x | = | statement represented by x is true (now, in the current situation) |
□x | = | the statement x is true in a set of different scenarios
what a "scenario" is depends on the application:
|
recursively defined from:
the modal operators □ and ◇ are applied to a single
formula,
which may contain ¬, ∧ and ∨ but also
other ◇ and □
examples:
usual recursive definition of syntax
(variable is a formula, if A and B are two formulae then
A ∧ B is a formula, etc.)
unary operators ¬, □ and ◇ have the same precedence
they take precedence over the binary connectives ∧ and ∨
therefore, □x ∨ y is:
◇ is complementary to □
depending on the application, ◇x therefore may mean:
it holds ◇F=¬□¬F
□F means that formula F is true in various situations
cannot be evaluated with a single interpretation
(an interpretation represents one situation)
evaluating a formula requires not just an interpretation, but a set of interpretations
semantics is complicated
varies with the application
(others exist)
evaluating a formula requires:
in both cases:
in order to evaluate a modal formula, a set of interpretations is required
they differ because:
interpretations are given as a set (S5), or as a sequence (TL)
Kripke model: a generalization including both, but also many other ones
the set of possible worlds is a set of elements
is not a set of propositional interpretations
why not?
worlds are just elements of a set…
… but then a propositional interpretation is associated to each world
why not defining a model to be: a set of interpretation plus their links?
for S5, the simplified definition would work:
no point in having two worlds with the same interpretation
in TL, it does not:
two different time points may have the same interpretation
(example on the next slide)
two different time points may have the same interpretation
example: the sequence of states:
ab, ¬ab, a¬b, ab, ab, ab, … (ab forever)
interpretation ab recurs in different time points (worlds)
□a is false in the first state, but true in the fourth
state one and state four have the same interpretation,
but
a modal formula evaluates differently on them
〈R,S,V〉
what is the relation?
how is it used?
intuition: □F means that F is true in a set of worlds
both cases can be encoded using the right accessibility relation
the point: the relation tells the worlds to look when evaluating □F (cont.)
formulae are evaluated in a model and one of its worlds
notation: 〈R,S,V〉,s ⊧ F
(diamond: later)
formulae are evaluated recursively in the same state except for boxed formulae
for boxed formulae, evaluate subformula in all linked states
in all worlds linked from s formula F is true ⇒ □F is true in s
truth in s:
truth in s':
a formula is true in a model if it is true in all states of the model
formally: 〈R,S,V〉 ⊧ F if 〈R,S,V〉,s ⊧ F for all worlds s∈S
◇F is true in s if F is true in a world s' such that 〈s,s'〉∈R
F is true in one world reachable from s ⇒ ◇F is true in s
one world is enough!
not all models are right for all kinds of applications:
the same formula may be true in all models of a logic but not of another
example: ◇F → □◇F
◇F → □◇F always true: proof
(for simplicity, we use lines in place of arrows;
can be done because the relation is symmetric)
(loops omitted)
◇F is true in a world s
⇓ there exists s' where F is true |
for every other s'',
it holds 〈s'',s'〉∈R ⇓ ◇F is true in all worlds |
◇F is true in all words;
as a result, □◇F is true as well (whichever world we choose, ◇F is true in all linked ones) |
◇F → □◇F false in TL: proof
counterexample:
loops and arcs that follow from transitivity omitted
(shown only when necessary)
F is true in the third state
reachable from initial state
⇓
◇F true in the initial state
F false from fourth state on
⇓
◇F false in the fourth state
◇F false in the fourth state
reachable from the initial state
⇓
□◇F false in the initial state
conclusion: ◇F → □◇F false
◇F → □◇F is:
validity depends on the assumptions on the relation R
(which vary with the logic: S5 or TL, in the example)
for simplicity, 〈s,t〉 ∈ R is denoted as sRt
the accessibility relation may be:
some formulae are true in all models only if some assumptions hold
some formulae are always true only if the relation is reflexive
since □F is true in the state, F is true in all its successors
but the state is one of the successors of itself, therefore F is true there
in the left state, □F is true
because F is true in the only successor
since F is true, in every predecessor ◇F is true
but the predecessors are also the successors
|
if □F is true in s,
and every state has a successor t,
then F is true in t;
therefore, F is true in at least a successor of s
does not hold on states having no successors:
s has no successors
⇒ but it has no successors
⇒ |
conclusion:
some formulae are always true only if the relation satisfies certain conditions
every modal logic (S5, TL, …) is specified by the assumptions on the accessibility relation
a formula is true in a logic if it is true in all models whose relation satisfies the corresponding assumptions (e.g., reflexivity, transitivity, …) of that logic
K is the modal logic with no assumption on the relation
formula F is satisfiable if there exist 〈R,S,V〉 and s such that 〈R,S,V〉,s ⊧ F
where:
by the second point, satisfiability depends on the considered modal logic