Skip to main content

Derivational Modal Logics with the Difference Modality

  • Chapter
  • First Online:

Part of the book series: Outstanding Contributions to Logic ((OCTR,volume 4))

Abstract

In this chapter we study modal logics of topological spaces in the combined language with the derivational modality and the difference modality. We give axiomatizations and prove completeness for the following classes: all spaces, \(T_1\)-spaces, dense-in-themselves spaces, a zero-dimensional dense-in-itself separable metric space, \(\mathbf R^n~ (n\ge 2).\) We also discuss the correlation between languages with different combinations of the topological, derivational, universal and difference modalities in terms of definability.

In memory of Leo Esakia

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

Buying options

Chapter
USD   29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD   84.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD   109.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Learn about institutional subscriptions

Notes

  1. 1.

    The early works of the second author in this field were greatly influenced by Leo Esakia.

  2. 2.

    Some other kinds of topomodal logics arise when we deal with topological spaces with additional structures, e.g. spaces with two topologies, spaces with a homeomorphism etc. (cf. [3]).

  3. 3.

    For the 1-modal case this lemma has been known as folklore since the 1970s; the second author learned it from Leo Esakia in 1975.

  4. 4.

    Sometimes we neglect this difference.

  5. 5.

    There is no common notation for this operation; some authors use \(\varvec{\uptau }\).

  6. 6.

    So we extend the definitions of the d-truth or the c-truth by adding the item for \([\forall ]\) or \([\ne ]\).

  7. 7.

    Shehtman [37] contains a stronger claim: \(\mathbf {Lc_\forall }(\mathfrak X)=\mathbf {S4U} + AC\) for any connected dense-in-itself separable metric \(\mathfrak X\). However, recently we found a gap in the proof of Lemma 17 from that paper. Now we state the main result only for the case \({\mathfrak X}=\mathbf R^n\); a proof can be obtained by applying the methods of the present chapter, but we are planning to publish it separately.

  8. 8.

    Recently Kremer [21] has showed that \(\mathbf{S4}\) is strongly complete w.r.t. any dense-in-itself metric space. His proof uses much of the construction from [33].

  9. 9.

    \(f_1\cup f_2\) is the map \(f\) such that \(f|X_i=f_i\); similarly for \(f_1\cup f_2\cup g\) (Fig. 11.2).

  10. 10.

    \(|\ldots |\) denotes the cardinality.

  11. 11.

    Basic frames were defined in Sect. 11.4.

  12. 12.

    Recall that \(R^\prime \) is the transitive closure of \({\underline{R}}\), \(R_D'=\underline{R_D}\).

  13. 13.

    In this chapter, as well as in [9, ‘countable’ means ‘of cardinality at most \(\aleph _0\)’.

  14. 14.

    In [31] neighbourhoods are assumed to be open, but this does not matter here since every neighbourhood contains an open neighbourhood.

References

  1. Abashidze M (1987) Ordinally complete normal extensions of the logic of provability. In: Logic, methodology and philosophy of science, abstracts, vol 5, part 1. Moscow, pp 9–10

    Google Scholar 

  2. Abashidze M (1988) Ordinal completeness of modal Gödel-Löb system (Russian). In: Intensional logics and logical structure of theories (Telavi, 1985), Tbilisi, pp 49–73

    Google Scholar 

  3. Aiello M, Pratt-Hartmann I, van Benthem J (eds) (2007) Handbook of spatial logics. Springer, Berlin

    Google Scholar 

  4. Aleksandrov P (1977) Introduction to set theory and general topology. Nauka, Moscow

    Google Scholar 

  5. Aull C, Thron W (1962) Separation axioms between \(T_0\) and \(T_1\). Indag Math 24:26–37

    Google Scholar 

  6. Bezhanishvili G, Esakia L, Gabelaia D (2005) Some results on modal axiomatization and definability for topological spaces. Studia Logica 81:325–355

    Google Scholar 

  7. Bezhanishvili G, Esakia L, Gabelaia D (2010) The modal logic of Stone spaces: diamond as derivative. Rev Symbolic Logic 3:26–40

    Google Scholar 

  8. Bezhanishvili G, Esakia L, Gabelaia D (2011) Spectral and \(T_0\)-spaces in d-semantics. In: Bezhanishvili N, Löbner S, Schwabe K, Spada L (eds) Lecture notes in artificial intelligence. Springer, Berlin, pp 16–29

    Google Scholar 

  9. Bezhanishvili G, Lucero-Bryan J (2012) More on d-logics of subspaces of the rational numbers. Notre Dame J Formal Logic 53:319–345

    Article  Google Scholar 

  10. Blass A (1990) Infinitary combinatorics and modal logic. J Symbolic Logic 55:761–778

    Article  Google Scholar 

  11. Burgess J (1985) The decision problem for linear temporal logic. Notre Dame J Formal Logic 26:115–128

    Article  Google Scholar 

  12. Chagrov A, Zakharyaschev M (1997) Modal logic. Oxford University Press, Oxford

    Google Scholar 

  13. de Rijke M (1993) Extending modal logic. PhD Thesis. ILLC Dissertation Series, Amsterdam

    Google Scholar 

  14. Esakia L (1981) Diagonal constructions, Löb formula, and scattered Cantor spaces. In: Logical and semantical investigations, Tbilisi, pp 128–143 (Russian)

    Google Scholar 

  15. Esakia L (2001) Weak transitivity—a restitution. In: Logical investigations, vol 8. Nauka, Moscow, pp 244–245 (Russian)

    Google Scholar 

  16. Fine K (1985) Logics containing K4, Part II. J Symbolic Logic 50:619–651

    Article  Google Scholar 

  17. Gabelaia D (2001) Modal definability in topology. Master’s thesis, ILLC

    Google Scholar 

  18. Gargov G, Goranko V (1991) Modal logic with names. In: Colloquium on modal logic. Dutch network for language, logic and information, Amsterdam, pp 81–103

    Google Scholar 

  19. Goranko V, Passy S (1992) Using the universal modality: gains and questions. J Logic Comput 2:5–30

    Article  Google Scholar 

  20. Hodkinson I (2012) On the Priorean temporal logic with ‘around now’ over the real line. Submitted

    Google Scholar 

  21. Kremer P (2013) Strong completeness of S4 for any dense-in-itself metric space. Rev Symbolic Logic 6:545–570

    Article  Google Scholar 

  22. Kudinov A (2005) Difference modality in topological spaces. Algebraic and topological methods in non-classical logics II, Barcelona, abstracts, pp 50–51

    Google Scholar 

  23. Kudinov A (2006) Topological modal logics with difference modality. In: Advances in modal logic, vol 6. College Publications, London, pp 319–332

    Google Scholar 

  24. Kudinov A (2011) Modal logic of R with the difference modality. In: Proceedings of the conference on information technologies and systems. Moscow, pp 335–339 (Russian)

    Google Scholar 

  25. Kudinov A (2008) Topological modal logic of R with inequality. Russ Math Surv 63:163–165

    Article  Google Scholar 

  26. Kudinov A, Shapirovsky I (2010) Some examples of modal logics without a finite axiomatisation. In: Proceedings of the conference on information technologies and systems. Moscow, pp 258–262 (Russian)

    Google Scholar 

  27. Kuratowski K (1922) Sur l’opération \(\bar{A}\) de l’Analysis Situs. Fundam Math 3:181–199

    Google Scholar 

  28. Kuratowski K (1966) Topology, vol 1. Academic Press, London

    Google Scholar 

  29. Litak T (2006) Isomorphism via translation. Advances in modal logic, vol 6. College Publications, London, pp 333–351

    Google Scholar 

  30. Lucero-Bryan J (2011) The d-logic of the rational numbers: a fruitful construction. Studia Logica 97:265–295

    Google Scholar 

  31. Lucero-Bryan J (2013) The d-logic of the real line. J Logic Comput 23:121–156

    Article  Google Scholar 

  32. McKinsey J, Tarski A (1944) The algebra of topology. Ann Math 45:141–191

    Google Scholar 

  33. Rasiowa H, Sikorski R (1963) The mathematics of metamathematics. Monografie matematyczne, Tom 41. PWN—Polish Scientific Publishers, Warsaw

    Google Scholar 

  34. Reynolds M (2010) The complexity of temporal logic over the reals. Ann Pure Appl Logic 161:1063–1096

    Article  Google Scholar 

  35. Segerberg K (1980) A note on the logic of elsewhere. Theoria 46:183–187

    Article  Google Scholar 

  36. Shehtman V (1990) Derived sets in Euclidean spaces and modal logic. ITLI Prepublication Series, X-90-05, University of Amsterdam

    Google Scholar 

  37. Shehtman V (1999) “Everywhere" and “here". J Appl Non-classical Logics 9:369–380

    Google Scholar 

  38. Shehtman V (2000) Modal logics of topological spaces. Habilitation Thesis. Moscow (Russian).

    Google Scholar 

  39. van Benthem J (1983) Modal logic and classical logic. Bibliopolis, Napoli

    Google Scholar 

  40. Zakharyaschev M (1993) A sufficient condition for the finite model property of modal logics above K4. Bull IGPL 1:13–21

    Article  Google Scholar 

Download references

Acknowledgments

We would like to thank the referee who has helped us to improve the first version of the manuscript. The work on this chapter was supported by RFBR grants 11-01-00281-a, 11-01-00958-a, 11-01-93107-CNRS-a and the Russian President’s grant NSh-5593.2012.1.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Andrey Kudinov .

Editor information

Editors and Affiliations

Appendix

Appendix

Finally let us give technical details of the proofs of Propositions 11, 19.

Proposition 11 Let \(\mathfrak X\) be a dense-in-itself separable metric space, \(B\subset X\) a closed nowhere dense set. Then there exists a d-morphism \(g: \mathfrak X\twoheadrightarrow ^d \Phi _{ml}\) with the following properties:

  1. (1)

    \(B \subseteq g^{-1}(b_1)\);

  2. (2)

    every \(g^{-1}(a_i)\) \((for i\le l)\) is a union of a set \(\alpha _i\) of disjoint open balls, which is dense at any point of \(g^{-1}(\{ b_1,\ldots ,b_m\})\).

The frame \(\Phi _{ml}\) is shown in Fig. 11.1.

Proof

Let \(X_1,\dots , X_n,\dots \) be a countable base of \(\mathfrak X\) consisting of open balls. We construct sets \(A_{ik}, ~B_{jk}\) for \(1\le i\le l, ~1\le j\le m, ~k\in \omega \), with the following properties:

  1. (1)

    \(A_{ik}\) is the union of a finite set \(\alpha _{ik}\) of nonempty open balls whose closures are disjoint;

  2. (2)

    \(\mathbf{C}A_{ik} \cap \mathbf{C}A_{i'k} = \varnothing \) for \(i\not = i'\);

  3. (3)

    \(\alpha _{ik} \subseteq \alpha _{i,k+1};~A_{ik} \subseteq A_{i,k+1}\);

  4. (4)

    \(B_{jk}\) is finite;

  5. (5)

    \(B_{jk} \subseteq B_{j,k+1};\)

  6. (6)

    \(A_{ik} \cap B_{jk} = \varnothing ;\)

  7. (7)

    \(X_{k+1} \subseteq \bigcup \limits _{i=1}^l A_{ik}\Rightarrow \alpha _{i,k+1} = \alpha _{ik}, ~B_{j,k+1} = B_{jk};\)

  8. (8)

    if \(X_{k+1}\not \subseteq \bigcup \limits _{i=1}^l A_{ik}\), there are closed nontrivial balls \(P_1,\dots ,P_l\) such that for any \(i\), \(j\)

    $$P_i \subseteq X_{k+1}-A_{ik},~ \alpha _{i,k+1} = \alpha _{ik} \cup \{ \mathbf{I}P_i\},~ (B_{j,k+1} - B_{jk}) \cap X_{k+1}\not = \varnothing ; $$
  9. (9)

    \(A_{ik} \subseteq X-B\);

  10. (10)

    \(B_{jk} \subseteq X-B\);

  11. (11)

    \(j\not = j'\Rightarrow B_{j'k} \cap B_{jk} = \varnothing \) .

We carry out both the construction and the proof by induction on \(k\).

Let \(k=0\); \((X-B)\) is infinite since it is nonempty and open in a dense-in-itself \(\mathfrak X\). Take distinct points \(v_1,\dots ,v_l \not \in B\) and disjoint closed nontrivial balls \( Z_1,\dots ,Z_l \subset X-B\) with centers at \(v_1,\dots ,v_l\), respectively (see Fig.11.6).

Fig. 11.6
figure 6

Case \(k=0\)

Put

$$\alpha _{i0}: =\{ \mathbf{I}Z_i\}; ~A_{i0}: = \mathbf{I}Z_i;$$

then \(Z_i = \mathbf{C}A_{i0}\). As above, since \((X-B) -\bigcup \limits ^l_{i=1}Z_i\) is nonempty and open, it is infinite. Pick distinct \(w_1,\dots ,w_m\in X-B\) and put \(B_{j0}: = \{ w_j\}\). Then the required properties hold for \(k=0\).

At the induction step we construct \(A_{i,k+1}, B_{j,k+1}\). Put \(Y_k :=\bigcup \limits ^l_{i=1}A_{ik}\) and consider two cases.

(a) \(X_{k+1}\subseteq Y_k\). Then put:

$$\alpha _{i,k+1}: = \alpha _{ik};~ A_{i,k+1}: = A_{ik}; ~B_{j,k+1}: = B_{jk}.$$

(b) \(X_{k+1}\not \subseteq Y_k\). Then \(X_{k+1} \not \subseteq \mathbf{C}Y_k\). Indeed, \(X_{k+1} \subseteq \mathbf{C}Y_k\) implies \(X_{k+1} \subseteq \mathbf{I}\mathbf{C}Y_k=Y_k\) since \(X_{k+1}\) is open and by (1) and (2). So we put

$$W_0 := X_{k+1} - \mathbf{C}Y_k -\bigcup ^m_{j=1}B_{jk},~ W := W_0 - B.$$

Since \((X_{k+1} - \mathbf{C}Y_k )\) is nonempty and open and every \(B_{jk}\) is finite by (4), \(W_0\) is also open and nonempty (by the density of \(\mathfrak X\)). By the assumption of Proposition 11, \(B\) is closed, and thus \(W\) is open.

\(W\) is also nonempty. Otherwise \(W_0 \subseteq B\), and then \(W_0 \subseteq \mathbf{I}B=\varnothing \) (since \(B\) is nowhere dense by the assumption of Proposition 11).

Now we argue similarly to the case \(k=0\). Take disjoint closed nontrivial balls \(P_1,\dots ,P_l\subset W\). Then \(W - \bigcup \limits ^l_{i=1}P_i\) is infinite, so we pick distinct \(b_{1,k+1},\dots , b_{m,k+1}\) in this set and put

$$B_{j,k+1}: = B_{jk} \cup \{ b_{j,k+1}\},~ \alpha _{i,k+1}: = \alpha _{ik} \cup \{ \mathbf{I}P_i\},~ A_{i,k+1}: = A_{ik} \cup \mathbf{I}P_i.$$

In case (a) all the required properties hold for \((k+1)\) by the construction.

In case (b) we have to check only (1), (2), (6), (8)–(11).

(8) holds since by construction we have

$$\begin{aligned} P_i\subset W\subset&X_{k+1}-\mathbf{C}Y_k\subset X_{k+1}-A_{ik};\\ b_{j,k+1}\in W \subseteq&X_{k+1},~ b_{j,k+1}\in (B_{j,k+1}-B_{jk}). \end{aligned}$$

(1): From IH it is clear that \(\alpha _{i,k+1}\) is a finite set of open balls and their closures are disjoint; note that \(P_i \cap \mathbf{C}A_{ik} = \varnothing \) since \(P_i \subseteq W \subseteq -\mathbf{C}A_{ik}.\)

(2): We have

$$\begin{aligned} \mathbf{C}A_{i,k+1} \cap \mathbf{C}A_{i^\prime ,k+1}&= (\mathbf{C}A_{ik} \cup P_i) \cap (\mathbf{C}A_{i^\prime k}\cup P_{i^\prime }) \\&=(\mathbf{C}A_{ik} \cap \mathbf{C}A_{i^\prime k}) \cup (\mathbf{C}A_{ik} \cap P_{i'}) \cup (\mathbf{C}A_{i^\prime k} \cap P_i) \cup (P_i \cap P_{i'})\\ {}&=\mathbf{C}A_{ik} \cap \mathbf{C}A _{i^\prime k} = \varnothing \end{aligned}$$

by IH and by the construction; note that \(P_i, P_i' \subseteq W \subseteq -\mathbf{C}Y_k\).

(6): We have

$$ A_{i,k+1} \cap B_{j,k+1} = (A_{ik} \cap B_{jk} ) \cup (\mathbf{I}P_i \cap \left\{ b_{j,k+1}\right\} ) \cup ( A_{ik} \cap \left\{ b_{j,k+1}\right\} ) \cup (\mathbf{I}P_i \cap B_{jk}) = \varnothing $$

by IH and since \(b_{j,k+1}\not \in P_i,~ b_{j,k+1}\in W \subseteq X-Y_k\), \(P_i \subset W \subseteq X-B_{jk}.\)

(9): We have \(A_{i,k+1} = A_{ik} \cup \mathbf{I}P_i \subseteq -B\) since \(A_{ik} \subseteq -B\) by IH, and \(P_i \subset W \subseteq -B\) by the construction.

Likewise, (10) follows from \(B_{jk} \subseteq -B\) and \(b_{j,k+1}\in W \subseteq -B\).

To check (11), assume \(j \not = j'\). We have \(B_{j',k+1} \cap B_{j,k+1} = B_{j'k} \cap B_{jk}\) since \(b_{j^\prime ,k+1} \not = b_{j,k+1} , ~ b_{j,k+1}\in W \subseteq -B_{j^\prime k}\) and \(b_{j^\prime ,k+1}\in W\subseteq -B_{jk}\). Then apply IH.

Therefore, the required sets \(A_{ik}, B_{jk}\) are constructed. Now put

$$\alpha _i: =\bigcup _k \alpha _{ik},~ A_i:= \bigcup \alpha _i=\bigcup _k A_{ik}, ~ B_j: =\bigcup _k B_{jk},$$
$$ B^\prime _1: = X - (\bigcup _i A_i \cup \bigcup _j B_j),$$

and define a map \(g: X\longrightarrow \Phi _{ml}\) as follows:

$$ g(x):=\left\{ \begin{array}{ll} a_i &{} \text{ if } x\in A_i, \\ b_j &{} \text{ if } x\in B_j,~j\ne 1,\\ b_1 &{} \text{ otherwise } \text{(i.e., } \text{ for } x\in B^\prime _1). \\ \end{array} \right. $$

By (2), (3), (5), (6), (11), \(g\) is well defined; by (9), (10), \(B \subseteq g^{-1}(b_1)\).

To prove that \(g\) is a d-morphism, we check some other properties.

$$(12)\quad X - \bigcup \limits ^l_{i=1}A_i \subseteq \mathbf{d}B_j.$$

Indeed, take an arbitrary \(x\not \in \bigcup \limits ^l_{i=1}A_i \) and show that \(x\in \mathbf{d}B_j\), i.e.,

$$(13)\quad (U-\{ x\}) \cap B_j \not = \varnothing $$

for any neighbourhood \(U\) of \(x\). First assume that \(x\not \in B_j\). Take a basic open \(X_{k+1}\) such that \(x\in X_{k+1} \subseteq U\). Then \(X_{k+1}\not \subseteq \bigcup \limits ^l_{i=1}A_i\), and (8) implies \(B_{j,k+1} \cap X_{k+1} \not = \varnothing .\) Thus \(B_j \cap U \not = \varnothing \). So we obtain (13).

Suppose \(x\in B_j\); then \(x\in B_{jk}\) for some \(k\). Since \(\mathfrak X\) is dense-in-itself and \(\left\{ X_1,\,X_2,\dots \right\} \) is its open base, \(\left\{ X_{s+1}\,\left| \,s\ge k \right. \right\} \) is also an open base (note that every ball in \(\mathfrak X\) contains a smaller ball). So \(x\in X_{s+1}\subseteq U\) for some \(s\ge k\). Since \(x\not \in \bigcup \limits ^l_{i=1}A_i \), we have \(X_{s+1}\not \subseteq \bigcup \limits ^l_{i=1}A_i\), and so \((B_{j,s+1}-B_{js}) \cap X_{s+1} \not = \varnothing \) by (8); thus \((B_j-B_{js}) \cap U \not = \varnothing \). Now \(x\in B_{jk} \subseteq B_{js}\) implies (13).

$$(14)\, \mathbf{d}B_j \subseteq X -\bigcup \limits ^l_{i=1}A_i.$$

Indeed, \(B_j\subseteq -A_i\) by (3), (5), (6). So \(\mathbf{d}B_j \subseteq \mathbf{d}(-A_i) \subseteq -A_i\) since \(A_i \) is open.

Similarly we obtain

$$(15) \, \mathbf{d}B^\prime _1\subseteq X -\bigcup \limits ^l_{i=1}A_i, \qquad \mathbf{d}A_i \subseteq X- \bigcup \limits _{r\not =i}A_r.$$

Also note that

$$(16)\, A_i \subseteq \mathbf{d}A_i$$

since \(A_i\) is open, \(\mathfrak X\) is dense-in-itself. As in (12) we have

$$(17)\, \alpha _i \text{ is } \text{ dense } \text{ at } \text{ every } \text{ point } \text{ of } B_j, B'_1\ \text{(and } \text{ thus } B_j,\ B'_1 \subseteq \mathbf{d}A_i).$$

To conclude that \(g\) is a d-morphism, note that

$$g^{-1}(a_i) = A_i, ~g^{-1}(b_j) = B_j~ (\text{ for } j\not = 1),~ g^{-1}(b_1) = B^\prime _1,$$

and so by (15), (16), (17)

$$\begin{aligned} \mathbf{d}g^{-1}(a_i)&= \mathbf{d}A_i = X - \bigcup \limits _{r\not =i}A_r = g^{-1}(R^{- 1}(a_i)), \end{aligned}$$

and by (12), (14), (15)

$$\begin{aligned} \mathbf{d}g^{-1}(b_j)&= \mathbf{d}B_j = X - \bigcup \limits ^l_{i=1}A_i = g^{-1}(R^{- 1}(b_j))\ \ \text{(for } j\ne 1\text{) },\\ \qquad \qquad \qquad \mathbf{d}g^{-1}(b_1)&= \mathbf{d}B^\prime _1= X - \bigcup \limits ^l_{i=1}A_i = g^{-1}(R^{- 1}(b_1)). \qquad \qquad \qquad \square \end{aligned}$$

Proposition 19 For a finite rooted \(\mathbf {DT_1CK}\) -frame \(F = (W, R, R_D)\) and \(R\) -reflexive points \(w', w'' \in W\), the following holds.

  1. (a)

    If \(X=\left\{ x\in \mathbf{R}^n\mid ||x|| \le r \right\} \), \(n \ge 2\), then there exists \(f: X \twoheadrightarrow ^{dd}F\) such that \(f(\partial X) = \left\{ w'\right\} ;\)

  2. (b)

    If \(0\le r_1<r_2\) and

    $$\begin{aligned} X&= \left\{ x\in \mathbf{R}^n\,\left| \,r_1\le ||x||\le r_2 \right. \right\} , \\ Y'&=\left\{ x\in \mathbf{R}^n\,\left| \,||x||=r_1 \right. \right\} ,\ Y'' =\left\{ x\in \mathbf{R}^n\,\left| \,||x||=r_2 \right. \right\} , \end{aligned}$$

    then there exists \(f: X \twoheadrightarrow ^{dd}F\) such that \(f(Y') = \left\{ w'\right\} \), \(f(Y'') = \left\{ w''\right\} .\)

Proof By induction on \(\left| W\right| \). Let us prove (a) first. There are five cases:

(a1) \(W = R(b)\) (and hence \(b R b\)) and \(b = w'\). Then there exists \(f:X\twoheadrightarrow ^d (W,R)\). Indeed, let \(C\) be the cluster of \(b\) (as a subframe of \((W,R)\)). Then \((W,R)=C\) or \((W,R)=C\cup F_1\cup \ldots \cup F_l\), where the \(F_i\) are generated by the successors of \(C\). If \((W,R)=C\), we apply Proposition 11; otherwise we apply Lemma 15 and IH.

By Proposition 8 it follows that \(R_D\) is universal. And so by Proposition 12(3) \(f\) is a dd-morphism.

(a2) \(W = R(b)\) and not \(w'Rb\). We may assume that \(r=3\). Put

$$ X_1 := \left\{ x \mid ||x|| \le 1\right\} ,~ Y := \left\{ x \mid 1\le ||x|| \le 2 \right\} ,~ X_2 := \left\{ x \mid 2 \le ||x|| \le 3 \right\} . $$

By the case (a1), there is \(f_1: X_1 \twoheadrightarrow ^{dd}F\) with \(f_1(\partial X_1) = \left\{ b\right\} \). Let \(C\) be a maximal cluster in \(R(w')\). By Proposition 11 there is \(g: \mathbf{I}Y \twoheadrightarrow ^d C\). Since \(R(w')\ne W\), we can apply IH to the frame \(F':=F^{w'}_{\forall }\) and construct a dd-morphism \(f_2: X_2 \twoheadrightarrow ^{dd}F'\) with \(f_2(\partial X_2) = \left\{ w'\right\} \). Now since \(f_i(\partial X_i)\subseteq R^{-1}(C)\), the Glueing lemma 16 is applicable. Thus \(f:X\twoheadrightarrow ^d F\) for \(f:=f_1\cup f_2\cup g\) [See Fig. 11.7, Case (a2)]. Note that \(\partial X\subset \partial X_2\), so \(f(\partial X)=f_2(\partial X)=\{w'\}\).

Fig. 11.7
figure 7

dd-morphism f

As in case (a1), \(f\) is a dd-morphism by Proposition 12.

(a3) \((W,R)\) is not rooted. By Lemma 33 there is a global path \(\alpha \) in \(F\) with a single occurrence of every \(R_D\)-irreflexive point. We may assume that \( \alpha = b_0 c_0 b_1 c_1\ldots c_{m-1} b_m\), \(b_m = w'\) and for any \(i<m\), \(c_i \in C_i \subseteq R(b_i) \cap R(b_{i+1})\), where \(C_i\) is an \(R\)-maximal cluster. Such a path is called reduced. For \(0 \le j \le m\) we put \(F_j := F|\overline{R}(b_j)\).

Since \((W,R)\) is not rooted, each \(F_j\) is of smaller size than \(F\), so we can apply the induction hypothesis to \(F_j\). We may assume that

$$X = \left\{ x \mid ||x||\le 2m+1\right\} , \ Y=\left\{ x \mid ||x||=2m+1\right\} .$$

Then put

$$ X_i := \left\{ x \mid ||x|| \le i+1 \right\} \text{ for } 0 \le i \le 2m, $$
$$ Y_i := \partial X_i,~ {\Delta }_i := \mathbf{C}(X_{i} -X_{i-1}) \text{ for } 0 \le i \le 2m. $$

By IH and Proposition 11 there exist

$$\begin{aligned} f_0:&X_0 \twoheadrightarrow ^{dd} F_0 \hbox { such that } f_0(Y_0) = \{c_0\},\\ f_{2j}:&{\Delta }_{2j} \twoheadrightarrow ^{dd}F_{j} \text{ such } \text{ that } f_{2j}(Y_{2j}) = \{c_j\}, \ f_{2j}(Y_{2j-1}) = \{c_{j-1}\}\ \text{ for } 1 \le j \le m,\\ f_{2j-1}:&\mathbf{I}{\Delta }_{2j+1} \twoheadrightarrow ^d{}C_j \text{ for } 0 \le j \le m-1. \end{aligned}$$

One can check that \(f:X\twoheadrightarrow ^{dd} F\) for \(f:=\bigcup \limits _{j=0}^{2m}f_j\) (Fig. 11.7).

(a4) \(W = \overline{R}(b)\), \(\lnot b R_D b\) (and so \(\lnot b R b\)). We may assume that

$$X = \left\{ x \mid ||x||\le 2\right\} , \ Y=\left\{ x \mid ||x||=2\right\} .$$

Then similar to case (a3) put

$$X_0 := X, ~Y_0 := Y, ~ X_i := \left\{ x \mid ||x|| \le \frac{1}{i} \right\} ,~ Y_i := \partial X_i,~ {\Delta }_i := \mathbf{C}(X_{i} -X_{i+1}), ~ (i>0). $$

Consider the frame \(F' {:}= F|W'\), where \(W' = W - \left\{ b\right\} \). Note that \(w'\in W'\) since \(w'Rw'\) by the assumption of Proposition 19. By Lemma 30 \(F'\) is connected, and thus \(F'\vDash \mathbf {DT_1CK}\). By Lemma 33 there is a reduced global path \(\alpha = a_1 \ldots a_m\) in \(F'\) such that \( a_1 = w'\). Let

$$ \upgamma = a_1 a_2 \ldots a_{m-1} a_m a_{m-1} \ldots a_2 a_1 a_2 \ldots $$

be an infinite path shuttling back and forth through \(\alpha \). Rename the points in \(\upgamma \):

(11.6)

Again as in case (a3) we put \(F_j {:}= F | \overline{R} (b_j)\), and assume that \(c_j \in C_j\) and \(C_j\) is an \(R\)-maximal cluster. By IH there exist

$$\begin{aligned} f_0:&{\Delta }_0 \twoheadrightarrow ^{dd} F_0 \hbox { such that } f_0(Y_0) = \{b_0\} = \{w'\}, \ f_1(Y_1) =\{ c_0\},\\ f_{2j}:&{\Delta }_{2j} \twoheadrightarrow ^{dd}F_{j} \hbox { such that } f_{2j}(Y_{2j}) = \{c_{j-1}\}, \ f_{2j}(Y_{2j+1}) = \{c_{j}\}\hbox { for }j>0, \end{aligned}$$

and by Proposition 11 there exist \(f_{2j+1}: \mathbf{I } {\Delta }_{2j+1} \twoheadrightarrow ^d{}C_j\). Put

$$ f(x) := \left\{ \begin{array}{ll} b &{}\hbox { if } x = \mathbf{0},\\ f_{2j}(x)&{}\hbox { if } x \in \Delta _{2j},\\ f_{2j+1}(x)&{}\hbox { if } x \in \mathbf{I}\Delta _{2j+1}, \end{array} \right. $$

One can check that \(f\) is a d-morphism (Fig. 11.8).

Fig. 11.8
figure 8

dd-morphism f

(a5) \(W = \overline{R}(b)\), \(\lnot b R b\) and \(b R_D b\). Then \(R_D\) is universal, \(w'\ne b\). Put

$$ X' := \left\{ x \mid ||x|| < 1\right\} , X_4 := \left\{ x \mid 1 \le ||x|| \le 2\right\} , $$

and let \(X_1,X_2\) be two disjoint closed balls in \(X'\), \(X_3 := X'- X_1 - X_2\).

Let \(C\) be a maximal cluster in \(R(w')\), \(F' := F|R(w')\). Then there exist:

$$\begin{aligned} f_i:&X_i \twoheadrightarrow ^d (W,R) \text{ for } i=1,2 \text{ such } \text{ that } \ f_i(\partial X_i) = \left\{ w'\right\} , \hbox { by case (a4),}\\ f_3:&X_3 \twoheadrightarrow ^d C, \hbox { by Proposition 11,}\\ f_4:&X_4 \twoheadrightarrow ^{dd}F' \text{ such } \text{ that } \ f_4(\partial X_4) = \left\{ w'\right\} , \hbox { by the induction hypothesis.} \end{aligned}$$

Put \(f:=f_1\cup f_2\cup f_3\cup f_4\) (Fig. 11.8). Then \(f(\partial X) = \left\{ w'\right\} \).

By Lemma 16 (b), \(f_1\cup f_2:X_1\cup X_2\twoheadrightarrow ^d F\), and hence \(f:X\twoheadrightarrow ^d F\) by Lemma 16 (a). \(f\) is manifold at \(b\), thus it is a dd-morphism by Lemma 18.

Now we prove (b). There are three cases.

(b1) \(w' = w'' = b\) and \(W = R(b)\). The argument is the same as in case (a1), using Proposition 11, Lemma 15, the induction hypothesis, and Proposition 12.

(b2) \(w' = w'' = b\), but \(W \ne R(b)\). Consider a maximal cluster \(C \subseteq R (b)\). Since all spherical shells for different \(r_1\) and \(r_2\) are homeomorphic, we assume that \(r_1 = 1\), \(r_2 = 4\). Consider the sets

$$\begin{aligned} X_1 := \left\{ x \mid 1 \le ||x|| \le 2\right\} ,\ \ X' := \left\{ x \mid 2 < ||x|| < 3\right\} ,\ \ X_3 := \left\{ x \mid 3 \le ||x|| \le 4\right\} , \end{aligned}$$

and let \(X_0\subset X'\) be a closed ball, \(X_2 := X' - X_0\). Let \(F' :=F| R (b)\). There exist

$$\begin{aligned} f_1:&X_1 \twoheadrightarrow ^{dd}F' \text{ such } \text{ that } f_1(\partial X_1) = \left\{ b\right\} \!, \hbox { by case (b1)},\\ f_2:&X_2 \twoheadrightarrow ^d C, \hbox { by Proposition 11},\\ f_3:&X_3 \twoheadrightarrow ^{dd}F' \text{ such } \text{ that } \ f_3(\partial X_3) = \left\{ b\right\} \!, \hbox { by case (b1)},\\ f_0:&X_0 \twoheadrightarrow ^{dd}F \text{ such } \text{ that } \ f_4(\partial X_0) = \left\{ b\right\} \!, \hbox { by statement (a) for }F. \end{aligned}$$

One can check that \(f:X\twoheadrightarrow ^{dd}F\) for \(f:=f_0\cup f_1\cup f_2\cup f_3\).

(b3) \(w' \ne w''\) and for some \(b\in W\), \(W = R(b)\), so \(F\) has an \(R\)-reflexive root. Let

$$ F_1 := F|R(w'), ~ F_2 := F|R(w''), $$

and let \(C_i\) be an \(R\)-maximal cluster in \(F_i\) for \(i\in \left\{ 1,2\right\} .\)

We assume that \(r_1 = 1\), \(r_2 = 6\) and consider the sets

$$\begin{aligned} X_i&:= \left\{ x \mid i \le ||x|| \le i+1\right\} , \ i \in \left\{ 1, \ldots , 5.\right\} \end{aligned}$$

By case (b1) and Proposition 11 we have

$$\begin{aligned} f_1:&~ X_1 \twoheadrightarrow ^{dd}F_1 \text{ such } \text{ that } f_1(\partial X_1) = \left\{ w'\right\} \!,&f_2:&~ \mathbf{I}X_2 \twoheadrightarrow ^d C_1,\\ f_3:&~ X_3 \twoheadrightarrow ^{dd}F \text{ such } \text{ that } \ f_3(\partial X_3) = \left\{ b\right\} \!,&f_4:&~ \mathbf{I}X_4 \twoheadrightarrow ^d C_2, \\ f_5:&~ X_5 \twoheadrightarrow ^{dd}F_2 \text{ such } \text{ that } f_1(\partial X_5) = \left\{ w''\right\} \!. \end{aligned}$$

One can check that \(f:X\twoheadrightarrow ^{dd}F\) for \(f:= \bigcup \limits _{i=1}^5 f_i\) [Fig. 11.9, Case (b3)].

Fig. 11.9
figure 9

dd-morphism f

(b4) \(w' \ne w''\) and \(W \ne R(b)\) for any \(b\in W\). By Lemma 32 there is a reduced path \(\alpha = b_0 c_0 b_1 \ldots c_{m-1} b_m\) from \(b_0 = w'\) to \(b_m = w''\) that does not contain \(R_D\)-irreflexive points, \(c_i \in C_i\), where \(C_i\) is an \(R\)-maximal cluster. We may also assume that

(11.7)

Indeed, if the frame \((W,R)\) is not rooted, then (11.7) obviously holds. If \((W,R)\) is rooted, then its root \(r\) is irreflexive and by Lemma 30, \(R(r)\) is connected, so there exists a path \(\alpha \) in \(R(r)\) satisfying (11.7). Put

$$ F_0 := F,~ F_j := F|R(b_j), 1 \le j \le m. $$

Assuming that \(r_1 = 1\), \(r_2 = 2m+1\) we define

$$\begin{aligned} X_i&:= \left\{ x\,\left| \,||x|| \le i +1 \right. \right\} ,~ Y_i := \partial X_i~ (\text{ for } 0 \le i \le 2m+1),\\ {\Delta }_i&:= \mathbf{C}(X_{i+1} -X_{i}) \ (\text{ for } 0\le i\le 2m). \end{aligned}$$

By cases (b2), (b1), Proposition 11, and the induction hypothesis there exist

$$\begin{aligned}&f_0: {\Delta }_0 \twoheadrightarrow ^{dd}F=F_0 \hbox { such that } f_0(Y_0) = f_0(Y_1) = \left\{ w'\right\} \!; \\&f_{2j}: {\Delta }_{2j} \twoheadrightarrow ^{dd}F_{j} \hbox { such that } f_{2j}(Y_{2j+1}) = \left\{ c_{j}\right\} , \ f_{2j}(Y_{2j}) = \{c_{j-1}\}\ (1 \le j \le m);\\&f_{2j-1}: \mathbf{I}{\Delta }_{2j-1} \twoheadrightarrow ^d C_{j-1} \ (1 \le j \le m),\\&f_{2m}: {\Delta }_{2m} \twoheadrightarrow ^{dd}F_m \hbox { such that } f_{2m}(Y_{2m}) = \left\{ c_{m}\right\} , \ f_{2m}(Y_{2m+1}) = \left\{ w''\right\} \!. \end{aligned}$$

We claim that \(f:X\twoheadrightarrow ^{dd}F\) for \(f:= \bigcup \limits _{i=0}^{2m} f_i\) (Fig. 11.10). First, we prove by induction using Lemma 16 (see the previous cases) that \(f\) is a d-morphism. Note that \(f(Y')= f(Y_0) = \left\{ w'\right\} \) and \(f(Y'')= f(Y_{2m+1}) = \left\{ w''\right\} \).

Fig. 11.10
figure 10

dd-morphism f, case (b4)

Second, there are no \(R_D\)-irreflexive points in \(\alpha \), so all preimages of \(R_D\)-irreflexive points are in \(\Delta _0\); since \(f_0\) is a dd-morphism, \(f\) is 1-fold at any \(R_D\)-irreflexive point and manifold at all the others. Thus \(f\) is a dd-morphism by Proposition 12. \(\square \)

Rights and permissions

Reprints and permissions

Copyright information

© 2014 Springer Science+Business Media Dordrecht

About this chapter

Cite this chapter

Kudinov, A., Shehtman, V. (2014). Derivational Modal Logics with the Difference Modality. In: Bezhanishvili, G. (eds) Leo Esakia on Duality in Modal and Intuitionistic Logics. Outstanding Contributions to Logic, vol 4. Springer, Dordrecht. https://doi.org/10.1007/978-94-017-8860-1_11

Download citation

Publish with us

Policies and ethics