Moving Arrows and Model Checking Results with Dynamic Operators

Moving Arrows and Four Model Checking Results
Carlos Areces
1,2
, Raul Fervari
1
& Guillaume Homann
1
1
FaMAF, Universidad Nacional de Cordoba, Argentina,
2
CONICET, Argentina
WoLLIC 2012, Buenos Aires, Argentina
C. Areces, R. Fervari & G. Homann: Moving Arrows and Four Model Checking Results WoLLIC 2012 1/19
Modal logics: we like to talk about models
Modal logics are known to describe models.
Choose the right paintbrush:
. . .
Now, what about operators that can modify models?
Change the domain of the model.
Change the properties of the elements of the domain while we are

evaluating a formula.
Evaluate after deleting/adding/swapping around an edge.

Logics that change the model 1/2
What about a swapping modal operator?
w
sw
v w v
What happens when you add that to the basic modal logic?
What about:
an edge-deleting modality?
an edge-adding modality?
What about:
an edge-deleting modality?
an edge-adding modality?
Sabotage Modal Logic [van Benthem 2002]
/, w [= gs i pair (u, v) of / such that /
{(u,v)}
, w [= ,
where /
{(u,v)}
is / without the edge (u, v).
Note: (u, v) can be anywhere in the model.
What we know [Loding & Rohde 03]:
Model checking is PSPACE-complete.
Satisability is undecidable.
Sabotage Modal Logic [van Benthem 2002]
/, w [= gs i pair (u, v) of / such that /
{(u,v)}
, w [= ,
where /
{(u,v)}
is / without the edge (u, v).
Note: (u, v) can be anywhere in the model.
What we know [Loding & Rohde 03]:
Model checking is PSPACE-complete.
Satisability is undecidable.
Epistemic Operators
Those are operators that also modify models!
[!]: announce that if is true, eliminate states of the model where

holds (Public Announcement Logic) [Plaza 89].
: there is a -free announcement such that [!] holds

(Arbitrary Public Announcement Logic) [Balbiani et al. 07].
In some way these operators are deleting states.
We will focus on operators that modify the accesibility relation.

Epistemic Operators



Epistemic Operators



Epistemic Operators



Epistemic Operators



Meet the new operators
Remember the Basic Modal Logic (B/L).
Syntax: propositional language + a modal operator .
Semantics of : traverse some edge, then evaluate .

Now add new dynamic operators:
Semantics of swap, global/local sabotage and bridge:
sw: traverse some edge, turn it around, then evaluate .
gs: delete some edge anywhere, then evaluate .
ls: traverse some edge, delete it, then evaluate .
br : add a new edge, traverse it, then evaluate .















Examples: no tree model property
Theorem
/L() lacks the tree model property, for sw, gs, ls, br .
Proof.
1. br w and v ,= w are unconnected.
2. [gs] w is reexive.
3. [ls] w is reexive.
4. p (
1i 3
i
p) swp w has a reexive successor.
Examples: no tree model property
Theorem
/L() lacks the tree model property, for sw, gs, ls, br .
Proof.
1. br w and v ,= w are unconnected.
2. [gs] w is reexive.
3. [ls] w is reexive.
4. p (
1i 3
i
p) swp w has a reexive successor.
Bisimulations
We want to learn more about the models that these logics can describe.
So we need:
Denition of -bisimilarity.
A bisimilarity theorem that says that two -bisimilar models are

undistinguishable by /L().
Conditions for -bisimulations 1/2
always (nontriv) Z is not empty
always (agree) If (w, S)Z(w
, S
), w and w
agree propositionally.
(zig) If wSv, there is v
s.t. w
and (v, S)Z(v
, S
)
(zag) If w
, there is vW s.t. wSv and (v, S)Z(v
, S
)
sw (sw-zig) If wSv, there is v
s.t. w
and (v, S
vw
)Z(v
, S
w
)
(sw-zag) If w
, there is vW s.t. wSv and (v, S
vw
)Z(v
, S
w
)
Conditions for -bisimulations 2/2
gs (gs-zig) If vSu, there is v
, u
s.t. v
and (w, S
vu
)Z(w
, S
)
(gs-zag) If v
, there is v, uW s.t. vSu and (w, S
vu
)Z(w
, S
)
ls (ls-zig) If wSv, there is v
s.t. w
and (v, S
wv
)Z(v
, S
)
(ls-zag) If w
wv
)Z(v
, S
)
br (br-zig) If wSv, there is v
s.t. w
and (v, S
+
wv
)Z(v
, S
+
w
)
(br-zag) If w

+
wv
)Z(v
, S
+
w
)
Invariance for Dynamic Logics
Theorem
For /L(), sw, gs, ls, br , /, w

ML()
/
, w
implies
/, w
ML()
/
, w
.
Comparing expressiveness
What if we want to show that all of these logics are uncomparable?
Find two
1
-bisimilar models distinguishable by /L(
2
).
Find two
2
-bisimilar models distinguishable by /L(
1
).
Then /L(
1
) and /L(
2
) are uncomparable.
Now lets have fun!
/ /
Distinct by Bisimilar for

w
w
br br
gs
/L(ls)
/L(sw)
w
w
ls
gs
/L(sw)
/L(br )
w
w
swsw
[br ][br ]
/L(gs)
/L(ls)
w
. . . . . .
w
. . .
sw /L(br )
w
. . .
. . .
w
. . .
. . .
ls /L(gs)
It all boils down to that. . .
Theorem
For all
1
,
2
sw, gs, ls, br with
1
,=
2
, /L(
1
) and
/L(
2
) are uncomparable.
Other results: Model checking modal logics
It is well known that model checking B/L is only polynomial.
But, what happens with dynamic operators?
Model checking PAL is PSPACE-complete [Balbiani et al. 07].
For global sabotage is PSPACE-complete [Loding & Rohde 03].
Let us prove PSPACE-completeness for local sabotage, bridge and

swap logic.

swap logic.

swap logic.

swap logic.

swap logic.
Model checking /L(sw) is PSPACE-hard
For a Quantied Boolean Formula with k variables:
1. Build /
k
as:
p
1
p
p
1
. . .
p
k
p
p
k
2. Build a /L(sw) formula from a QBF as follows:
(x
i
.)
= sw(p
i
()
)
(x
i
)
= (p
i
p
)
()
= ()
( ) = ()
()
3. is true i /
k
, w [= ()
1. Build /
k
as:
p
1
p
p
1
. . .
p
k
p
p
k
(x
i
.)
= sw(p
i
()
)
(x
i
)
= (p
i
p
)
()
= ()
( ) = ()
()
3. is true i /
k
, w [= ()
1. Build /
k
as:
p
1
p
p
1
. . .
p
k
p
p
k
(x
i
.)
= sw(p
i
()
)
(x
i
)
= (p
i
p
)
()
= ()
( ) = ()
()
3. is true i /
k
, w [= ()
1. Build /
k
as:
p
1
p
p
1
. . .
p
k
p
p
k
(x
i
.)
= sw(p
i
()
)
(x
i
)
= (p
i
p
)
()
= ()
( ) = ()
()
3. is true i /
k
, w [= ()
Model checking is PSPACE-complete
We have similar translations for /L(gs), /L(ls) and /L(br ).
Being in PSPACE is shown with a depth-rst algorithm that follows the
denition of [=.
Theorem
For sw, gs, ls, br , model checking for any of the logics
/L() is PSPACE-complete.
denition of [=.
Theorem
denition of [=.
Theorem
Conclusions
These logics have similar features to Sabotage Modal Logic:
Lack of tree model property.
PSPACE-complete model checking problem.
They are all uncomparable in expressivity.
Decidability of the satisability problem?
We have a proof that /L(sw) is undecidable.
We dont know yet about /L(ls) and /L(br ).
Further step: axiomatizations.

Conclusions

Conclusions

Conclusions

Conclusions

Conclusions

Conclusions

Conclusions


Moving Arrows and Model Checking Results with Dynamic Operators

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Moving Arrows and Model Checking Results with Dynamic Operators

Uploaded by

Copyright:

Available Formats

Moving Arrows and Four Model Checking Results

Modal logics are known to describe models.

Choose the right paintbrush:

Now, what about operators that can modify models?

Change the domain of the model.

Change the properties of the elements of the domain while we are

Evaluate after deleting/adding/swapping around an edge.

Model checking is PSPACE-complete.

Model checking is PSPACE-complete.

Those are operators that also modify models!

[!]: announce that if is true, eliminate states of the model where

: there is a -free announcement such that [!] holds

In some way these operators are deleting states.

We will focus on operators that modify the accesibility relation.

Those are operators that also modify models!

[!]: announce that if is true, eliminate states of the model where

: there is a -free announcement such that [!] holds

In some way these operators are deleting states.

We will focus on operators that modify the accesibility relation.

Those are operators that also modify models!

[!]: announce that if is true, eliminate states of the model where

: there is a -free announcement such that [!] holds

In some way these operators are deleting states.

We will focus on operators that modify the accesibility relation.

Those are operators that also modify models!

[!]: announce that if is true, eliminate states of the model where

: there is a -free announcement such that [!] holds

In some way these operators are deleting states.

We will focus on operators that modify the accesibility relation.

Those are operators that also modify models!

[!]: announce that if is true, eliminate states of the model where

: there is a -free announcement such that [!] holds

In some way these operators are deleting states.

We will focus on operators that modify the accesibility relation.

Syntax: propositional language + a modal operator .

Semantics of : traverse some edge, then evaluate .

Semantics of swap, global/local sabotage and bridge:

sw: traverse some edge, turn it around, then evaluate .

gs: delete some edge anywhere, then evaluate .

ls: traverse some edge, delete it, then evaluate .

br : add a new edge, traverse it, then evaluate .

Syntax: propositional language + a modal operator .

Semantics of : traverse some edge, then evaluate .

Semantics of swap, global/local sabotage and bridge:

sw: traverse some edge, turn it around, then evaluate .

gs: delete some edge anywhere, then evaluate .

ls: traverse some edge, delete it, then evaluate .

br : add a new edge, traverse it, then evaluate .

Syntax: propositional language + a modal operator .

Semantics of : traverse some edge, then evaluate .

Semantics of swap, global/local sabotage and bridge:

sw: traverse some edge, turn it around, then evaluate .

gs: delete some edge anywhere, then evaluate .

ls: traverse some edge, delete it, then evaluate .

br : add a new edge, traverse it, then evaluate .

Syntax: propositional language + a modal operator .

Semantics of : traverse some edge, then evaluate .

Semantics of swap, global/local sabotage and bridge:

sw: traverse some edge, turn it around, then evaluate .

gs: delete some edge anywhere, then evaluate .

ls: traverse some edge, delete it, then evaluate .

br : add a new edge, traverse it, then evaluate .

Syntax: propositional language + a modal operator .