^{ a_2}\) and \(\beta \in \mathbb {L}\) such that \(( y_1, y_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also .
Assume that , then \(N_2\in \Diamond _2( x_2)\). Hence, there must be \(N_1\in \Diamond _1( x_1)\) such that \(\forall ( a_1, y_1)\in N_1: \exists ( a_2, y_2)\in N_2, \beta \in \mathbb {L}:( y_1, y_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also .
\(\underline{\smash {d_\textsf {m} ^\mathbb {L}}( \mathcal N_1, \mathcal N_2)\sqsubseteq _\mathbb {L}\smash {d_\textsf {m} ^\mathbb {L}}( \textit{nd}( \mathcal N_1), \textit{nd}( \mathcal N_2)):}\)
Let \(\mathcal N_1=\left( X_1, X_1^0, \varDelta _1\right) \), \(\mathcal N_2=\left( X_2, X_2^0, \varDelta _2\right) \) be \(\nu \)-calculus expressions in normal form, with DMTS translations and . There is a DMTS refinement family \(R=\{ R_\alpha \subseteq X_1\times X_2\mid \alpha \in \mathbb {L}\}\) such that for all \(x_1^0\in X_1^0\), there is \(x_2^0\in X_2^0\) for which \(\left( x_1^0, x_2^0\right) \in R_{ \smash {d_\textsf {m} ^\mathbb {L}}( \mathcal N_1, \mathcal N_2)}\).
Let \(\alpha \in \mathbb {L}\), \(( x_1, x_2)\in R_\alpha \), \(a_1\in \varSigma \), and \(y_1\in \Box ^{ a_1}_1( x_1)\). Then ; hence, we have and \(\beta \in \mathbb {L}\) so that \(( y_1, y_2)\in R_\beta \) and \(F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also \(y_1\in \Box ^{ a_2}_2( x_2)\).
Let \(N_2\in \Diamond _2( x_2)\), then also . Hence, we must have with \(\forall ( a_1, y_1)\in N_1: \exists ( a_2, y_2)\in N_2, \beta \in \mathbb {L}:( y_1, y_2)\in R_\beta , F( a_1, a_2, \beta )\sqsubseteq _\mathbb {L}\alpha \), but then also \(N_1\in \Diamond _1( x_1)\). \(\square \)
Rights and permissions
About this article
Cite this article
Fahrenberg, U., Křetínský, J., Legay, A. et al. Compositionality for quantitative specifications. Soft Comput 22, 1139–1158 (2018). https://doi.org/10.1007/s00500-017-2519-5
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00500-017-2519-5