Parameterized Complexity of Even Set (and others)
Dániel Marx
Hungarian Academy of Sciences Dagstuhl Seminar 19041
January 25, 2019
[1] Bingkai Lin, SODA 2015, JACM 2018 [2] Bonnet, Egri, Lin, M. , unpublished
[3] Bhattacharyya, Ghoshal, Kartik C.S., Manurangsi, ICALP 2018
L
BELM BGKM
BBEGKLMM
Hitting Set and Variants
MINIMUM EXACT
HITTING
UNIQUE HITTING ODD
EVEN
ODD/EVEN
{ SET
} } { }
{ × ×
All of them are known to be W[1]-hard, except MINIMUM EVEN SET (note: solution is required to be nonempty).
Dual form: SET COVER and friends.
EVEN SET definitions
• Hypergraphs (EVEN HITTING SET)
• Bipartite Graphs (RED-BLUE DOMINATING SET)
• Linear Algebra (MINIMUM DEPENDENT SET)
• Find a nonempty vector with Hamming weight at most such that .
• Matroid Theory (SHORTEST CIRCUIT):
• Find a circuit of size at most in a binary matroid
• Coding Theory (MINIMUM DISTANCE):
• Find a nonempty codeword of Hamming weight at most .
•
Coding theory
}
Generating matrix: �∈ �
2h �
Codewords:
Fact:
distance, systematic code, Sphere packing bound,…
W[1]-hardness of ODD SET
�1 �2 �3 �4
�1,2 �1,3 �1,4 �2,3 �2,4 �3,4
Odd Set ? Even Set
Gap Odd Set Gap Even Set
Odd Set ? Even Set
Odd Set ? Even Set
Gap Odd Set Gap Even Set
BICLIQUE
• Given a bipartite graph and integer , find two sets and of size that are fully adjacent to each other.
• Should have been on the Downey-Fellows list.
• Easy to give an incorrect proof (which works only for PARTITIONED BICLIQUE)
• Theorem: BICLIQUE is W[1]-hard
[Bingkai Lin, SODA’15, JACM’18]
•
Template graph
-threshold property for a bipartite graph on and :
• For any vertices in have at most common neighbors.
• For any choice of sets , we can find one vertex in each set such that these vertices have at least common neighbors.
•
There are randomized constructions for such graphs with and (deterministic construction is worse).
The reduction
clique
vertices with common neighbors
no clique
Any vertices has common
neighbors
Hardness of BICLIQUE
• We reduced -CLIQUE to -BICLIQUE.
• Theorem: BICLIQUE is W[1]-hard and no algorithm assuming randomized ETH.
• Approximation version:
ONE SIDED BICLIQUE
Find vertices maximizing the common intersection
• Hard to distinguish between optimum of and
•
Randomized construction
• Random graph on vertices with edge probability
• Claim 1: probability of vertices with common vertices is at most .
• Claim 2: probability of a given set of vertices do not have common neighbors is less than .
•
LINEAR DEPENDENT SET
• Given vectors , find vectors that are linearly dependent.
s.t. has Hamming weight at most
• Hard to approximate for any constant: proof by reduction from ONE SIDED BICLIQUE.
•
Encoding vertices
Vertex set
vectors in such that
• any are linearly independent
• any are linearly dependent
Vertex set B
vectors in such that
• any are linearly independent
• any are linearly dependent (Vandermonde matrices)
•
Reduction
edge
�(��) block
�(��) block
REALLY!
REALLY!
F(e)
Reduction
biclique
dependent sets linearlyno biclique
no linearly dependent sets (if is sufficiently large)COLORED LINEAR DEPENDENT SET
Vectors partitioned into color classes, distinguish between
• exists dependent set of vectors, one from each color,
• no dependent set of vectors (or arbitrary colors!) Simple reduction from the uncolored version using
Color Coding
(Turing or not)
•
MAXIMUM LIKELIHOOD DECODING
Given and , find of Hamming weight at most such that .
= ODD/EVEN SET
Reduction from COLORED LINEAR DEPENDENT SET.
Tool: as -bit vectors.
•
Reduction
Colored dependent set
MLD solution of weight
no dependent set of size
No MLD solution of weight
NEAREST CODEWORD
Given and find such that has Hamming weight at most
• Sparse version: in a YES instance, exists solution of Hamming weight at most .
• Inapproximability by an easy reduction from MAXIMUM LIKELIHOOD DECODING.
•
Minimum Distance (finally!)
SPARSE NEAREST CODEWORD
Given , find such that has Hamming weight at most
•
MINIMUM DISTANCE
Given and , find of such that has Hamming weight at most .
Reduction
codeword at distance
Distance is at most
no codeword at distance
Distance is at least
Increasing the gap
Codes with matrices , the tensor product is the code }
Fact:
Any inapproximability for MINIMUM DISTANCE can be boosted to and hence to any constant.
Theorem: MINIMUM DISTANCE (EVEN SET etc.) is randomized W[1]-hard to approximate for any .
•
Integer Lattice problems
SHORTEST VECTOR
Given a matrix and integer , find a vector with .
NEAREST VECTOR
Given a matrix , vector and integer , find an with .
Theorem: SHORTEST VECTOR for any and NEAREST VECTOR for any is randomized W[1]-hard to approximate for any
constant .
•
Summary
CLIQUE
ONE SIDED BICLIQUE LINEAR DEPENDENT SET
MAXIMUM LIKELIHOOD DECODING (SPARSE) NEAREST CODEWORD
MINIMUM DISTANCE
•
Transferring inapproximability results can be easier than transferring exact hardness!