Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

On Local Domain Symmetry for Model

Expansion
Jo Devriendt, Bart Bogaerts,
Maurice Bruynooghe, Marc Denecker
University of Leuven / Aalto University

October 21, 2016

1 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Intro
General symmetry definition
Given a vocabulary , theory T , and domain D, a symmetry
for T is a permutation on the set of D,-structures D such
that for all I D :
I |= T iff (I ) |= T

2 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Intro
General symmetry definition
Given a vocabulary , theory T , and domain D, a symmetry
for T is a permutation on the set of D,-structures D such
that for all I D :
I |= T iff (I ) |= T
Why study symmetry?
speeding up search symmetry breaking
avoid parts of the search space symmetrical to failed parts

3 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Outline
Intro
Theory symmetry
MX symmetry
Efficient breaking
More symmetry
Conclusion

4 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Prelims: First-Order Logic (FO)

vocabulary of (function and predicate) symbols S/k
-theory T
-structure I
domain D
interpretations S I for all S

Semantics captured by satisfiability relation:

I |= T

5 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Prelims: First-Order Logic (FO)

vocabulary of (function and predicate) symbols S/k
-theory T
-structure I
domain D
interpretations S I for all S

Semantics captured by satisfiability relation:

I |= T
In ASP: program theory, set of facts structure.

6 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Prelims: First-Order Logic (FO)

vocabulary of (function and predicate) symbols S/k
-theory T
-structure I
domain D
interpretations S I for all S

Semantics captured by satisfiability relation:

I |= T
In ASP: program theory, set of facts structure.
For the rest of the talk: vocabulary and domain are implicit
and fixed.
7 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Running example: graph coloring

Tgc :

x1 y1 : Edge(x1 , y1 ) Color (x1 ) 6= Color (y1 )

x2 y2 : Edge(x2 , y2 ) V (x2 ) V (y2 )
x3 : C (Color (x3 ))

Igc :
V Igc = {t, u, v , w } C Igc = {r , g , b}
Edge Igc = {(t, u), (u, v ), (v , w ), (w , t)}
Color Igc = t 7 r , u 7 g , v 7 b, w 7 g , r 7 r , g 7 g , b 7 b
u

Note: Igc |= Tgc

8 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Symmetry for a theory

General symmetry definition

Given a vocabulary , theory T , and domain D, a symmetry
for T is a permutation on the set of D,-structures D such
that for all I D :
I |= T iff (I ) |= T
Symmetries compose to symmetry groups

9 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Global Domain Symmetry

Permutation on D induces permutation on D :

((d1 ), . . . , (dn )) P (I ) iff (d1 , . . . , dn ) P I
f (I ) ((d1 ), . . . , (dn )) = (d0 ) iff f I (d1 , . . . , dn ) = d0
Lets call such induced a Global Domain Symmetry for T .
Intuitively, domain renaming preserves satisfiability:
(I ) |= T iff I |= T

10 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Let = (v r ), so maps
u

dr

dv

to

which still models Tgc .

11 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Connectively closed argument positions

More precise notion of domain symmetry:
apply only on limited set of argument positions A.
Argument position S|n with S/k and 1 n k
denotes Ss nth argument. f |0 is output argument
position.

12 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Connectively closed argument positions

More precise notion of domain symmetry:
apply only on limited set of argument positions A.
Argument position S|n with S/k and 1 n k
denotes Ss nth argument. f |0 is output argument
position.
Argument positions are connected under theory T if one
occurs as subterm of the other, if they are connected by
=, or if they are connected by quantified variables.

13 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Connectively closed argument positions

More precise notion of domain symmetry:
apply only on limited set of argument positions A.
Argument position S|n with S/k and 1 n k
denotes Ss nth argument. f |0 is output argument
position.
Argument positions are connected under theory T if one
occurs as subterm of the other, if they are connected by
=, or if they are connected by quantified variables.
A set of argument positions A is connectively closed
under T if no other argument positions of are
connected to A under T .
Intuitively, a partition of connectively closed argument
positions under T corresponds to a well-defined typing of T .
14 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Tgc :
x1 y1 : Edge(x1 , y1 ) Color (x1 ) 6= Color (y1 )
x2 y2 : Edge(x2 , y2 ) V (x2 ) V (y2 )
x3 : C (Color (x3 ))

15 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Tgc :
x1 y1 : Edge(x1 , y1 ) Color (x1 ) 6= Color (y1 )
x2 y2 : Edge(x2 , y2 ) V (x2 ) V (y2 )
x3 : C (Color (x3 ))
Connectively closed argument position partition under Tgc :
{V |1, Edge|1, Edge|2, Color |1}
{C |1, Color |0}

16 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Local Domain Symmetry

Permutation on D and argument position set A induce
permutation A on D :
A

( A (d1 ), . . . , A (dn )) P (I ) iff (d1 , . . . , dn ) P I

A

f (I ) ( A (d1 ), . . . , A (dn )) = A (d0 ) iff f I (d1 , . . . , dn ) = d0

where A applies only on argument positions in A.

17 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Local Domain Symmetry

Permutation on D and argument position set A induce
permutation A on D :
A

( A (d1 ), . . . , A (dn )) P (I ) iff (d1 , . . . , dn ) P I

A

f (I ) ( A (d1 ), . . . , A (dn )) = A (d0 ) iff f I (d1 , . . . , dn ) = d0

where A applies only on argument positions in A.
If A is connectively closed under T , A is a local domain
symmetry of T .
Intuitively, A permutes domain element tuples according to
some type A of T .
18 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Let = (v r ) and A = {V |1, Edge|1, Edge|2, Color |1}, so A
maps
u

dr

to

which still models Tgc .

19 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Local Domain Symmetry

So far, so good: more or less known in literature (MACE,

SEM, Paradox, ...)

20 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Local Domain Symmetry

So far, so good: more or less known in literature (MACE,

SEM, Paradox, ...)

What if we have to take pre-interpreted symbols into
account? Model eXpansion

21 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

First-Order Model Expansion (MX)

In:
vocabulary = in out (in out = )
-theory T
in -structure Iin
domain D
interpretations S Iin to S in

Out:
out -structure Iout such that Iin t Iout |= T
same domain D
Iin t Iout merges both structures to a -structure
or unsat

22 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

First-Order Model Expansion (MX)

In:
vocabulary = in out (in out = )
-theory T
in -structure Iin
domain D
interpretations S Iin to S in

Out:
out -structure Iout such that Iin t Iout |= T
same domain D
Iin t Iout merges both structures to a -structure
or unsat

Shortened as MX (T, Iin ).

23 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Tgc :

x1 y1 : Edge(x1 , y1 ) Color (x1 ) 6= Color (y1 )

x2 y2 : Edge(x2 , y2 ) V (x2 ) V (y2 )
x3 : C (Color (x3 ))

Igcin :
V Igcin = {t, u, v , w } C Igcin = {r , g , b}
Edge Igcin = {(t, u), (u, v ), (v , w ), (w , t)}
u

24 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Igcout :
Color Igcout = t 7 r , u 7 g , v 7 b, w 7 g , r 7 r , g 7 g , b 7 b
Igcin t Igcout :
u

Note: Igcin t Igcout |= Tgc

25 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Symmetry for MX

Symmetry for MX (T, Iin )

A symmetry for MX (T, Iin ) is a permutation on the set of
D,out -structures D such that for all I D :
Iin t Iout |= T iff Iin t (Iout ) |= T
Intuitively, symmetry must preserve input structure.

26 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Sufficient condition for MX-symmetry?

Local domain symmetry A where A is connectively

closed under T and such that A (Iin ) = Iin

27 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Sufficient condition for MX-symmetry?

Local domain symmetry A where A is connectively

closed under T and such that A (Iin ) = Iin

Does not work well for independent symbols connected
under T :
u

Both graphs Edge/2 and Edge 0 /2 have different symmetries,

but since their argument positions typically are connected by a
V /1 argument position, no exists that is consistent with the
symmetry of both graphs.
28 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Sufficient condition for MX-symmetry?

We defined transformation for MX (T, Iin ) to MX (T , Iin ) such
that

Sufficient condition for MX-symmetry

A is a local domain symmetry for MX (T, Iin ) if A is
connectively closed under T and A (Iin ) = Iin .
Intuitively, MX (T, Iin ) decouples all occurrences of
pre-interpreted symbols.

29 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Finding structure-preserving ?
given MX (T, Iin ), finding good A is easy
Partition argument positions in T in connectively closed
classes

30 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Finding structure-preserving ?
given MX (T, Iin ), finding good A is easy
Partition argument positions in T in connectively closed
classes
how about such that A (Iin ) = Iin ?

31 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Finding structure-preserving ?
given MX (T, Iin ), finding good A is easy
Partition argument positions in T in connectively closed
classes
how about such that A (Iin ) = Iin ?

Answer:
generate-and-test for domain element swaps (d1 d2 )

32 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Finding structure-preserving ?
given MX (T, Iin ), finding good A is easy
Partition argument positions in T in connectively closed
classes
how about such that A (Iin ) = Iin ?

Answer:
generate-and-test for domain element swaps (d1 d2 )
encode to graph automorphism problem for more
complicated
In: Iin , A
Out: generators that induce symmetry A
Size of graph depends on size of Iin , not on size of T

33 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

t

t.0

u.0

v .0

t.2

t.1

u.2

u.1

v .2

v .1

Edge1 (t, u)

Edge1 (u, v )

Edge1 (v , w )

w .0

w .2 w .1

Edge1 (w , t)

r .0

g .0

b.0

r .2

r .1

g .2

g .1

b.2

b.1

34 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Symmetry breaking
Given symmetry group G, construct symmetry breaking
formula (G).
(G) is sound if for each Iout , there exists some G such
that Iin t (Iout ) |= (G).
(G) is complete if for each Iout , there exists exactly one
G such that Iin t (Iout ) |= (G).

35 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Symmetry breaking
Given symmetry group G, construct symmetry breaking
formula (G).
(G) is sound if for each Iout , there exists some G such
that Iin t (Iout ) |= (G).
(G) is complete if for each Iout , there exists exactly one
G such that Iin t (Iout ) |= (G).
What is the size of (G) to break G completely?

36 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Breaking local domain symmetry completely

GA is a local domain interchangeability group if A is a set of
argument positions, D, and each permutation over
induces a local domain symmetry A .
GA is broken completely with O(||2 ) sized symmetry breaking
formula if A contains at most one argument position S|i for
each symbol S out .
Intuitively, after grounding, GA represents row
interchangeability of a Boolean variable matrix. Ordering the
rows breaks all symmetry.

37 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Let A = {C |1, Color |0}, = {r , b, g }. GA represents the

interchangeability of colors, and can be broken completely with
a small symmetry breaking formula.

38 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Graph coloring ctd.

Let A = {C |1, Color |0}, = {r , b, g }. GA represents the

interchangeability of colors, and can be broken completely with
a small symmetry breaking formula.
However, e.g. for the Ramsey number problem,
A = {Edge|1, Edge|2, Color |1, . . .}, so A0 does not satisfy the
one argument position condition.

39 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Symmetry notions not captured by local domain

symmetry
E.g. swapping colors of node v and t:
u

dv

dt

to

which still models Tgc .

40 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Conclusion
Notion of local domain symmetry
Sufficient condition for symmetry detection in the context

of an input structure
Symmetry detection approach on predicate level
Completeness guarantee for symmetry breaking
Limits of our approach
Notion can be extended to aggregates, non-monotonic

rules, etc.
Implementation in IDP

41 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Conclusion
Notion of local domain symmetry
Sufficient condition for symmetry detection in the context

of an input structure
Symmetry detection approach on predicate level
Completeness guarantee for symmetry breaking
Limits of our approach
Notion can be extended to aggregates, non-monotonic

rules, etc.
Implementation in IDP

Future work
Extend local domain symmetry to capture other notions of
symmetry.
42 / 43

Intro

Theory symmetry

MX symmetry

Efficient breaking

More symmetry

Conclusion

Thanks for your attention!

Questions?

43 / 43