You say itโs only constant? Then something must be hyperfinite! And I say your title is too long. Hendrik Fichtenberger July 20, 2019
Property Testing in a Nutshell planar 1
Property Testing in a Nutshell planar non-planar 1
Property Testing in a Nutshell planar non-planar non-planar 1
Property Testing in a Nutshell planar non-planar non-planar non-planar 1
Property Testing in a Nutshell planar non-planar non-planar non-planar time complexity: ฮฉ(|๐|) 1
Property Testing in a Nutshell planar non-planar non-planar non-planar time complexity: ฮฉ(|๐|) 1
Property Testing in a Nutshell planar non-planar non-planar non-planar time complexity: ฮฉ(|๐|) very quite slightly 1
Property Testing in a Nutshell planar non-planar non-planar non-planar time complexity: ฮฉ(|๐|) very quite slightly 1
Property Testing in a Nutshell planar non-planar non-planar non-planar ๐ 1 0 # edge edits |๐|+|๐น| 1
Property Testing in a Nutshell planar non-planar non-planar non-planar 3 3 ๐ 1 0 # edge edits |๐|+|๐น| 1 accept w.p. > 2 reject w.p. > 2
Property Testing in a Nutshell 1 ๐ -far ๐ -far ๐ -close |๐|+|๐น| # edge edits 0 ๐ planar 3 3 non-planar non-planar non-planar 1 accept w.p. > 2 reject w.p. > 2
Property Testing in a Nutshell 1 complexity: # queries to adjacency list entries ๐ -far ๐ -far ๐ -close |๐|+|๐น| # edge edits 0 ๐ planar 3 3 non-planar non-planar non-planar 1 accept w.p. > 2 reject w.p. > 2
Property Testing of Bounded Degree Graphs bounded degree graphs: โ๐ค โ ๐ โถ ๐(๐ค) โค ๐, ๐ โ ๐(1) ๐(๐) planar 2
Property Testing of Bounded Degree Graphs bounded degree graphs: โ๐ค โ ๐ โถ ๐(๐ค) โค ๐, ๐ โ ๐(1) ๐(๐) planar, degree-regular, cycle-free, subgraph-free, connected, minor-free, hyperfinite, ... 2
Property Testing of Bounded Degree Graphs bounded degree graphs: โ๐ค โ ๐ โถ ๐(๐ค) โค ๐, ๐ โ ๐(1) ๐(๐) planar, degree-regular, cycle-free, subgraph-free, connected, minor-free, hyperfinite, ... bipartite, expander ฮ(โ๐) 2
Property Testing of Bounded Degree Graphs bounded degree graphs: โ๐ค โ ๐ โถ ๐(๐ค) โค ๐, ๐ โ ๐(1) ๐(๐) planar, degree-regular, cycle-free, subgraph-free, connected, minor-free, hyperfinite, ... bipartite, expander ฮ(โ๐) big picture? 2
Property Testing of Bounded Degree Graphs bounded degree graphs: โ๐ค โ ๐ โถ ๐(๐ค) โค ๐, ๐ โ ๐(1) ๐(๐) planar, degree-regular, cycle-free, subgraph-free, connected, minor-free, hyperfinite, ... bipartite, expander ฮ(โ๐) big picture? bounded-degree graphs: litule known about constant-time testability 2
๐ -Disks and Frequency Vectors disk ๐ (๐ค) : subgraph induced by BFS (๐ค) of depth ๐ 3
๐ -Disks and Frequency Vectors disk 1 ( ) disk ๐ (๐ค) : subgraph induced by BFS (๐ค) of depth ๐ 3
๐ -Disks and Frequency Vectors disk 2 ( ) disk 1 ( ) disk ๐ (๐ค) : subgraph induced by BFS (๐ค) of depth ๐ 3
๐ -Disks and Frequency Vectors disk 2 ( 1 โ ) โฎ 0.6 0.4 freq 2 ( type calculate its share of vertices freq ๐ (๐ป) : for each ๐ -disk isomorphism by BFS (๐ค) of depth ๐ disk ๐ (๐ค) : subgraph induced ) disk 1 ( ) 3 ) = (
๐ -Disks and Frequency Vectors 0.4 locality feature frequency vector is a 1 โ ) โฎ 0.6 3 disk 2 ( freq 2 ( type calculate its share of vertices freq ๐ (๐ป) : for each ๐ -disk isomorphism by BFS (๐ค) of depth ๐ disk ๐ (๐ค) : subgraph induced ) disk 1 ( ) ) = (
Constant-Qvery Testers accept Goldreich, Ron, STOCโ09 1 2. accepts ifg โฬ freq ฮ(๐) (๐ป) of freq ฮ(๐) (๐ป) can be transformed into an algorithm that Every property tester with constant query complexity ๐ โถ= ๐(๐) Theorem [GRโ09, โฆ] ฮ ฮ ฮ ฮ ฮ 1 1 reject 4 1. computes an approximation ฬ ฮ(๐) for any ๐ป โฒ โ ฮ freq ฮ(๐) (๐ป) โ freq ฮ(๐) (๐ป โฒ )โ 1 โค
Hyperfinite Graphs Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5 In every planar graph, there exists a set of โ๐ separators
Hyperfinite Graphs Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5 In every planar graph, there exists a set of โ๐ separators
Hyperfinite Graphs 3 |๐| 3 |๐| Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5 In every planar graph, there exists a set of โ๐ separators > 1 > 1
Hyperfinite Graphs remove ๐๐ edges components of size ๐ โ2 Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5 In every planar graph, there exists a set of โ๐ separators
Hyperfinite Graphs ๐(0.5) ๐(0.5) ๐(0.5) ๐ = 0.5 Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5
Hyperfinite Graphs ๐ = 0.5 ๐(0.5) ๐(0.5) ๐(0.5) Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5
Hyperfinite Graphs ๐ = 0.5 ๐ = 0.6 ๐(0.6) ๐(0.6) ๐(0.6) Definition (๐, ๐ญ) -hyperfinite: can remove at most ๐๐๐ edges to obtain connected components of size at most ๐ก ๐ -hyperfinite: (๐, ๐(๐)) -hyperfinite for all ๐ โ (0, 1] 5
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite [GRโ09] Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite [GRโ09] [NSโ11] Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite [GRโ09] [NSโ11] Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite [GRโ09] [NSโ11] e.g. ฮ = connectivity contains expanders Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6
The Story so Far ฮ has constant query complexity ฮ is characterized by its k-disk distribution ฮ is ๐ -hyperfinite [GRโ09] [NSโ11] e.g. ฮ = connectivity contains expanders [FPSโ19] Newman, Sohler, STOCโ11; Fichtenberger, Peng, Sohler, SODAโ19 6 โ ฮ โฒ โ ฮ : ฮ โฒ is ๐ โฒ -hyperfinite
Small Frequency-Preserver Graphs freq ๐ (๐ป) freq ๐ (๐ผ) ๐ ๐ป ๐ผ ๐ ๐(๐, ๐) Theorem [Alonโ11] For every ๐, ๐ > 0 , there exists ๐(๐, ๐) such that for every ๐ป there Alon, โ11, see: Lovรกsz, Large Networks and Graph Limits , Proposition 19.10 7 exists ๐ผ of size at most ๐(๐, ๐) and โ freq ๐ (๐ป) โ freq ๐ (๐ผ)โ 1 < ๐ .
Small Frequency-Preserver Graphs freq ๐ (๐ป) freq ๐ (๐ผ) ๐ ๐ป ๐ผ ๐ ๐(๐, ๐) For ฯ-hyperfinite graphs M(ฮด,k) is at most Theorem [Alonโ11] For every ๐, ๐ > 0 , there exists ๐(๐, ๐) such that for every ๐ป there Alon, โ11, see: Lovรกsz, Large Networks and Graph Limits , Proposition 19.10 7 exists ๐ผ of size at most ๐(๐, ๐) and โ freq ๐ (๐ป) โ freq ๐ (๐ผ)โ 1 < ๐ .
Putuing Everything Into The Property Testing Blender in ฮ ๐ rel. ๐ธ size hypf. ? freq. v. change original start w/ ๐ป โ ฮ 8
Putuing Everything Into The Property Testing Blender original < ๐/2 ? ? freq. pres. ๐(1) start w/ ๐ป โ ฮ change in ฮ freq. v. ? hypf. size rel. ๐ธ ๐ 8
Putuing Everything Into The Property Testing Blender freq. pres. = 0 ๐ -close to ฮ blow-up hyperfinite (0, ๐(๐, ๐)) - ๐ < ๐/2 ? ? ๐(1) in ฮ start w/ ๐ป โ ฮ original change freq. v. ? hypf. size rel. ๐ธ ๐ 8
Recommend
More recommend