randomized algorithms for cuts and colouring david pritchard, nserc post-doctoral fellow
TRANSCRIPT
Randomized Algorithms for Cuts and Colouring
David Pritchard,NSERC Post-doctoral Fellow
What Can Randomness Do?
Part 1: Check 3-edge-connectivityin a distributed networkJoint with Ramakrishna Thurimella
(Denver)
Part 2: Find many disjoint setcovers as a function of min degreeJoint w/ Béla Bollobás (Cambridge &
Memphis), Thomas Rothvoß (MIT), Alex Scott (Oxford)
Distributed ComputationVertices are computers that
communicate using edges♫ initially, local/no knowledge♫ goal: compute global graph property
Distributed Computation
Message passing happens in rounds♫ time complexity: # rounds elapsed
Diameter := maximum distance (# hops) between two nodes
e.g., Diam = 5
Distributed Computation
Message passing happens in rounds♫ time complexity: # rounds elapsed
Diameter := maximum distance (# hops) between two nodes
♫ if message lengths are unrestricted, we can compute anything in O(Diameter) rounds
♫ “CONGEST” model: limit message lengths to O(log |V|) bits
Known Time Complexities
Synchronizer w/polylog overhead, AP’90Breadth-first spanning tree:♫ O(Diam) time♫ “Universally optimal”
Known Time Complexities
Synchronizer w/polylog overhead, AP’90Breadth-first spanning tree:♫ O(Diam) time♫ “Universally optimal”
Depth-first spanning tree:♫ O(|V|) time; no good lower bound
Min-cost spanning tree (KP’98, PR’99):♫ O(√V log*V + Diam), Ω(√V/log V)
Main Result
Distributed algorithm to check 3-edge-connectivity in O(Diam) time
♫ explicit: finds 2-edge-cuts
♫ application: reinforcement♫ beats prior O(Diam+V2); optimal
Main tool: sample cycle space randomly
Main Tool: Cycle Space
connected network/graph (V, E)♫ (V, F) Eulerian if ∀v, degF(v) is even
Cycle space := the vector space {F : (V, F) is Eulerian}
♫ notation abuse: F is a subset of E and also its characteristic vector ∈ {0, 1}E
♫ why is it a vector space?♫ even deg. ⊕ even deg. ≡ even deg.⊕ =
Random Sampling
Claim: if T is a spanning tree, any 0-1 vector on E\T extends uniquely to an Eulerian subgraph F
Corollary: sampling from the cycle space uniformly is as easy as sampling 2E\T
thick: T
green: in Fred: not in Fgrey: undecided
Cuts and Cycles
Claim. |δ(S)∩F| is even for all Eulerian F
♫ use Euler tour(s)!Claim. Unless E’ = δ(S) for some S,
for a random F from the cycle space, |E’ ∩ F| is even exactly ½ of the time.
Finding 2-Edge Cuts?
2 edges {e, e’} form a cut
♫ Implies transitivity: if {e, f} and {f, g} are 2-edge cuts, so is {e, g}
♫ Call equivalence classes “cut classes”some edges not in any2-edge-cuts
cut class
⇔ |{e, e’} ∩ F| even for all Eulerian F⇔ e and e’ always both or neither in F
Algorithm for 2-Edge Cuts
Set k = O(log |V|).Sample F1, F2,… Fk
from cycle space.Group equal rows
and output themas cut classes.
♫ {ei, ej} is a 2-edge-cut ⇔ ei, ej have equal rows, with error probability 2-k
♫ Pr[any error] ≤ |E|22-k = 1/poly(V)
F1 F2 F3 F4 F5 F6…
1 0 1 1 1 1…0 1 0 1 0 1…1 0 1 0 1 0…1 0 1 0 1 0…0 1 0 1 0 1… ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
e1
e2
e3
e4
e5
10110 ⋮
Questions♫ Leads to O(Diam + Δ/log V)-time
algorithm for cut vertices♫ O(Diam + √V log*V) known before
(Thurimella ’97)♫ Is O(Diam) possible or not?
♫ Can this be derandomized?♫ in a distributed way?
Festival Scheduling♫ Set V of musicians, list of bands ⊆ V
♫ A schedule maps each band to a day♫ Each musician must play every day♫ What is max # days in schedule?
The Basic Question
Input: A set system/hypergraph (V, H)♫ each S ∈ H is a hyperedge S ⊆ V
A cover-decomposition is a partitionH = H1 H⨄ 2 … H⨄ ⨄ k
s.t. each Hi covers V, i.e. ⋃{S | S∈Hi} = V♫ equivalently, a polychromatic coloring
goal: maximize #parts/#colours
cd: Cover-Decomposition Number
cd(H) ≔ largest k for which there is a cover-decomposition into k parts
♫ Easy: cd ≤ minimum degree ≕ δ
Does a converse hold? If δ ≥ 100 (H covers every point 100 times), can we get many disjoint covers?
♫ In general, no: cd = 1 is still possible♫ But for specific families…
Cover-Decomposition in Graphs
R. P. Gupta, 1970s:In a graph, cd ≥ δ-1.
In a multigraph, cd ≥ ⌊3δ+1/4⌋.
Tight multigraph examples:
δ=2 cd=1
δ=3 cd=2
…δ=4 cd=3
Ground set = finite X ⊆ Rd,edges = subsets of X covered by shapes
Cover-Decomposition in Geometry
cd = Ω(δ) not cover-decomposable
Translates of any convex polygon
Translates of any non-convex quadrilateral
Axis-aligned strips Axis-aligned rectangles
Halfspaces in 2D Unit strips in 2D
3D orthants 4D orthants
Cover-Decomposition in Geometry
Do all hypergraph families* satisfy this
dichotomy? [Pálvölgyi, Keszegh]*closed under edge deletion, duplication
Conjecture [Pach, 1980]:♫ for any fixed convex set S, there is δS,
so that hypergraphs with a finite ground set in R2 and translates of S as edges, with δ ≥ δS, have cd(H) ≥ 2.
♪ “The family is cover-decomposable.”
Our Results
Hypergraphs with bounded edge size R♫ cd ≳ δ/log R; this is tight
Techniques: LLL, discrepancy, LPsHypergraphs of paths in trees♫ cd ≥ δ/5
Hypergraphs of VC-dimension ≤ D♫ cover-decomposable only for D = 1
Goal: find out how cover-decomposition number (cd) depends on minimum degree (δ) in as many natural hypergraph families as possible.
Lovasz Local Lemma:
There are any number of “bad”events, but each is independentof all but D others.
♫ LLL: If each bad event has individual probability at most 1/eD, then
Pr[no “bad” events happen] > 0.
Natural to try in our setting: randomly k-colour the edges
/
Edge size ≤ R
Pick some k, randomly k-colour edges.♫ bad event: “vertex v misses colour c.”
Dependence degree ≤ k×R×(max degree)
♫ set all degrees to δ by “shrinking”
Analyze: Pr[v misses c] ≤ (1-1/k)δ ≤ e-δ/k
♫ Need δRk × e-δ/k < 1/e dependency degree
♫ ∴ cd ≥ Ω(δ/(log R + log δ))
vS S\{v}→
Splitting the Hypergraph
Ω(δ/log Rδ) is already Ω(δ/log R)if δ ≤ poly(R)
♫ Idea: partition edges to H1,H2,…,HM with δ(Hi) ≤ poly(R), δ(Hi) ~ δ(H)/M
=Ω(δ(H)/M/log R) covers
Ω(δ(H)/M/log R) covers
Ω(δ(H)/M/log R) coversM=3
~δ/log R covers
Ω(δ(Hi)/log R) covers
Beck-Fiala 1981: there is anassignment with discrepancy ≤ 2R
Iterated Pairwise Splitting
Split H into H+, H- so that
∀v, ±: deg(v, H±) ≥ deg(v, H)/2 - ε
♫ Equivalent view: assign ±1 to edges, s.t. |total weight on each vertex| ≤ 2ε
discrepancy!
Beck-Fiala Algorithm
LP variables: ∀S: 0 ≤ xS = 1 - yS ≤ 1
∀v: ΣS:v∈S xS ≥ δ/2, ΣS:v∈S yS ≥ δ/2
1. find extreme point LP solution2. “fix” variables with values 0 or 13. discard all constraints involving ≤ R
non-fixed variables♫ Termination lemma
♪ basis of tight degree constraints has size ≤ |Hnonfixed|; each var is in ≤R constraints
Remarks
Maximum edge size R:♫ use better discrepancy bound to get
right multiplicative constant♫ Concentration/LLL instead of B-F
cd ≥ δ /5 for paths in trees:♫ B-F, different termination lemma♪ linear independence of basis
Sparse Hypergraphs[Alon-Berke-Buchin2-Csorba-Shannigrahi-Speckmann-Zumstein]
(α, β)-sparse hypergraph:= incidences(U ⊆ V, F ⊆ H) ≤ α|U|+β|
F|♫ ⇔: “α-vertex-sparse” incidences ⨄
“β-edge-sparse” incidences♫ idea: shrink off β-edge-sparse ones,
obtaining cd ≳ (δ-α)/log β vertices
hyperedges
bipartiteincidence
graph
≤ α
≤ β
Cover Scheduling
Hyperedges are sensors monitoring V♫ Each hyperedge S has battery life dS
♫ Goal: schedule when each should turn on, so V is covered from time 0 to T
♫ How large can T be?(Cover-decomposition: d ≡ 1)♫ Result: Ω(min point coverage/R)
schedule is possible♫ Open Q: improve R to log R!