Announcement as effort on topological spaces

Hans van Ditmarsch

hans.vanditmarsch@loria.fr

Sophia Knight
sophia.knight@gmail.com

Aybke zgn
aybuke.ozgun@loria.fr

LORIA, CNRS - Universit de Lorraine

ABSTRACT
We propose a multi-agent logic of knowledge, public and
arbitrary announcements, that is interpreted on topological
spaces in the style of subset space semantics. The arbitrary announcement modality functions similarly to the effort modality in subset space logics, however, it comes with
intuitive and semantic differences. We provide axiomatizations for three logics based on this setting, and demonstrate
their completeness.

Keywords
Topology, subset space logic, dynamic epistemic logic, arbitrary (public) announcements

1.

INTRODUCTION

In [13], Moss et al. introduce a bi-modal logic with language

::= p | | | K | 2,
called subset space logic (SSL), in order to formalize reasoning about sets and points together in one modal system.
The main interest in their investigation lies in spatial structures such as topological spaces and using modal logic and
the techniques behind for spatial reasoning, however, they
also have a strong motivation from epistemic logic. While
the modality K is interpreted as knowledge, 2 intends to
capture the notion of effort, i.e., any action that results in
increase in knowledge. They propose subset space semantics for their logic. A subset space is defined to be a pair
(X, O), where X is a non-empty domain and O is a collection of subsets of X (not necessarily a topology), wherein
the modalities K and 2 are evaluated with respect to pairs
of the form (x, U ), where x U O. According to subset
space semantics, given a pair (x, U ), the modality K quantifies over the elements of U , whereas 2 quantifies over all
open subsets of U that include the actual world x. Therefore,
while knowledge is interpreted locally in a given observation set U , effort is read as open-set-shrinking where more
effort corresponds to a smaller neighbourhood, thus, a possible increase in knowledge. The schema 3K states that
after some effort the agent comes to know where effort can
be in the form of measurement, observation, computation,
approximation [13, 8, 14, 5], or announcement [15, 1, 16].
The epistemic motivation behind the subset space semantics and the dynamic nature of the effort modality suggests
TARK 2015, CMU, Pittsburg, USA.
Copyright 2015 by the authors.

a link between SSL and dynamic epistemic logic, in particular dynamics known as public announcement [4, 5, 3,
19, 6]. The works [4, 5, 3] propose modelling public announcements on subset spaces by deleting the states or the
neighbourhoods falsifying the announcement. This dynamic
epistemic method is not in the spirit of the effort modality:
dynamic epistemic actions result in global model change,
whereas the effort modality results in local neighbourhood
shrinking. Hence, it is natural to search for an open-setshrinking-like interpretation of public announcements on
subset spaces. To best of our knowledge, Wang and
Agotnes
[19] were the first to propose semantics for public announcements on subset spaces in the style of the effort modality,
although this is not necessarily on topological spaces. Bjorndahl [6] then proposed a revised version of the [19] semantics.
In contrast to the aforementioned proposals, Bjorndahl uses
models based on topological spaces to interpret knowledge
and information change via public announcements. He considers the language
::= p | | | K | int() | [],
where int() means is true and can be announced, and
where [] means after public announcement of , .
In [1], Balbiani et al. introduce a logic to quantify over
announcements in the setting of epistemic logic based on
the language (with single-agent version here)
::= p | | | K | [] | 2.
In this case, unlike above, 2 means after any announcement, (is true) so that 2 quantifies over epistemically definable subsets (2-free formulas of the language) of a given
model. In this case, 3K again means that the agent comes
to know , but in the interpretation that there is a formula
such that after announcing it the agent knows . What
becomes true or known by an agent after an announcement
can be expressed in this language without explicit reference
to the announced formula.
Clearly, the meaning of the effort 2 modality and of the arbitrary announcement 2 modality are related in motivation.
In both cases, interpreting the modality requires quantification over sets. Subset-space-like semantics provides natural
tools for this. In [16], we extended Bjorndahls proposal [6]
with an arbitrary announcement modality
::= p | | | K | int() | [] | 2
and provided topological semantics for the 2 modality, and
proved completeness for the corresponding single-agent logic
AP ALint .

In the current proposal we generalize this approach to a

multi-agent setting. Multi-agent subset space logics have
been investigated in [11, 12, 4, 18]. There are some challenges with such a logic concerning the evaluation of higherorder knowledge. The general setup is for any finite number of agents, but to demonstrate the challenges, consider
the case of two agents. Suppose for each of two agents i
and j there is an open set such that the semantic primitive
becomes a triple (x, Ui , Uj ) instead of a pair (x, U ). Now
j Ki p, for agent i knows that
consider a formula like Ki K
agent j considers possible that agent i knows proposition p.
j Ki p must be
If this is true for a triple (x, Ui , Uj ), then K
true for any y Ui ; but y may not be in Uj , in which case
j Ki p.
(y, Ui , Uj ) is not well-defined: we cannot interpret K
Our solution to this dilemma is to consider neighbourhoods
that are not only relative to each agent, as usual in multiagent subset space logics, but that are also relative to each
state. This amounts to, when shifting the viewpoint from
x to y Ui , in (x, Ui , Uj ), we simultaneously have to shift
the neighbourhood (and not merely the point in the actual
neighbourhood) for the other agent. So we then go from
(x, Ui , Uj ) to (y, Ui , Vj ), where Vj may be different from Uj .
If they are different, their intersection should be empty.
In order to define the evaluation neighbourhood for each
agent with respect to the state in question, we employ a
technique inspired by the standard neighbourhood semantics
[7]. We use a set of neighbourhood functions, determining the
evaluation neighbourhood relative to both the given state
and the corresponding agent. These functions need to be
partial in order to render the semantics well-defined for the
dynamic modalities in the system.
In Section 2 we define the syntax, structures, and semantics of our multi-agent logic of arbitrary public announcements, AP ALint , interpreted on topological spaces equipped
with a set of neighbourhood functions. Without arbitrary
announcements we get the logic P ALint , and with neither
arbitrary nor public announcements, the logic ELint . In this
section we also show some typical validities of the logic, and
give a detailed example. In Section 3 we give axiomatizations for the logics: P ALint extends ELint and AP ALint
extends P ALint . In Section 4 we demonstrate completeness
for these logics. The completeness proof for the epistemic
version of the logic, ELint , is rather different from the completeness proof for the full logic AP ALint . We then compare
our work to that of others (Section 5) and conclude.

2.

THE LOGIC AP ALint

We define the syntax, structures, and semantics of our

logic. From now on, Prop is a countable set of propositional
variables and A a finite and non-empty set of agents.

2.1

Syntax

Definition 1. The language LAP ALint is defined by

::= p | | | Ki | int() | [] | 2
where p Prop and i A. Abbreviations for the connectives , and are standard, and is defined as abbre i for Ki , and 3 for
viation by p p. We employ K
2. We denote the non-modal part of LAP ALint (without
the modalities Ki , int, [] and 2) by LP l , the part without
2 by LP ALint , and the part without 2 and [] by LELint .

Necessity forms [10] allow us to select unique occurrences of

a subformula in a given formula (unlike in uniform substitution). They will be used in the axiomatization (Section
3).
Definition 2. Let LAP ALint . The necessity forms
are inductively defined as
(]) := ] | (]) | Ki (]) | int((])) | [](]).
It is not hard to see that each necessity form (]) has a
unique occurrence of ]. Given a necessity form (]) and a
formula LAP ALint , the formula obtained by replacing ]
by is denoted by ().
In the completeness proof (Section 4) we use a complexity measure on formulas based on the size and 2-depth of
formulas where the size of a formula is a weighted count of
the number of symbols and 2-depth counts the number of
the 2-modalities occurring in a formula. The measure was
first introduced in [2].
Definition 3. The size S() of a formula LAP ALint
is defined as: S(p) = 1, S() = S() + 1, S( ) =
S() + S(), S(Ki ) = S() + 1, S(int()) = S() + 1,
S([]) = S() + 4S(), and S(2) = S() + 1.
The factor 4 in the clause for [] is to ensure Lemma 7.
Although the choice of the number 4 might seem arbitrary,
it is the smallest natural number guaranteeing the desired
result (see the proof of Lemma 7).
Definition 4. The 2-depth of a formula LAP ALint ,
denoted by d(), is defined as: d(p) = 0, d() = d(),
d( ) = max{d(), d()}, d(Ki ) = d(), d(int()) =
d(), d([]) = max{d(), d()}, and d(2) = d() + 1.
We now define three order relations on LAP ALint based on
the size and 2-depth of the formulas.
Definition 5. For any , LAP ALint ,
<S iff S() < S()
<d iff d() < d()
<S
d iff (either d() < d(), or d() = d() and
S() < S())
We let Sub() denote the set of subformulas of a given formula .
Lemma 6. For any , LAP ALint ,
1. <S , <d , <S
d are well-founded strict partial orders between formulas in LAP ALint ,
2. Sub() implies <S
d ,
3. int() <S
[],
d
4. LP ALint iff d() = 0,
5. LP ALint implies [] <S
d 2.
Lemma 7. For any , , LAP ALint and i A,
1. [] <S
d [],
2. int([]) <S
d []int(),
3. Ki [] <S
d []Ki ,
4. [[]int()] <S
d [][].
Proof. We only prove Lemma 7.4. The proof demonstrates why in the [] clause of Definition 3, 4 is the smallest natural number guaranteeing the result.

By Definition 3, we have that S([[]int()]) = S()+

4S() + 4S() + 9 and that S([][]) = S() + 4S() +
16S(). As for any LAP ALint , 1 S(), it follows that
4S()+9 4S()+9S() = 13S() < 16S(). Further, we
observe that d([[]int()]) = max{d(), d(), d()} =
d([][]). (This is similar in the first three items.)

2.2

Background

In this section, we introduce the topological concepts that

will be used throughout this paper. All the concepts in this
section can be found in [9].
Definition 8. A topological space (X, ) is a pair consisting of a non-empty set X and a family of subsets of X
satisfying and X , and closed under finite intersections and arbitrary unions.
The set X is called the space. The subsets of X belonging
to are called open sets (or opens) in the space; the family
of open subsets of X is also called a topology on X. If for
some x X and an open U X we have x U , we say
that U is an open neighborhood of x.
A point x is called an interior point of a set A X if
there is an open neighborhood U of x such that U A. The
set of all interior points of A is called the interior of A and
denoted by Int(A). We can then easily observe that for any
A X, Int(A) is the largest open subset of A.
Definition 9. A family B is called a base for a
topological space (X, ) if every non-empty open subset of X
can be written as a union of elements of B.
Given any family = {A | I} of subsets of X,
there exists a unique, smallest topology () with ()
[9, Th. 3.1]. The family () consists of , X, all finite
intersections of the A , and all arbitrary unions of these
finite intersections. is called a subbase for (), and ()
is said to be generated by . The set of finite intersections
of members of forms a base for ().

2.3

Structures

In this section we define our multi-agent models based on

topological spaces.
Definition 10. Given a topological space (X, ), a neighbourhood function set on (X, ) is a set of partial functions : X * A such that for all x, y Dom(), for
all i A, and for all U :
1. (x)(i) ,
2. x (x)(i),
3. (x)(i) Dom(),
4. if y (x)(i) then (x)(i) = (y)(i),
5. |U ,
where |U is the partial function with Dom(|U ) = Dom() U
and |U (x)(i) = (x)(i) U . We call the elements of
neighbourhood functions.
Definition 11. A topological model with functions (or
in short, a topo-model) is a tuple M = (X, , , V ), where
(X, ) is a topological space, a neighbourhood function set,
and V : P rop X a valuation function. We refer to the
part X = (X, , ) without the valuation function as a topoframe.

A pair (x, ) is a neighbourhood situation if x Dom()

and (x)(i) is called the epistemic neighbourhood at x of
agent i. If (x, ) is a neighbourhood situation in M we write
(x, ) M. Similarly, if (x, ) is a neighbourhood situation
in X we write (x, ) X .
Lemma 12. For any (X, , ) and , Dom() .

2.4

Semantics

Definition 13. Given a topo-model M = (X, , , V ) and

a neighbourhood situation (x, ) M, the semantics for the
language LAP ALint is defined recursively as:
M, (x, ) |= p
M, (x, ) |=
M, (x, ) |=
M, (x, ) |= Ki
M, (x, ) |= int()
M, (x, ) |= []
M, (x, ) |= 2

x V (p)
not M, (x, ) |=
M, (x, ) |= and M, (x, ) |=
(y (x)(i))(M, (y, ) |= )
x Int[[]]
M, (x, ) |= int()
M, (x, ) |=
iff ( LP ALint )(M, (x, ) |= [])

iff
iff
iff
iff
iff
iff

where p Prop, [[]] = {y Dom() | M, (y, ) |= }

and : X * A such that Dom( ) = Int[[]] and
(x)(i) = (x)(i) Int[[]] .
The updated neighbourhood function is the restriction of
to the open set Int[[]] , i.e., for all x X, (x)(i) =
|Int[[]] (x)(i).
A formula LAP ALint is valid in a topo-model M, denoted M |= , iff M, (x, ) |= for all (x, ) M; is valid,
denoted |= , iff for all topo-models M we have M |= .
Soundness and completeness with respect to topo-models
are defined as usual.
Let us now elaborate on the structure of topo-models and
the above semantics we have proposed for LAP ALint . Given
a topo-model (X, , , V ), the epistemic neighbourhoods of
each agent at a given state x are determined by (partial)
functions : X * A assigning an open neighbourhood
to the state in question for each agent. We allow for partial
functions in , and close under taking restricted functions |U where U (see Definition 10, condition 5), so
that updated neighbourhood functions are guaranteed to be
well-defined elements of . As in the standard subset space
semantics, by picking a neighbourhood situation (x, ), we
first localize our focus to an open subdomain, in fact to
Dom(), including the state x and the epistemic neighbourhood of each agent at x determined by . Then the function
(x) designates an epistemic neighbourhood for each agent
i in A. It is guaranteed that every agent i is assigned a
neighbourhood by at every state x in Dom(), since each
(x) is defined to be a total function from A to . Moreover,
condition 2 of Definition 10 ensures that cannot be an epistemic neighbourhood, i.e., (x)(i) 6= for all x Dom().
Finally, conditions 2 and 4 of Definition 10 make sure that
the S5 axioms for each Ki are sound with respect to all
topo-models.
We now provide some semantic results. As usual in the
subset space setting, truth of non-modal formulas only depends on the state in question.
Proposition 14. Give a topo-model M = (X, , , V ),
neighbourhood situations (x, 1 ), (x, 2 ) M, and a formula
LP l . Then (x, 1 ) |= iff (x, 2 ) |= .

Proposition 15. Given M = (X, , , V ), and

LAP ALint . Then [[int()]] = Int[[]] .
Proof.
[[int()]]

= {y Dom() | (y, ) |= int()}

= {y Dom() | y Int[[]] }
= Int[[]] (since Int[[]] Dom())

A corollary is that Int[[int()]] = IntInt[[]] = Int[[]] .

Proposition 16.
1. |= [] [int()]
2. |= (int() hiint()) hiint()
Proposition 17.

1. [[]] = [[hi]]
2. = int()
3. ( ) = hiint()

2.5

Example

We illustrate our logic by a multi-agent version of Bjorndahls convincing example in [6] about the jewel in the tomb.
Indiana Jones (i) and Emile Belloq (e) are both scouring
for a priceless jewel placed in a tomb. The tomb could either contain a jewel or not, the tomb could have been rediscovered in modern times or not, and (beyond [6]), the
tomb could be in the Valley of Tombs in Egypt or not. The
propositional variables corresponding to these propositions
are, respectively, j, d, and t. We represent a valuation of
these variables by a triple xyz, where x, y, z {0, 1}. Given
carrier set X = {xyz | x, y, z {0, 1}}, the topology that
we consider is generated by the base consisting of the subsets
{000, 100, 001, 101}, {010}, {110}, {011}, {111}. The idea
is that one can only conceivably know (or learn) about the
jewel or the location, on condition that the tomb has been
discovered. Therefore, {000, 100, 001, 101} has no strict subsets besides empty set: if the tomb has not yet been discovered, no one can have any information about the jewel or
the location.
A topo-model M = (X, , , V ) for this topology (X, )
has as the set of all neighbourhood functions that are
partitions of X for both agents, and restrictions of these
functions to open sets. A typical describes complete ignorance of both agents and is defined as (s)(i) =
(s)(e) = X. This corresponds most to the situation described in [6]. A more interesting neighbourhood situation
in this model is one wherein Indiana and Emile have different knowledge. Let us assume that Emile has the advantage over Indiana so far, as he knows the location of the
tomb but Indiana doesnt. This is the 0 such that for all
x X, 0 (x)(i) = X whereas the partition for Emile consists of sets {101, 100, 001, 000}, {111, 011}, {110, 010}, i.e.,
0 (111)(e) = {111, 011}, etc.
We now can evaluate what Emile knows about Indiana at
111, and confirm that this goes beyond Emils initial epistemic neighbourhood. This situation however does not create any problems in our setting since Indianas epistemic
neighbourhoods will be determined relative to the states in
Emiles initial neighbourhood. Firstly, Emile knows that the
tomb is in the Valley of Tombs in Egypt
M, (111, 0 ) |= Ke t

and he also knows that Indiana does not know that

M, (111, 0 ) |= Ke (Ki t Ki t)
i t and
The latter involves verifying M, (111, 0 ) |= K
0

M, (111, ) |= Ki t. And this is true because 0 (111)(i) =

X, and 000, 001 X, and while M, (001, 0 ) |= t, we also
have M, (000, 0 ) |= t. We can also check that Emile knows
that Indiana considers it possible that Emile doesnt know
the tombs location
i (Ke t Ke t)
M, (111, 0 ) |= Ke K
Announcements will change their knowledge in different ways.
Consider the announcement of j. This results in Emile knowing everything but Indiana still being uncertain about the
location.
M, (111, 0 ) |= [j](Ke (j d t) Ki (j d) Ki (t Ki t))
Model checking this involves computing the epistemic neighbourhoods of both agents given by the updated neighbour0
hood function (0 )j at 111. Observe that Int[[j]] = {111, 110}.
0
Therefore, (0 )j (111)(e) = Int[[j]] 0 (111)(e) = {111} and
0
(0 )j (111)(i) = Int[[j]] 0 (x)(i) = {111, 110}.
There is an announcement after which Emile and Indiana
know everything (for example the announcement of j t):
M, (111, ) |= 3(Ke (j d t) Ki (j d t))
As long as the tomb has not been discovered, nothing will
make Emile (or Indiana) learn that it contains a jewel or
where the tomb is located:
M |= d 2((Ke j Ke j) (Ke t Ke t))

3.

AXIOMATIZATION

We now provide the axiomatizations of ELint , P ALint ,

and AP ALint , and prove their soundness and completeness
with respect to the proposed semantics.
(P)
(K-K)
(K-T)
(K-4)
(K-5)
(int-K)
(int-T)
(int-4)
(Kint )
(R1)
(R2)
(R3)
(R4)
(R5)
(R6)
(R7)
(DR1)
(DR2)
(DR3)
(DR4)
(DR5)

all instantiations of propositional tautologies

Ki ( ) (Ki Ki )
Ki
Ki Ki Ki
Ki Ki Ki
int( ) (int() int())
int()
int() int(int())
Ki int()
[]p (int() p)
[] (int() [])
[]( ) [] []
[]int() (int() int([]))
[]Ki (int() Ki [])
[][] [[]int()]
2 []
where LP ALint
From and , infer
From , infer Ki
From , infer int()
From , infer []
From ([]) for all LP ALint , infer (2)

Table 1: Axiomatizations ELint , P ALint , and AP ALint

Definition 18. The axiomatization AP ALint is given in

Table 1. The axiomatization P ALint is the one without
(DR5) and (R7). We get ELint if we further remove axioms (R1)-(R6) and the rule (DR4).
The parts (DR1) to (DR5) are the derivation rules and
the other parts are the axioms. A formula is a theorem of
AP ALint , notation ` , if it belongs to the smallest set of
formulas containing the axioms and closed under the derivation rules. (Similarly for ELint and P ALint .)
Lemma 19. Axiomatization AP ALint satisfies substitution of equivalents. If ` , then ` [p/] [p/].
Proof. In the above, [p/] means uniform substitution
of for p. The proof is not trivial but proceeds along similar
lines as for public announcement logic, see [17].
Proposition 20. [] int() is a theorem of AP ALint .
Proposition 21. AP ALint is sound with respect to the
class of all topo-models.
Proof. Let M = (X, , , V ) be a topo-model, (x, )
M and , , LAP ALint . We show three cases.
(Kint ) Suppose (x, ) |= Ki . This means, (y, ) |=
for all y (x)(i). Hence, (x)(i) [[]] . By Definition 10,
(x)(i) is an open neighbourhood of x, therefore we obtain
x Int[[]] , i.e., (x, ) |= int().
(R7) Let LP ALint and suppose (x, ) |= 2. By the
semantics, we have (x, ) |= 2 iff ( LP ALint )((x, ) |=
[]). Therefore, in particular, (x, ) |= [].
(DR5) Suppose ([]) is valid for all LP ALint .
The proof follows by induction on the complexity of (]).
In case (]) = ], we have ([]) = []. By assumption,
we have that [] is valid for all LP ALint . This implies
M, (x, ) |= [] for all LP ALint , all topo-models M,
and (x, ) M. Therefore, by the semantics, M, (x, ) |=
2, i.e., M, (x, ) |= (2). All other, inductive, cases are
elementary.
Corollary 22. The axiomatizations ELint and P ALint
are sound with respect to the class of all topo-models.

4.

COMPLETENESS

We now show completeness for ELint , P ALint , and AP ALint

with respect to the class of all topo-models. Completeness of
ELint is shown in a standard way via a canonical model construction and a Truth Lemma that is proved by induction on
formula complexity. Completeness for P ALint is shown by
reducing each formula in LP ALint to an equivalent formula of
LELint . The proof of the completeness for AP ALint becomes
more involved. Reduction axioms for public announcements
no longer suffice in the AP ALint case, and the inductive
proof needs a subinduction where announcements are considered. Moreover, the proof system of AP ALint has an
infinitary derivation rule, namely the rule (DR5), and given
the requirement of closure under this rule, the maximally
consistent sets for that case are defined to be maximally consistent theories (see, Section 4.2). Lastly, the Truth Lemma
requires the more complicated complexity measure on formulas defined in Section 2. There, we need to adapt the
completeness proof of [2] to our setting.

4.1

Completeness of ELint and P ALint

For LELint we define consistent and maximally consistent

sets in the usual way, see e.g. [6] for details, and the multiagent aspect does not complicate the definition. Let X c be
the set of all maximally consistent sets of ELint . We define relations i on X c as x i y iff LELint (Ki
x iff Ki y). Notice that the latter is equivalent to:
LELint (Ki x implies y) since Ki is an S5
modality. As each Ki is of S5 type, every i is an equivalence relation, hence, it induces equivalence classes on X c .
Let [x]i denote the equivalence class of x induced by the
relation i . Moreover, we define
b = {y X c | y}.
Observe that x
b iff x.
Lemma 23 (Lindenbaums Lemma). Each consistent
set can be extended to a maximally consistent set.
Definition 24. We define the canonical model
X c = (X c , c , c , V c ) as follows:
X c is the set of all maximally consistent sets;
c is the topological space generated by the subbase
\ | x X c , LEL and i A};
= {[x]i int()
int
x V c (p) iff p x, for all p Prop;
c = { |U | U c }, where we define : X c
A c as (x)(i) = [x]i , for x X c and i A.
\ = X c , we have [x]i int(>)
\ =
Observe that, since int(>)
[x]i for each i. Therefore, each [x]i is an open subset
of X c . Moreover, the elements of c satisfy the required
properties given in Definition 10.
Lemma 25 (Truth Lemma). For every LELint and
for each x X c , x iff X c , (x, ) |= .
Proof. Cases for the propositional variables and Booleans
are straightforward. We only show the cases for Ki and int.
Case := Ki
()
Suppose Ki x and let y (x)(i). Since
y (x)(i) = [x]i , by definition of i , we have Ki y.
Then, by T-axiom for Ki , we obtain y. Then, by IH,
X c , (y, ) |= . Therefore X c , (x, ) |= Ki .
() Suppose Ki 6 x. Then, {Ki | Ki x} {}
is a consistent set. We can then extend it to a maximally
consistent set y. As {Ki | Ki x} y, we have y
[x]i meaning that y (x)(i). Moreover, since y,
6 y. Therefore, we have a maximally consistent set y
(x)(i) such that 6 y. By (IH), X c , (y, ) 6|= . Hence,
X c , (x, ) 6|= Ki .
Case := int()
\
() Suppose int() x. Consider the set [x]i int()
\ and [x]i int()
\
for some i A. Obviously, x [x]i int()
\ Since
is open (since it is in ). Now let y [x]i int().
\ int() y. Then, by (int -T), since y is maximal
y int(),
consistent, we have y. Thus, by IH, we have (y, ) |= .

\ [[]] .
Therefore, y [[]] . This implies [x]i int()
\ c , we have x Int[[]] , i.e.,
And, since x [x]i int()
(x, ) |= int().

() Suppose (x, ) |= int(), i.e., x Int[[]] . Recall

that the set of finite intersections of the elements of forms
a base, which we denote by B , for c . x Int[[]] implies

that there exists an open U B such that x U [[]] .

Given the construction of B , U is of the form
\

U=

[x1 ]i . . .

[xk ]i

iIn

iI1

\
int()

Formfin

where I1 , . . . , In are finite subsets of A, x1 . . . xk X c and

Formfin is a finite subset of LELint . Since int is a normal
modality, we can simply write
\
\
\
U=
[x1 ]i . . .
[xk ]i int(),
iI1

where

iIn

:= . Since x is in each [xj ]i with 1 j

Formfin

k, we have [xj ]i = [x]i for all such j. Therefore, we have

\
\ [[]] ,
x U = ( [x]i ) int()

A theory x is said to be consistent iff 6 x. A theory x

is maximally consistent iff x is consistent and any set of
formulas properly containing x is inconsistent.
Observe that AP ALint constitutes the smallest theory. Moreover, maximally consistent theories of AP ALint posses the
usual properties of maximally consistent sets:
Proposition 29. For any maximally consistent theory x,
6 x iff x, and x iff x and x.
In the setting of our axiomatization based on the infinitary
rule (DR5), we will say that a set x of formulas is consistent
iff there exists a consistent theory y such that x y. Obviously, maximal consistent theories are maximal consistent
sets of formulas. Under the given definition of consistency
for sets of formulas, maximal consistent sets of formulas are
also maximal consistent theories.

iI

where I = I1 In .
T
\ then y.
This implies, for all y ( [x]i ), if y int()
S iI
{Ki | Ki x} ` int() .
From this, we can say
iI
S
{Ki | Ki x} such
Then, there is a finite subset
iI
V
that `
(int() ). It then follows:

1. ` int(

(int() ))

(DR3)

2. ` int(
) int(int() ))
(int-K) and (DR1)
V
3. ` (
int()) int(int() )) (int-K)

ObserveSthat each is of the form Kj for some

Kj
{Ki | Ki x} and we have ` Ki int(Ki ).
iI
V
Therefore, ` (
) int(int() )). Thus, since

V
x (by x), we have int(int() )) x. Then,

Definition 30. Let LAP ALint and i A. Then x +

:= { | x} and Ki x := { | Ki x}.
Lemma 31. For any theory x of AP ALint and
LAP ALint , x + is a theory and it contains x and ,
and Ki x is a theory.
Lemma 32. Let LAP ALint . For all theories x, x +
is consistent iff 6 x.
Proof. Let LAP ALint and x be a theory. Then
x iff x (as is a theorem) iff x+.
Therefore, x + is inconsistent iff x, i.e., x + is
consistent iff 6 x.
Lemma 33 (Lindenbaums Lemma [1]). Each consistent theory can be extended to a maximal consistent theory.
Lemma 34. If Ki 6 x, then there is a maximally consistent theory y such that Ki x y and 6 y.

Theorem 26. ELint is complete with respect to the class

of all topo-models.

Proof. Let LAP ALint and x be such that Ki 6

x. Thus, 6 Ki x. Hence, by Lemma 32, Ki x + is
consistent. Then, by Lemma 33, there exists a maximally
consistent set y such that Ki x + y. Therefore Ki x y
and 6 y.

Theorem 27. P ALint is complete with respect to the class

of all topo-models.

Lemma 35. For all LAP ALint and all maximally consistent theories x, 2 x iff for all LP ALint , [] x.

Proof. This follows from Theorem 26 by reduction in a

standard way. The occurrences of the modality int on the
right-hand-side of the reduction axioms (axioms (R1)-(R6))
should not lead to any confusion: extending the complexity
measure defined in [17, Definition 7.21 p. 187] to the language LP ALint by adding the same complexity measure for
the modality int as for Ki gives us the desired result.

Proof. Let LAP ALint and x be a maximally consistent theory.

() Suppose 2 x. Then, by (R7) and (DR1), we
have [] x for all LP ALint .
() Suppose [] x for all LP ALint . Consider the
necessity form ]. By assumption, ]([]) for all LP ALint .
Then, since x is closed under (DR5), ](2) x, i.e., 2 x
as well.

by (int-K), (DR1) and since ` int(int()) int() and

\ (i.e., int() x) , we obtain int() x.
x int()

4.2

Completeness of AP ALint
We now reuse the technique of [2] in the setting of topological semantics. Given the closure requirement under derivation rule (DR5) it seems more proper to call maximally consistent sets of AP ALint maximally consistent theories, as
further explained below.

The definition of the canonical model for AP ALint is the

same as for ELint , except that the maximally consistent
sets are maximally consistent theories. We now come to
the Truth Lemma for the logic AP ALint . Here we use the
complexity measure <S
d .

Definition 28. A set x of formulas is called a theory

iff AP ALint x and x is closed under (DR1) and (DR5).

Lemma 36 (Truth Lemma). For every LAP ALint

and for each x X c , x iff X c , (x, ) |= .

Proof. Let LAP ALint and x X c . The proof is by

on , where the case = [] is proved by a
subinduction on . We therefore consider 14 cases.
Case := p

6.4 and Lemma 6.5). Therefore, we obtain

<S
d -induction

xp

iff x c (p)
iff (x, ) |= p

Lemma 35
(IH)
semantics

Theorem 37. AP ALint is complete with respect to the

class of all topo-models.

Induction Hypothesis (IH): For all formulas LAP ALint ,

c

if <S
d , then x iff X , (x, ) |= .
The cases negation, conjunction, and interior modality are
as in Truth Lemma 25 for ELint , where we observe that the
subformula order is subsumed in the <S
d order (see Lemma
6.2). We proceed with the knowledge operator, i.e., case
:= Ki , and then with the subinduction on for case
announcement := [], and finally with the case := 2.
Case := Ki
This case is also similar to the one in Truth Lemma 25 for
ELint , however, using maximally consistent theories in the
canonical model creates some differences. For the direction
from left-to-right, see Truth Lemma 25. For (), suppose
Ki 6 x. Then, by Lemma 34, there exists a maximally
consistent theory y such that Ki x y and 6 y. By <S
d
Ki and (IH), (y, ) 6|= . Since Ki x y, we have y [x]i
meaning that y (x)(i). Therefore, by the semantics,
X c , (x, ) 6|= Ki .
Case := []p
[]p x iff int() p x
iff int()
6 x or p x
iff (x, ) 6|= int() or (x, ) |= p
iff (x, ) |= []p

(R1)
Prop. 29
()
(R1)

S
(*): By (IH), int() <S
d []p and p <d []p (Lemma 6.3 and
Lemma 6.2).
Case := [] Use (R2) and (IH) and, by Lemma 6.3
S
and Lemma 7.1, int() <S
d [] and [] <d [].
Case := []( ) Use (R3) and (IH), [] <S
d [](
) and [] <S
d []( ).
Case := []int() Use (R4) and (IH) and, by Lemmas
S
6.3, 7.2, int() <S
d []int() and int([]) <d []int().
Case := []Ki Use (R5) and (IH) and, by Lemmas
S
6.3, 7.3, int() <S
d []Ki and Ki [] <d []Ki .
Case := [][] Use (R6) and (IH) and, by Lemma
7.4, [[]int()] <S
d [][].
Case := []2 For all LP ALint , [][] <S
d []2,
as []2 has one more 2 than [][] . Therefore, it suffices
to show []2 x iff LP ALint , [][] x.
() Consider the necessity form []] and assume that
for all LP ALint , [][] x, i.e., for all LP ALint ,
[]]([]) x . As x is closed under (DR5), we obtain
[]](2) x, i.e., []2 x.
() Suppose []2 x. We have

` 2 [], for all LP ALint

` [](2 []) for all LP ALint
` []2 [][], for all LP ALint

2 x iff ( LP ALint )([] x)

iff ( LP ALint )(x, ) |= []
iff (x, ) |= 2

(R7)
(DR4)
(DR1), (R1-R3)

Therefore, for all LP ALint , [][] x. As [][] <S

d
[]2 for all LP ALint , by (IH), we have for all
LP ALint , (x, ) |= [][]. Then, by the semantics, we obtain (details omitted) that (x, ) |= []2.
Case := 2 Again note that for all LP ALint ,
[] <S
d 2, as 2 has one more 2 than [] (see Lemma

Proof. Let LAP ALint such that 6` , i.e., 6 AP ALint

(Recall that AP ALint is the smallest theory). Then, by
Lemma 32, AP ALint + is a consistent theory and, by
Lemma 31, AP ALint + . By Lemma 33, the consistent theory AP ALint + can be extended to a maximally
consistent theory y such that AP ALint + y. Since
y is maximally consistent and y, we obtain 6 y
(by Proposition 29). Then, by Lemma 36 (Truth Lemma),
X c , (y, ) 6|= .

5.

COMPARISON TO OTHER WORK

Multi-agent epistemic systems with subset space-like semantics have been proposed in [11, 12, 4, 18], however, none
of these are concerned with arbitrary announcements. Our
goal in this paper is not to provide a multi-agent generalization of SSL per se, but to work with the effort-like modality 2 intended to capture the information change brought
about by any announcements (subject to some restrictions)
in a multi-agent setting and modelling it by way of openset shrinking similar to the effort modality, rather than by
deleting states or neighbourhoods, so that the intuitive link
between the two becomes more transparent on a semantic
level. In [3], Balbiani et al. proposed subset space semantics for arbitrary announcements, however, their approach
does not go beyond the single-agent case and the semantics
provided is in terms of model restriction. An unorthodox
approach to multi-agent knowledge is proposed in [11, 12].
Roughly speaking, instead of having a knowledge modality
Ki for each agent in his syntax, Heinemann uses additional
operators to define Ki and his semantics only validate the
S4-axioms for Ki . The necessitation rule for Ki does not
preserve validity under the proposed semantics [11, 12]. In
[18] a multi-agent semantics for knowledge is provided, but
no announcements or further generalizations (unlike in their
other, single-agent, work [19]), and not in a topological setting. Their use of partitions for each agent instead of a
single neighbourhood is compatible with our requirement
that all neighbourhoods for a given agent be disjoint. A further difference from the existing literature is that we restrict
our attention to topological spaces and prove our results by
means of topological tools.
We applied the new completeness proof for arbitrary public announcement logic of [2] to a topological setting. The
canonical modal construction is as in [6] with some multiagent modifications. The modality int in our system demands a different complexity measure in the Truth Lemma
of the completeness proof than in [2].

6.

CONCLUSIONS

We have proposed topological semantics for the multiagent extensions of the public announcement logic of [6], and
further extended the logic with arbitrary announcements.
We showed topological completeness of these logics. Our
work can be seen as a step toward discovering the interplay
between dynamic epistemic logic and topological reasoning.

For further research, we envisage a finitary axiomatization for AP ALint wherein the infinitary derivation rule
(DR5) is replaced by a finitary rule. The obvious derivation
rule would derive something after any announcement if it
can be derived after announcing a fresh variable [1]. Under
subset space semantics, it is unclear how to prove that this
rule is sound.
We are still investigating expressivity and (un)decidability.
If the logic AP ALint is undecidable, this would contrast
nicely with the undecidability of arbitrary public announcement logic. Otherwise, there may be interesting decidable
versions when restricting the class of models to particular
topologies.
The logic AP ALint is also axiomatizable on the class where
the K modalities have S4 properties, a result we have not
reported in this paper for consistency of presentation. This
class is of topological interest.
In our setup all agents have the same observational powers. If agents can have different observational powers, we
can associate a topology with each agent and generalize the
logic to an arbitrary epistemic action logic.
Furthermore, we would like to explore the exact difference
between the effort modality and the arbitrary announcement
modality (in the single agent case, see [16]) by constructing a
topological model which distinguishes the two: a topological
model might have more than epistemically definable opens
with respect to the proposed semantics.

Acknowledgements
We thank Philippe Balbiani for various detailed suggestions
over the past year on how to improve our single-agent and
multi-agent results in subset space logic and topological logics. We have found him very supportive of our efforts. We
also thank the TARK reviewers for their valuable comments.
Hans van Ditmarsch is also affiliated to IMSc (Institute of
Mathematical Sciences), Chennai, as research associate. We
acknowledge support from European Research Council grant
EPS 313360.

7.

REFERENCES

[1] Balbiani, P., Baltag, A., van Ditmarsch, H., Herzig,

A., Hoshi, T., and de Lima, T. (2008) Knowable as
known after an announcement. The Review of
Symbolic Logic, 1, 305334.
[2] Balbiani, P. and van Ditmarsch, H. (2015) A simple
proof of the completeness of APAL. Studies in Logic,
8 (1), 6578.
[3] Balbiani, P., van Ditmarsch, H., and Kudinov, A.
(2013) Subset space logic with arbitrary
announcements. Proc. of the 5th ICLA, pp. 233244,
Springer.
[4] Baskent, C. (2007) Topics in Subset Space Logic.
Masters thesis, University of Amsterdam.
[5] Baskent, C. (2012) Public announcement logic in
geometric frameworks. Fundam. Inform., 118,
207223.
[6] Bjorndahl, A. (2013) Subset space public
announcement logic revisited. CoRR, abs/1302.4009.
[7] Chellas, B. F. (1980) Modal logic. Cambridge
University Press, Cambridge.
[8] Dabrowski, A., Moss, L. S., and Parikh, R. (1996)
Topological reasoning and the logic of knowledge.

Ann. Pure Appl. Logic, 78, 73110.

[9] Dugundji, J. (1965) Topology. Allyn and Bacon Series
in Advanced Mathematics, Prentice Hall.
[10] Goldblatt, R. (1982) Axiomatising the Logic of
Computer Programming. Springer-Verlag.
[11] Heinemann, B. (2008) Topology and knowledge of
multiple agents. Proc. of the 11th IBERAMIA, pp.
110, Springer.
[12] Heinemann, B. (2010) Logics for multi-subset spaces.
Journal of Applied Non-Classical Logics, 20, 219240.
[13] Moss, L. S. and Parikh, R. (1992) Topological
reasoning and the logic of knowledge. Proc. of the 4th
TARK , pp. 95105, Morgan Kaufmann.
[14] Parikh, R., Moss, L., and Steinsvold, C. (2007)
Topology and epistemic logic. Handbook of Spatial
Logics, pp. 299341.
[15] Plaza, J. A. (1989) Logics of public communications.
Proc. of the 4th ISMIS , pp. 201216, Oak Ridge
National Laboratory.
un, A. (2014)
[16] van Ditmarsch, H., Knight, S., and Ozg
Arbitrary announcements on topological subset spaces.
Proc. of the 12th EUMAS , pp. 252266, Springer.
[17] van Ditmarsch, H., van der Hoek, W., and Kooi, B.
(2007) Dynamic Epistemic Logic, vol. 337 of Synthese
Library. Springer.
[18] Wang, Y. N. and
Agotnes, T. (2013) Multi-agent
subset space logic. Proc. of the 23rd IJCAI , pp.
11551161, IJCAI/AAAI.
[19] W
ang, Y. N. and
Agotnes, T. (2013) Subset space
public announcement logic. Proc. of 5th ICLA, pp.
245257, Springer.