Radu Curticapean Holger Dell Dániel Marx NEW INSIGHTS INTO Saarland University Cluster of Excellence MMCI Institute for Computer Science and Control Hungarian Academy of ID: 544158
Download Presentation The PPT/PDF document "COUNTING SUBGRAPHS" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
COUNTING SUBGRAPHS
Radu Curticapean, Holger Dell, Dániel Marx
NEW INSIGHTS INTO
Saarland University,Cluster of Excellence (MMCI)
Institute for Computer Science and Control,Hungarian Academy of Sciences (MTA SZTAKI)Slide2
vertices
pattern
vertices
h
ost
count
subgraphs
F
of
G
isomorphic
to H
NOT NECESSARILY INDUCED
COUNTING SUBGRAPHSSlide3
-clique
-
cycle
-
matching
-star
with
vtx
-cover
of
size
any
[NP 85]
[BKK 14]
[BKK 14]
#W[1]-hard
refutes ETH
[CCFHJKX 04]
#W[1]-hard
[FG 04]
#W[1]-hard,
refutes ETH
vertices
pattern
vertices
h
ost
count
subgraphs
F
of
G
isomorphic
to
H
[BHKK 09]
[C 13, CM 14, CM 15]
ETH:
SAT on
-variable formulas
has no
time algorithm
parameterized problem X:
input comes with some parameter
X is FPT:
problem can be solved in time
is #W[1]-hard:
FPT-algorithm for #
-clique with
-oracle
and
all queries have parameter
for some
X most likely not FPT
Slide4
vertices
pattern
vertices
h
ost
count
subgraphs
F
of
G
isomorphic
to H
poly-time
#W[1]-hard.
with vtx
-coverof
size
-star
-clique
-
cycle
-
matching
Slide5
poly-time
#W[1]-hard
.
Input:
Graphs
and
Parameter:
Output:
maximum matching in
bounded
for
recursively
enumerable
unbounded
with
vtx
-cover
of
size
-star
[CM 14]
-clique
-
cycle
-
matching
classifies
w
is
poly
-time
assuming
despite
-intermediate cases!
[CTW 08]
Slide6
poly-time
-
hard
finite
count
/find
induced
subgraphs
from
else
[CTW 08]
count/find
colorful subgraphs
from
has bounded
treewidth
else
[GSS 01][CM 14]
count
homomorphisms from
has bounded
treewidth
else
[G 07]
[DJ 04]
count
subgraphs from
has bounded
matchings
else
[CM 14]
tree-decomposition of H:
tree
with bags
at
Each
is in some bag
Each
is in some bag
If
appears in bags of
,
then
is connected in
width of
: max. bag size
treewidth of
: min. width over all
David Eppstein, Wikipedia
Slide7
COUNTING
HOMOMORPHISMS
poly-time
-
hard
count
homomorphisms
from
has bounded
treewidth
else
[G 07]
[DJ 04]
time
via standard D
Slide8
COUNTING
SUBGRAPH EMBEDDINGS
COUNTING
HOMOMORPHISMSSlide9
COUNTING
HOMOMORPHISMS
COUNTING
INJECTIVE HOMOMORPHISMSSlide10
COUNTING
HOMOMORPHISMS
COUNTING
INJECTIVE HOMOMORPHISMS
COUNTING LINEAR
COMBINATIONS OF
COUNTING LINEAR
COMBINATIONS OF
Slide11
1
2
3
4
Define
for
partition
of
Contract every block of
in
.
Delete multi-
edges
,
keep
self
-loops.
12
3
4
12|3|4
34|1|2
341|2
1234
1|2|3|4
13|2|4
14|2|3
23|1|4
24|1|3
12|34
13|24
14|23
23|14
24|13
34|12
123|4
142|3
234|1
LINEAR COMBINATIONS OF
HOMOMORPHISMS
COUNTING
INJECTIVE HOMOMORPHISMS
Spasm
1
2
34
2
341
1
2
3
4Slide12
1|2|3|4
12|3|4
13|2|4
14|2|3
23|1|4
24|1|3
34|1|2
12|34
13|24
14|23
23|14
24|13
34|12
123|4
142|3
234|1
341|2
1234
1
2
3
4
LINEAR COMBINATIONS OF
HOMOMORPHISMS
COUNTING
INJECTIVE HOMOMORPHISMS
12
3
4
1
2
34
2
341
1
2
3
4Slide13
First
application
:
Counting
any
-
edge
subgraph
in time
best
known
upper
bound
on
treewidth
of
-
edge
graph
-
paths
(
before
: time
)
-
matchings
(
before
: time
)
Lovász
1967Slide14
Scott, Sorkin 2016
:
Graphs
with
edges have
time
poly
time
graphs,
each
with
edges
Slide15
Second application:
Hardness of counting subgraphs
via fine-grained reductions
Slide16
#
PartitionedSub
(
Pick
any
.
Given
oracle
for
this
,
we
can
compute
this
in
steps.
Slide17
#
PartitionedSub
(
under #ETH
[Marx 2010]
even when fixing
and allowing only
under #ETH
even when fixing
and allowing only
Pick
any
.
Given
oracle
for
this
,
we
can
compute
this
in
steps.
has
-
matching
pick expander
with
Slide18
always in time
.
Exact value of
for
fixed ?
exponent in running time
for #PartitionedSub
exponent in running time
for computing #H
Can
you
beat
treewidth
:
Clique
conjecture
:
if
Lovász
1967