Skip to main content
Log in

Satisfiability of ECTL with Local Tree Constraints

  • Published:
Theory of Computing Systems Aims and scope Submit manuscript

Abstract

Recently, we have shown that satisfiability for the temporal logic E C T L with local constraints over (ℤ, <, =) is decidable using a new technique (Carapelle et al., 2013). This approach reduces the satisfiability problem of E C T L with constraints over some structure \(\mathcal {A}\) (or class of structures) to the problem whether \(\mathcal {A}\) has a certain model theoretic property that we called EHD (for “existence of homomorphisms is definable”). Here we apply this approach to structures that are tree-like and obtain several results. We show that satisfiability of E C T L with constraints is decidable over (i) semi-linear orders (i.e., tree-like structures where branches form arbitrary linear orders), (ii) ordinal trees (semi-linear orders where the branches form ordinals), and (iii) infinitely branching trees of height h for each fixed \(h\in \mathbb {N}\). We prove that all these classes of structures have the property EHD. In contrast, we introduce Ehrenfeucht-Fraïssé-games for W M S O + B (weak M S O with the bounding quantifier) and use them to show that the infinite (order) tree does not have the EHD-property. As a consequence, our technique cannot be used to establish whether satisfiability of E C T L with constraints over the infinite (order) tree is decidable. A preliminary version of this paper has appeared as (Carapelle et al., 2015).

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. A structure \(\mathcal {U}\) is universal for a class Γ if there is a homomorphic embedding of every structure from Γ into \(\mathcal {U}\) and \(\mathcal {U}\) belongs to Γ.

  2. We call (A, <, ⊥ ) a graph to emphasize that here the binary relation symbols < and ⊥ can have arbitrary interpretations and they need not be a partial order and its incomparability relation. We can instead see them as two different kinds of edges in an arbitrary graph.

  3. For the ease of presentation we assume that \(\mathcal {A}\) and \(\mathcal {B}\) are infinite structures.

References

  1. Bojaṅczyk, M., Parys, P., Toruṅczyk, S.: The MSO+U theory of (N, < ) is undecidable. In: Proceedings of the 33rd Symposium on Theoretical Aspects of Computer Science (STACS 2016), Leibniz International Proceedings in Informatics (LIPIcs), vol. 47, pp. 21:1–21:8. Schloss Dagstuhl–Leibniz-Zentrum für Informatik (2016)

  2. Bojańczyk, M., Toruńczyk, S.: Weak MSO+U over infinite trees (long version). http://www.mimuw.edu.pl/bojan/papers/wmsou-trees

  3. Bojańczyk, M., Toruńczyk, S.: Weak MSO+U over infinite trees. In: Proceedings of the 29th International Symposium on Theoretical Aspects of Computer Science (STACS 2012), Leibniz International Proceedings in Informatics (LIPIcs), vol. 14, pp. 648–660. Schloss Dagstuhl–Leibniz-Zentrum für Informatik (2012)

  4. Bozzelli, L., Gascon, R.: Branching-time temporal logic extended with qualitative presburger constraints. In: Proceedings of the 13th International Conference on Logic for Programming, Artificial Intelligence, and Reasoning (LPAR 2006), Lecture Notes in Computer Science, vol. 4246, pp. 197–211. Springer (2006)

  5. Bozzelli, L., Pinchinat, S.: Verification of gap-order constraint abstractions of counter systems. Theor. Comput. Sci. 523, 1–36 (2014)

    Article  MathSciNet  MATH  Google Scholar 

  6. Carapelle, C., Feng, S., Kartzow, A., Lohrey, M.: Satisfiability of ECTL* with Tree Constraints. In: Proceedings of the 10th International Computer Science Symposium in Russia (CSR 2015), Lecture Notes in Computer Science, vol. 9139, pp. 94–108. Springer (2015)

  7. Carapelle, C., Kartzow, A., Lohrey, M.: Satisfiability of CTL* with Constraints. In: Proceedings of the 24th International Conference on Concurrency Theory (CONCUR 2013), Lecture Notes in Computer Science, vol. 8052, pp. 455–469. Springer (2013)

  8. Carapelle, C., Kartzow, A., Lohrey, M.: Satisfiability of ECTL with constraints. J. Comput. Syst. Sci. 82 (5), 826–855 (2016). http://www.sciencedirect.com/science/article/pii/S002200001600012X

    Article  MathSciNet  MATH  Google Scholar 

  9. Demri, S., Deters, M.: Temporal logics on strings with prefix relation. J. Log. Comput. 26(3), 989–1017 (2016). doi:10.1093/logcom/exv028 http://logcom.oxfordjournals.org/content/26/3/989

    Article  MathSciNet  MATH  Google Scholar 

  10. Demri, S., D’Souza, D.: An automata-theoretic approach to constraint LTL. Inf. Comput. 205(3), 380–415 (2007)

    Article  MathSciNet  MATH  Google Scholar 

  11. Demri, S., Gascon, R.: Verification of qualitative Z constraints. Theor. Comput. Sci. 409(1), 24–40 (2008)

    Article  MathSciNet  MATH  Google Scholar 

  12. Demri, S., Lazić, R.: LTL with the freeze quantifier and register automata. ACM Trans. Comput. Log. 10(3), 16:1–16:30 (2009)

    Article  MathSciNet  MATH  Google Scholar 

  13. Droste, M.: Partially ordered sets with transitive automorphism groups. Proc. Lond. Math. Soc. 54(3), 517–543 (1987)

    Article  MathSciNet  MATH  Google Scholar 

  14. Ebbinghaus, H., Flum, J.: Finite Model Theory. Springer Monographs in Mathematics. Springer Berlin Heidelberg (2005)

  15. Gascon, R.: An automata-based approach for CTL* with constraints. Electron. Notes Theor. Comput. Sci. 239, 193–211 (2009)

    Article  MathSciNet  MATH  Google Scholar 

  16. Gottschalk, W.H.: Choice functions and Tychonoff’s theorem. Proc. Am. Math. Soc. 2, 172 (1951)

    MathSciNet  MATH  Google Scholar 

  17. Rado, R.: Axiomatic treatment of rank in infinite sets. Can. J. Math. 1, 337–343 (1949)

    Article  MathSciNet  MATH  Google Scholar 

  18. Thomas, W.: Computation tree logic and regular omega-languages. In: REX Workshop, pp. 690–713 (1988)

  19. Thomas, W.: Languages, automata, and logic. In: Handbook of Formal Languages, pp. 389–455. Springer (1996)

  20. Vardi, M.Y., Wolper, P.: Yet another process logic (preliminary version). In: Logic of Programs, pp. 501–512 (1983)

  21. Wolk, E.S.: The comparability graph of a tree. Proc. Am. Math. Soc. 13(5), 789–795 (1962)

    Article  MathSciNet  MATH  Google Scholar 

  22. Wolk, E.S.: A note on The comparability graph of a tree. Proc. Am. Math. Soc. 16(1), 17–20 (1965)

    MathSciNet  MATH  Google Scholar 

Download references

Acknowledgements

We thank Manfred Droste for fruitful discussions on universal structures and semi-linear orders.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Claudia Carapelle.

Additional information

This work is supported by the DFG Research Training Group 1763 (QuantLA) and the DFG research project GELO.

Appendix: A universal semi-linear order

Appendix: A universal semi-linear order

In the following we define a semi-linear order which is universal for the class of all countable semi-linear orders. The fact that this order is universal is known to the experts in the field of semi-linear orders. Unfortunately, to our best knowledge there is no proof of this fact in the literature. Hence we provide a proof for this result.

Definition 31

Let \(\mathcal {U}=(U, <, \mathrel {\bot })\) be the countable semi-linear order with:

  • \(U = (\mathbb {N}\mathbb {Q})^{*}\) (the set of all finite sequences n 1 q 1n k q k with k ≥ 0, \(n_{1}, \ldots , n_{k} \in \mathbb {N}\) and \(q_{1}, \ldots , q_{k} \in \mathbb {Q}\)),

  • < is the strict order induced by n 1 p 1 n 2 p 2n k p k m 1 q 1 m 2 q 2m l q l iff

    • kl, n i = m i for all 1 ≤ ik, p i = q i for all 1 ≤ ik − 1 and p k q k , and

    • ⊥ = ⊥<.

We call \(\mathcal {U}\) the universal countable semi-linear order.

Note that Droste [13] has already studied this and similar orders.

For u = n 1 p 1 n 2 p 2n k p k U (with k ≥ 1) and \(q\in \mathbb {Q}\), we define

$$ u+q = n_{1}p_{1}n_{2}p_{2}{\cdots} n_{k-1}p_{k-1}n_{k}(p_{k}+q) . $$
(8)

We say that a countable semi-linear order (A, <, ⊥ ) is closed under finite infima if for each finite set \(S \subseteq A\) the linear order {aAas for allsS} has a maximal element, which is denoted by \(\inf (S)\). Let \(E=(a_{i})_{i\in \mathbb {N}}\) be a repetition-free enumeration of A. We say E is closed under infima if for each initial subset \(A_{i}=\{a_{1}, a_{2}, \dots , a_{i}\}\) and each \(S\subseteq A_{i}\) we have \(\inf (S) \in A_{i}\).

Lemma 32

Let \(\mathcal {A} = (A, < ,\mathrel {\bot })\) be a countable semi-linear order. There is a countable semi-linear order \(\mathcal {B}\) that is closed under finite infima and an injective homomorphism from \(\mathcal {A}\) to \(\mathcal {B}\).

Proof

For a nonempty subset \(S\subseteq A\) we set ↓S = {aA∣∀sS(as)}. Let \(\bar A\) be the set of finite nonempty subsets of A, which is obviously countable. We define an equivalence on \(\bar A\) by setting ST iff ↓S = ↓T. For all \(S\in \bar A\), [S] denotes its equivalence class. Let B be the set of all equivalence classes. We define an order ⊏ on B by [S] ⊏ [T] if and only if ↓S ⊊ ↓T.

We claim that \(\mathcal {B} = (B, \sqsubset , \mathrel {\bot }_{\sqsubset })\) is a semi-linear order that is closed under finite infima and that the map ϕ given by ϕ(a) ↦ [{a}] is an injective homomorphism from \(\mathcal {A}\) to \(\mathcal {B}\).

  • \(\mathcal {B}\) is obviously a partial order. Moreover, note that ↓S is a linear and downwards closed suborder of \(\mathcal {A}\) for every nonempty finite set \(S \subseteq A\). In order to show that \(\mathcal {B}\) is semi-linear, assume that \([S_{1}] \sqsubset [S]\) and \([S_{2}] \sqsubset [S]\), i.e., \({\downarrow } S_{1} \subsetneq {\downarrow } S\) and \({\downarrow } S_{2} \subsetneq {\downarrow } S\). Thus, all elements from ↓S 1 and all elements from ↓S 2 are comparable. Since both sets are downwards closed, this directly implies that either [S 1]=[S 2], \([S_{1}] \sqsubset [S_{2}]\) or \([S_{2}] \sqsubset [S_{1}]\).

  • Let us show that \(\mathcal {B}\) is closed under finite infima: Let S, S 1, … , S n be finite nonempty subsets of A and assume that \([S] \sqsubseteq [S_{i}]\) for all 1 ≤ in. Thus, \({\downarrow } S \subseteq {\downarrow } S_{i}\). Hence, \({\downarrow } S \subseteq \bigcap _{i=1}^{n} {\downarrow } S_{i} = {\downarrow } \bigcup _{i=1}^{n} S_{i}\). Since \({\downarrow } \bigcup _{i=1}^{n} S_{i} \subseteq {\downarrow } S_{i}\) for all 1 ≤ in, \([\bigcup _{i=1}^{n} S_{i}] = \inf (\{ [S_{1}], \ldots , [S_{n}] \})\).

  • For a, bA with ab we have b ∉ ↓{a} or a ∉ ↓{b}. Thus, ϕ is an injective map from A to B. Moreover, a < b implies \({\downarrow }\{a\} \subsetneq {\downarrow }\{b\}\), i.e., \(\phi (a) \sqsubset \phi (b)\). Similarly, ab implies a ∉ ↓{b} and b ∉ ↓{a}, i.e., \(\phi (a) \mathrel {\bot }_{\sqsubset } \phi (b)\).

Lemma 33

Let \(\mathcal {A} =(A, < , \mathrel {\bot })\) be a countable semi-linear order that is closed under finite infima. There is a repetition-free enumeration of \(\mathcal {A}\) , which is closed under infima.

Proof

Fix an arbitrary repetition-free enumeration \((a_{i})_{i\in \mathbb {N}}\) of A. Assume that we have constructed a sequence \(b_{1}, b_{2}, \dots , b_{i}\) such that \(B_{j} = \{b_{1}, b_{2}, \dots , b_{j}\}\) is closed under infima for every ji. Let k ∈ ℕ be minimal with a k B i . Let \(b^{\prime }_{1} < b^{\prime }_{2} < {\dots } b^{\prime }_{m} < a_{k}\) be the list of all infima of the form \(\inf (S\cup \{a_{k}\})\) for \(S\subseteq B_{i}\) that are not contained in B i . This list is indeed linearly ordered by < since all elements in the list are bounded by a k . Now set \(b_{i+l} = b^{\prime }_{l}\) for all 1 ≤ lm and set b i + m+1 = a k . The resulting sequence b 1, … , b i + m + 1 contains a k and \(B_{j} = \{b_{1}, b_{2}, \dots , b_{j}\}\) is closed under infima for every ji + m + 1. This can be easily shown using the fact that \(\inf (X \cup \{ \inf (Y)\}) = \inf (X \cup Y)\) for all sets X and Y.

Repeating this construction leads to an enumeration \((b_{i})_{i\in \mathbb {N}}\) of A with the desired property. □

Lemma 34

Let \(\mathcal {A}=(A, \sqsubset , \mathrel {\bot }_{\sqsubset })\) be a countable semi-linear order. There exists an injective homomorphism from \(\mathcal {A}\) into \(\mathcal {U}\).

Proof

Due to Lemma 32 and 33, we may assume that \(\mathcal {A}\) is closed under finite infima and that \((a_{i})_{i\in \mathbb {N}}\) is a repetition-free enumeration of A which is closed under finite infima. Set \(A_{i}=\{a_{1}, \dots , a_{i}\}\) for i ≥ 1. Inductively, we construct injective homomorphisms ϕ i :A i U (i ≥ 1) such that

  1. 1.

    ϕ i+1 extends ϕ i , and

  2. 2.

    for all \(u = n_{1}p_{1}n_{2}p_{2}{\dots } n_{k}p_{k} \in {\mathsf {im}}(\phi _{i})\) and all 1 ≤ jk we have \(p_{j}\in \frac {1}{2^{i}}\mathbb {Z}\).

Define ϕ 1:A 1U by \(\phi _{1}(a_{1}) = 00 \in \mathbb {N}\mathbb {Q}\). Assume that ϕ i has already been constructed. We distinguish two cases.

  1. 1.

    If there is some aA i with \(a_{i+1} \sqsubset a\) let \(u = \inf \{a \in A_{i}\mid a_{i+1} \sqsubset a\}\). Note that \(a_{i+1} \sqsubseteq u\). Since the enumeration is closed under infima, we have uA i (and thus \(a_{i+1} \sqsubset u\)) and we can define \(\phi _{i+1}(a_{i+1}) = \phi _{i}(u)+ (\frac {-1}{2^{i+1}})\), where we add according to (8). Note that ϕ i+1(a i+1) < ϕ i (u) = ϕ i+1(u). In order to prove that this defines a homomorphism, we distinguish the following cases:

    1. (a)

      If \(a_{i+1} \sqsubset a\) for some aA i then ua. Hence, ϕ i+1(a i+1) < ϕ i+1(u) = ϕ i (u) ≤ ϕ i (a) = ϕ i+1(a) as desired.

    2. (b)

      If \(a \sqsubset a_{i+1}\) for some aA i , then au. Hence, ϕ i+1(a) = ϕ i (a) < ϕ i (u). Since ϕ i uses only rationals from \(\frac {1}{2^{i}}\mathbb {Z}\), we conclude that \(\phi _{i+1}(a) \leq \phi _{i}(u) + \frac {-1}{2^{i}} < \phi _{i+1}(a_{i+1})\) as desired.

    3. (c)

      If \(a_{i+1} \mathrel {\bot }_{\sqsubset } a\) for some aA i , then \(a \mathrel {\bot }_{\sqsubset } u\). By induction, ϕ i+1(a) = ϕ i (a) ⊥< ϕ i (u) = ϕ i+1(u). Thus, the assumption ϕ i+1(a) ≤ ϕ i+1(a i+1) leads by transitivity of ≤ to the contradiction ϕ i+1(a) ≤ ϕ i+1(u). Similarly, the assumption ϕ i+1(a) > ϕ i+1(a i+1) yields \(\phi _{i+1}(a) \geq \phi _{i+1}(a_{i+1})+\frac {1}{2^{i+1}} = \phi _{i+1}(u)\). We can conclude that ϕ i+1(a) ⊥< ϕ i+1(a i+1) as desired.

  2. 2.

    Otherwise, for all ji we know that \(\inf \{a_{j},a_{i+1}\}\) is strictly below a i+1 and hence belongs to A i (since the enumeration is closed under infima). In particular, the set {aA i a < a i+1} is not empty. By semi-linearity, \(u=\max \{ a\in A_{i}\mid a< a_{i+1}\}\) is well-defined. Since i m(ϕ i ) is finite, there is some \(n\in \mathbb {N}\) such that φ i (u)n0 is incomparable to all elements from the set φ i (A i ∖{aA i au}). Extending ϕ i by setting ϕ i+1(a i+1) = φ i (u)n0 is easily shown to be a homomorphism.

Finally, the limit of \((\phi _{i})_{i\in \mathbb {N}}\) clearly defines an injective homomorphism from \(\mathcal {A}\) into \(\mathcal {U}\). □

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Carapelle, C., Feng, S., Kartzow, A. et al. Satisfiability of ECTL with Local Tree Constraints. Theory Comput Syst 61, 689–720 (2017). https://doi.org/10.1007/s00224-016-9724-y

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00224-016-9724-y

Keywords

Navigation