Upload
others
View
37
Download
0
Embed Size (px)
Citation preview
Approximation Algorithms forMulticommodity-Type Problems with Guarantees
Independent of the Graph Size
Ankur Moitra, MIT
October 23, 2009
Ankur Moitra () Multicommodity October 23, 2009 1 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
Ankur Moitra () Multicommodity October 23, 2009 2 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
Ankur Moitra () Multicommodity October 23, 2009 3 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
Ankur Moitra () Multicommodity October 23, 2009 4 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
The cost is 6
Ankur Moitra () Multicommodity October 23, 2009 5 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
Ankur Moitra () Multicommodity October 23, 2009 6 / 116
Background
The Minimum Bisection Problem
Goal: Minimize cost of bisection
Ankur Moitra () Multicommodity October 23, 2009 7 / 116
Background
The Minimum Bisection Problem
4
Goal: Minimize cost of bisection
The cost is
Ankur Moitra () Multicommodity October 23, 2009 8 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
History of the Minimum Bisection Problem
1 Applications through Divide-and-Conquer: VLSI design, sparse matrixcomputations, approximation algorithms
2 [Kernighan, Lin 1970] Local search heuristic
3 [Garey, Johnson, Stockmeyer 1976] NP-Complete
4 [Leighton, Rao 1988] O(log n) approximate minimum bisection
5 [Saran, Vazirani 1995] n2 -approximation algorithm
6 [Arora, Karger, Karpinski 1999] PTAS for dense graphs
7 [Feige, Krauthgamer 2001] O(log1.5 n)-approximation algorithm
8 [Khot 2004] No PTAS, unless P = NP
9 [Racke, 2008] O(log n)-approximation algorithm
Ankur Moitra () Multicommodity October 23, 2009 9 / 116
Background
The Steiner Minimum Bisection Problem
Goal: Minimize cost of a bisection of the k blue nodes
Ankur Moitra () Multicommodity October 23, 2009 10 / 116
Background
The Steiner Minimum Bisection Problem
Goal: Minimize cost of a bisection of the k blue nodes
Ankur Moitra () Multicommodity October 23, 2009 11 / 116
Background
The Steiner Minimum Bisection Problem
Goal: Minimize cost of a bisection of the k blue nodes
The cost is 4
Ankur Moitra () Multicommodity October 23, 2009 12 / 116
Background
Question
Can we find a poly(log k)-approximation algorithm?
Some approximation guarantees can be made independent of the graphsize:
1 O(log k) generalized sparsest cut for k commodities[Linial, London, Rabinovich 1995] and [Aumann, Rabani 1997]
2 O(log k) multicut for k terminals[Garg, Vazirani, Yannakakis 1996]
3 O( log klog log k ) 0-extension for k terminals
[Fakcharoenphol, Harrelson, Rao, Talwar 2003]
Ankur Moitra () Multicommodity October 23, 2009 13 / 116
Background
Question
Can we find a poly(log k)-approximation algorithm?
Some approximation guarantees can be made independent of the graphsize:
1 O(log k) generalized sparsest cut for k commodities[Linial, London, Rabinovich 1995] and [Aumann, Rabani 1997]
2 O(log k) multicut for k terminals[Garg, Vazirani, Yannakakis 1996]
3 O( log klog log k ) 0-extension for k terminals
[Fakcharoenphol, Harrelson, Rao, Talwar 2003]
Ankur Moitra () Multicommodity October 23, 2009 13 / 116
Background
Question
Can we find a poly(log k)-approximation algorithm?
Some approximation guarantees can be made independent of the graphsize:
1 O(log k) generalized sparsest cut for k commodities[Linial, London, Rabinovich 1995] and [Aumann, Rabani 1997]
2 O(log k) multicut for k terminals[Garg, Vazirani, Yannakakis 1996]
3 O( log klog log k ) 0-extension for k terminals
[Fakcharoenphol, Harrelson, Rao, Talwar 2003]
Ankur Moitra () Multicommodity October 23, 2009 13 / 116
Background
Question
Can we find a poly(log k)-approximation algorithm?
Some approximation guarantees can be made independent of the graphsize:
1 O(log k) generalized sparsest cut for k commodities[Linial, London, Rabinovich 1995] and [Aumann, Rabani 1997]
2 O(log k) multicut for k terminals[Garg, Vazirani, Yannakakis 1996]
3 O( log klog log k ) 0-extension for k terminals
[Fakcharoenphol, Harrelson, Rao, Talwar 2003]
Ankur Moitra () Multicommodity October 23, 2009 13 / 116
Background
Question
Can we find a poly(log k)-approximation algorithm?
Some approximation guarantees can be made independent of the graphsize:
1 O(log k) generalized sparsest cut for k commodities[Linial, London, Rabinovich 1995] and [Aumann, Rabani 1997]
2 O(log k) multicut for k terminals[Garg, Vazirani, Yannakakis 1996]
3 O( log klog log k ) 0-extension for k terminals
[Fakcharoenphol, Harrelson, Rao, Talwar 2003]
Ankur Moitra () Multicommodity October 23, 2009 13 / 116
Background
A Meta Question
Given
A poly(log n) approximation algorithm (integrality gap or competitiveratio) for a multicommodity-type problem
Let k be the number of ”interesting” nodes
Meta Question
Can we give a poly(log k) approximation algorithm (integrality gap orcompetitive ratio)?
Yes we can, and ...
Ankur Moitra () Multicommodity October 23, 2009 14 / 116
Background
A Meta Question
Given
A poly(log n) approximation algorithm (integrality gap or competitiveratio) for a multicommodity-type problem
Let k be the number of ”interesting” nodes
Meta Question
Can we give a poly(log k) approximation algorithm (integrality gap orcompetitive ratio)?
Yes we can, and ...
Ankur Moitra () Multicommodity October 23, 2009 14 / 116
Background
A Meta Question
Given
A poly(log n) approximation algorithm (integrality gap or competitiveratio) for a multicommodity-type problem
Let k be the number of ”interesting” nodes
Meta Question
Can we give a poly(log k) approximation algorithm (integrality gap orcompetitive ratio)?
Yes we can, and ...
Ankur Moitra () Multicommodity October 23, 2009 14 / 116
Background
A Meta Question
Given
A poly(log n) approximation algorithm (integrality gap or competitiveratio) for a multicommodity-type problem
Let k be the number of ”interesting” nodes
Meta Question
Can we give a poly(log k) approximation algorithm (integrality gap orcompetitive ratio)?
Yes we can, and ...
Ankur Moitra () Multicommodity October 23, 2009 14 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extension
And Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Background
Our Results
We give the first poly(log k)-approximation algorithms (or competitiveratios) for:
1 Steiner minimum bisection
2 requirement cut
3 l-multicut
4 oblivious 0-extensionAnd Steiner generalizations of:
5 oblivious routing
6 min-cut linear arrangement
7 minimum linear arrangement
Our approach is to prove the existence of vertex sparsifiers that preservecuts...
Ankur Moitra () Multicommodity October 23, 2009 15 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
b
d
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
Ankur Moitra () Multicommodity October 23, 2009 16 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
b
d
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
Ankur Moitra () Multicommodity October 23, 2009 17 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
b
d
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
Ankur Moitra () Multicommodity October 23, 2009 18 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
h (a) = 5
d
K
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
c
a
b
Ankur Moitra () Multicommodity October 23, 2009 19 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
h (a) = 5
d
K
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
c
a
b
Ankur Moitra () Multicommodity October 23, 2009 20 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
h (a) = 5
d
K
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
c
a
b
Ankur Moitra () Multicommodity October 23, 2009 21 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 22 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 23 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 24 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (b) = 2
Ankur Moitra () Multicommodity October 23, 2009 25 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (b) = 2
Ankur Moitra () Multicommodity October 23, 2009 26 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (b) = 2
Ankur Moitra () Multicommodity October 23, 2009 27 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 28 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 29 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Ankur Moitra () Multicommodity October 23, 2009 30 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (ac) = 4
Ankur Moitra () Multicommodity October 23, 2009 31 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (ac) = 4
Ankur Moitra () Multicommodity October 23, 2009 32 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
a
c d
b
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
Kh (ac) = 4
Ankur Moitra () Multicommodity October 23, 2009 33 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
2
1__2
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
a
c d
b
3
__2
3__2
5__2
1
1__
Ankur Moitra () Multicommodity October 23, 2009 34 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
1__2
h (a) = 5
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
a
c d
b
3
__2
3__2
5__2
1
1__2
Ankur Moitra () Multicommodity October 23, 2009 35 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
__2
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
a
c d
b
3
__2
3__2
5__2
1
1__2
1
Ankur Moitra () Multicommodity October 23, 2009 36 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
__2
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
a
c d
b
3
__2
3__2
5__2
1
1__2
1
Ankur Moitra () Multicommodity October 23, 2009 37 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
__2
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
Kh (b) = 2
a
c d
b
3
__2
3__2
5__2
1
1__2
1
Ankur Moitra () Multicommodity October 23, 2009 38 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
2
h’(b) = 2.5
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
Kh (b) = 2
a
c d
b
3
__2
3__2
5__2
1
1__2
1__
Ankur Moitra () Multicommodity October 23, 2009 39 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
2
h’(b) = 2.5
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
Kh (b) = 2
a
c d
b
3
__2
3__2
5__2
1
1__2
1__
Ankur Moitra () Multicommodity October 23, 2009 40 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
2
h’(b) = 2.5
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
h (b) = 2
h (ac) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__
Ankur Moitra () Multicommodity October 23, 2009 41 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
h’(b) = 2.5
h’(ac) = 4.5
h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
h (b) = 2
h (ac) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
Ankur Moitra () Multicommodity October 23, 2009 42 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 43 / 116
Vertex Sparsifiers
General Approach: Vertex Sparsifiers
h (b) = 2
h’(a) = 6
c
The of the Sparsifer isQuality ___54
a
b
d
K
K
K
K
K
K
K
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
h’(ad) = 5h’(d) = 5h (a) = 5
Ankur Moitra () Multicommodity October 23, 2009 44 / 116
Vertex Sparsifiers
Vertex Sparsifiers, Informally
Definition
G ′ = (K ,E ′) is a Vertex Sparsifier for G = (V ,E ) if all cuts in G ′ are atleast as large as the corresponding min-cut in G .
Definition
The Quality of a Vertex Sparsifier is the maximum ratio of a cut in G ′ tothe corresponding min-cut in G .
Ankur Moitra () Multicommodity October 23, 2009 45 / 116
Vertex Sparsifiers
Vertex Sparsifiers, Informally
Definition
G ′ = (K ,E ′) is a Vertex Sparsifier for G = (V ,E ) if all cuts in G ′ are atleast as large as the corresponding min-cut in G .
Definition
The Quality of a Vertex Sparsifier is the maximum ratio of a cut in G ′ tothe corresponding min-cut in G .
Ankur Moitra () Multicommodity October 23, 2009 45 / 116
Vertex Sparsifiers
Vertex Sparsifiers
Good Vertex Sparsifiers exist!
And can be computed in polynomial time!
Theorem
For all weighted graphs G = (V ,E ), and all K ⊂ V there is a weightedgraph G ′ = (K ,E ′) such that G ′ is a poly(log k)-quality Vertex Sparsifierand such a graph can be computed in time polynomial in n and k.
Ankur Moitra () Multicommodity October 23, 2009 46 / 116
Vertex Sparsifiers
Vertex Sparsifiers
Good Vertex Sparsifiers exist!
And can be computed in polynomial time!
Theorem
For all weighted graphs G = (V ,E ), and all K ⊂ V there is a weightedgraph G ′ = (K ,E ′) such that G ′ is a poly(log k)-quality Vertex Sparsifierand such a graph can be computed in time polynomial in n and k.
Ankur Moitra () Multicommodity October 23, 2009 46 / 116
Vertex Sparsifiers
Vertex Sparsifiers
Good Vertex Sparsifiers exist!
And can be computed in polynomial time!
Theorem
For all weighted graphs G = (V ,E ), and all K ⊂ V there is a weightedgraph G ′ = (K ,E ′) such that G ′ is a poly(log k)-quality Vertex Sparsifierand such a graph can be computed in time polynomial in n and k.
Ankur Moitra () Multicommodity October 23, 2009 46 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 47 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 48 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 49 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 50 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 51 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 52 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 53 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 54 / 116
Vertex Sparsifiers
An Application to Steiner Minimum Bisection
c
h’(ad) = 5h’(d) = 5h (a) = 5 h’(a) = 6
Graph G=(V,E) Sparsifier G’=(K,E’)
a
b
d
K
K
K
K
K
K
K
h (b) = 2
h (c) = 3
h (ab) = 7
h (ad) = 5
h (ac) = 4
h (d) = 4
a
c d
b
3
__2
3__2
5__2
1
1__2
1__2
h’(b) = 2.5
h’(c) = 3.5
h’(ab) = 7.5
h’(ac) = 4.5
Ankur Moitra () Multicommodity October 23, 2009 55 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Reducibility in Multicommodity-Type Problems
This is a general strategy!
Reducibility for Multicommodity-Type Problems:
1 Construct G ′ so OPT ′ ≤ poly(log k)OPT
2 Run approximation algorithm on G ′
3 Map solution back to G
Informally: If the problem is Lipschitz w.r.t. hK , then this is a generalreduction!
Ankur Moitra () Multicommodity October 23, 2009 56 / 116
Vertex Sparsifiers
Definition
Let f : V → K , denote a 0-extension if for all a ∈ K , f (a) = a.
G=(V,E)
Ankur Moitra () Multicommodity October 23, 2009 57 / 116
Vertex Sparsifiers
Definition
Let f : V → K , denote a 0-extension if for all a ∈ K , f (a) = a.
G=(V,E)
Ankur Moitra () Multicommodity October 23, 2009 58 / 116
Vertex Sparsifiers
Definition
Let f : V → K , denote a 0-extension if for all a ∈ K , f (a) = a.
G =(K,E )f f
Ankur Moitra () Multicommodity October 23, 2009 59 / 116
Vertex Sparsifiers
Definition
Let f : V → K , denote a 0-extension if for all a ∈ K , f (a) = a.
G =(K,E )f f
Ankur Moitra () Multicommodity October 23, 2009 60 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
G =(K,E )f f
Ankur Moitra () Multicommodity October 23, 2009 61 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
G =(K,E )K−A
A
f f
Ankur Moitra () Multicommodity October 23, 2009 62 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
G =(K,E )K−A
A
f f
Ankur Moitra () Multicommodity October 23, 2009 63 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
A
G=(V,E)K−A
Ankur Moitra () Multicommodity October 23, 2009 64 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
A
G=(V,E)K−A
Ankur Moitra () Multicommodity October 23, 2009 65 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
A
G=(V,E)K−A
Ankur Moitra () Multicommodity October 23, 2009 66 / 116
Vertex Sparsifiers
Lemma
Gf is a vertex sparsifier
A
G=(V,E)K−A
Ankur Moitra () Multicommodity October 23, 2009 67 / 116
An Example
An Example
Ankur Moitra () Multicommodity October 23, 2009 68 / 116
An Example
An Example
Ankur Moitra () Multicommodity October 23, 2009 69 / 116
An Example
An Example
Ankur Moitra () Multicommodity October 23, 2009 70 / 116
An Example
An Example
k−1The cost is
Ankur Moitra () Multicommodity October 23, 2009 71 / 116
An Example
An Example
k−1The cost is The cost is1
Ankur Moitra () Multicommodity October 23, 2009 72 / 116
An Example
An Example
of the Sparsifier is k−1
The cost is The cost is1 k−1
The Quality
Ankur Moitra () Multicommodity October 23, 2009 73 / 116
An Example
An Example: A Second Attempt
k
p = 1__k
p = 1__
Ankur Moitra () Multicommodity October 23, 2009 74 / 116
An Example
An Example: A Second Attempt
k
__k
1__k
1__k
1k
__
1__
p = 1__k
p = 1__k
1
Ankur Moitra () Multicommodity October 23, 2009 75 / 116
An Example
An Example: A Second Attempt
k
k
2
1__
k1__k
1__k
1__p = 1__k
p = 1__k
__k
1__k
1__k
1k
__
1__
Ankur Moitra () Multicommodity October 23, 2009 76 / 116
An Example
An Example: A Second Attempt
k
p = 1__k
p = 1__k
2__
Ankur Moitra () Multicommodity October 23, 2009 77 / 116
An Example
An Example: A Second Attempt
k
2__
Ankur Moitra () Multicommodity October 23, 2009 78 / 116
An Example
An Example: A Second Attempt
min(|A|, |K−A|)
2__
k
A
The cost is
Ankur Moitra () Multicommodity October 23, 2009 79 / 116
An Example
An Example: A Second Attempt
The cost is at most
2__
k
A
The cost is min(|A|, |K−A|) 2min(|A|, |K−A|)
Ankur Moitra () Multicommodity October 23, 2009 80 / 116
An Example
An Example: A Second Attempt
of the Sparsifer is < 2
2__
k
A
The cost is min(|A|, |K−A|) 2min(|A|, |K−A|)The cost is at most
The Quality
Ankur Moitra () Multicommodity October 23, 2009 81 / 116
An Example
Approximate Vertex Sparsifiers via 0-Extensions
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a distribution γ on0-extensions such that G ′ =
∑f γ(f )Gf is a O( log k
log log k )-quality VertexSparsifier for G ,K .
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a polynomial (in nand k) time algorithm to construct G ′ that is a poly(log k)-quality VertexSparsifier for G ,K .
Ankur Moitra () Multicommodity October 23, 2009 82 / 116
An Example
Approximate Vertex Sparsifiers via 0-Extensions
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a distribution γ on0-extensions such that G ′ =
∑f γ(f )Gf is a O( log k
log log k )-quality VertexSparsifier for G ,K .
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a polynomial (in nand k) time algorithm to construct G ′ that is a poly(log k)-quality VertexSparsifier for G ,K .
Ankur Moitra () Multicommodity October 23, 2009 82 / 116
An Example
Approximate Vertex Sparsifiers via 0-Extensions
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a distribution γ on0-extensions such that G ′ =
∑f γ(f )Gf is a O( log k
log log k )-quality VertexSparsifier for G ,K .
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a polynomial (in nand k) time algorithm to construct G ′ that is a poly(log k)-quality VertexSparsifier for G ,K .
Ankur Moitra () Multicommodity October 23, 2009 83 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 84 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 85 / 116
Non-constructive Proof
The Extension-Cut Game
P1
P2
Ankur Moitra () Multicommodity October 23, 2009 86 / 116
Non-constructive Proof
The Extension-Cut Game
f
P2
P1A
Ankur Moitra () Multicommodity October 23, 2009 87 / 116
Non-constructive Proof
The Extension-Cut Game
N(f,A)
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 88 / 116
Non-constructive Proof
The Extension-Cut Game
N(f,A)
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 89 / 116
Non-constructive Proof
The Extension-Cut Game
A=aA
f N(f,A)
K−A
P2
P1
Ankur Moitra () Multicommodity October 23, 2009 90 / 116
Non-constructive Proof
The Extension-Cut Game
A=aA
f N(f,A)
K−A
P2
P1
Ankur Moitra () Multicommodity October 23, 2009 91 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
P2
P1A
f N(f,A)
Kh (a) = 5
K−A
Ankur Moitra () Multicommodity October 23, 2009 92 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
N(f,A)
Kh (a) = 5
K−A
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 93 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
N(f,A)
Kh (a) = 5
K−A
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 94 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
N(f,A)
Kh (a) = 5
K−A
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 95 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
N(f,A)
Kh (a) = 5
K−A
P2
P1A
f
Ankur Moitra () Multicommodity October 23, 2009 96 / 116
Non-constructive Proof
The Extension-Cut Game
A=a
h (a) = 5
h (a) = 7f
K−A
P2
P1A
f N(f,A)
K
Ankur Moitra () Multicommodity October 23, 2009 97 / 116
Non-constructive Proof
The Extension-Cut Game
5A=a
N(f,A) = 7__P2
P1A
f N(f,A)
Kh (a) = 5
h (a) = 7f
K−A
Ankur Moitra () Multicommodity October 23, 2009 98 / 116
Non-constructive Proof
Definition
Let ν denote the game value of the extension-cut game
So ∃ a distribution γ on 0-extensions s.t. for all A ⊂ K :
Ef←γ [N(f ,A)] ≤ ν
Ef←γ [N(f ,A)] =∑
f ∈supp(γ)
γ(f )hf (A)
hK (A)=
Ef←γ [hf (A)]
hK (A)≤ ν
Let G ′ =∑
f γ(f )Gf ; for all A ⊂ K :
h′(A) = Ef←γ [hf (A)] ≤ νhK (A)
Ankur Moitra () Multicommodity October 23, 2009 99 / 116
Non-constructive Proof
Definition
Let ν denote the game value of the extension-cut game
So ∃ a distribution γ on 0-extensions s.t. for all A ⊂ K :
Ef←γ [N(f ,A)] ≤ ν
Ef←γ [N(f ,A)] =∑
f ∈supp(γ)
γ(f )hf (A)
hK (A)=
Ef←γ [hf (A)]
hK (A)≤ ν
Let G ′ =∑
f γ(f )Gf ; for all A ⊂ K :
h′(A) = Ef←γ [hf (A)] ≤ νhK (A)
Ankur Moitra () Multicommodity October 23, 2009 99 / 116
Non-constructive Proof
Definition
Let ν denote the game value of the extension-cut game
So ∃ a distribution γ on 0-extensions s.t. for all A ⊂ K :
Ef←γ [N(f ,A)] ≤ ν
Ef←γ [N(f ,A)] =∑
f ∈supp(γ)
γ(f )hf (A)
hK (A)=
Ef←γ [hf (A)]
hK (A)≤ ν
Let G ′ =∑
f γ(f )Gf ; for all A ⊂ K :
h′(A) = Ef←γ [hf (A)] ≤ νhK (A)
Ankur Moitra () Multicommodity October 23, 2009 99 / 116
Non-constructive Proof
Definition
Let ν denote the game value of the extension-cut game
So ∃ a distribution γ on 0-extensions s.t. for all A ⊂ K :
Ef←γ [N(f ,A)] ≤ ν
Ef←γ [N(f ,A)] =∑
f ∈supp(γ)
γ(f )hf (A)
hK (A)=
Ef←γ [hf (A)]
hK (A)≤ ν
Let G ′ =∑
f γ(f )Gf ; for all A ⊂ K :
h′(A) = Ef←γ [hf (A)] ≤ νhK (A)
Ankur Moitra () Multicommodity October 23, 2009 99 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 100 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 101 / 116
Non-constructive Proof
Best Response?
Let µ = 12a, b+ 1
2a, d
d
a
b
c
Ankur Moitra () Multicommodity October 23, 2009 102 / 116
Non-constructive Proof
Best Response?
Let µ = 12a, b+ 1
2a, d
p =
d
1__
2a
b
c
Ankur Moitra () Multicommodity October 23, 2009 103 / 116
Non-constructive Proof
Best Response?
Let µ = 12a, b+ 1
2a, d
2
p =
p = 1__
a
b
c
d
1__
2
Ankur Moitra () Multicommodity October 23, 2009 104 / 116
Non-constructive Proof
Best Response?
Let µ = 12a, b+ 1
2a, d
2
p =
p = 1__
a
b
c
d
1__
2
Ankur Moitra () Multicommodity October 23, 2009 105 / 116
Non-constructive Proof
Best Response?
Let µ = 12a, b+ 1
2a, d
2
p =
p = 1__
a
b
c
d
1__
2
Ankur Moitra () Multicommodity October 23, 2009 106 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 107 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 108 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value
[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 109 / 116
Non-constructive Proof
Proof Outline
1 Define a Zero-Sum Game
2 The Best Response is a 0-Extension Problem
3 Construct a Feasible Solution for the Linear
Programming Relaxation
4 Round the solution to bound the Game Value[Fakcharoenphol, Harrelson, Rao, Talwar 2003][Calinescu, Karloff, Rabani 2001]
Ankur Moitra () Multicommodity October 23, 2009 110 / 116
Constructive Results
Approximate Vertex Sparsifiers via 0-Extensions
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a distribution γ on0-extensions such that G ′ =
∑f γ(f )Gf is a O( log k
log log k )-quality VertexSparsifier for G ,K .
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a polynomial (in nand k) time algorithm to construct G ′ that is a poly(log k)-quality VertexSparsifier for G ,K .
Ankur Moitra () Multicommodity October 23, 2009 111 / 116
Constructive Results
Approximate Vertex Sparsifiers via 0-Extensions
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a distribution γ on0-extensions such that G ′ =
∑f γ(f )Gf is a O( log k
log log k )-quality VertexSparsifier for G ,K .
Theorem
For all graphs G = (V ,E ) and all sets K ⊂ V , there is a polynomial (in nand k) time algorithm to construct G ′ that is a poly(log k)-quality VertexSparsifier for G ,K .
Ankur Moitra () Multicommodity October 23, 2009 112 / 116
Constructive Results
Constructively Finding G ′
Unfortunately, this is a whole other talk!
Ankur Moitra () Multicommodity October 23, 2009 113 / 116
Improvements and Open Questions
Improvements and Open Questions
Let PG ⊂ <(k2) be the set of all demands that can be routed in G (where
demands are restricted to K ).
Theorem
There is a graph G ′ = (K ,E ′) such that PG ⊂ PG ′ ⊂ O( log klog log k )PG
This is a strengthening of the results presented here!
Open Question
What is the integrality gap of the linear program for 0-extensions?
Is it√
log k , log klog log k , ...?
Ankur Moitra () Multicommodity October 23, 2009 114 / 116
Improvements and Open Questions
Improvements and Open Questions
Let PG ⊂ <(k2) be the set of all demands that can be routed in G (where
demands are restricted to K ).
Theorem
There is a graph G ′ = (K ,E ′) such that PG ⊂ PG ′ ⊂ O( log klog log k )PG
This is a strengthening of the results presented here!
Open Question
What is the integrality gap of the linear program for 0-extensions?
Is it√
log k , log klog log k , ...?
Ankur Moitra () Multicommodity October 23, 2009 114 / 116
Improvements and Open Questions
Improvements and Open Questions
Let PG ⊂ <(k2) be the set of all demands that can be routed in G (where
demands are restricted to K ).
Theorem
There is a graph G ′ = (K ,E ′) such that PG ⊂ PG ′ ⊂ O( log klog log k )PG
This is a strengthening of the results presented here!
Open Question
What is the integrality gap of the linear program for 0-extensions?
Is it√
log k , log klog log k , ...?
Ankur Moitra () Multicommodity October 23, 2009 114 / 116
Improvements and Open Questions
Improvements and Open Questions
Let PG ⊂ <(k2) be the set of all demands that can be routed in G (where
demands are restricted to K ).
Theorem
There is a graph G ′ = (K ,E ′) such that PG ⊂ PG ′ ⊂ O( log klog log k )PG
This is a strengthening of the results presented here!
Open Question
What is the integrality gap of the linear program for 0-extensions?
Is it√
log k , log klog log k , ...?
Ankur Moitra () Multicommodity October 23, 2009 114 / 116
Improvements and Open Questions
Improvements and Open Questions
Theorem (Leighton, M)
There is a graph G and a set K so that for any graph G ′ = (K ,E ′) whichis a better flow network - i.e. PG ⊂ PG ′ , PG ′ * Ω(log log k)PG
Open Question
What if we are only interested in cuts, and not all flows? Is there an O(1)upper bound?
Equivalently, is there an O(1) upper bound for the linear program for0-extensions when ∆ is an `1 semi-metric?
Ankur Moitra () Multicommodity October 23, 2009 115 / 116
Improvements and Open Questions
Improvements and Open Questions
Theorem (Leighton, M)
There is a graph G and a set K so that for any graph G ′ = (K ,E ′) whichis a better flow network - i.e. PG ⊂ PG ′ , PG ′ * Ω(log log k)PG
Open Question
What if we are only interested in cuts, and not all flows? Is there an O(1)upper bound?
Equivalently, is there an O(1) upper bound for the linear program for0-extensions when ∆ is an `1 semi-metric?
Ankur Moitra () Multicommodity October 23, 2009 115 / 116
Improvements and Open Questions
Improvements and Open Questions
Theorem (Leighton, M)
There is a graph G and a set K so that for any graph G ′ = (K ,E ′) whichis a better flow network - i.e. PG ⊂ PG ′ , PG ′ * Ω(log log k)PG
Open Question
What if we are only interested in cuts, and not all flows? Is there an O(1)upper bound?
Equivalently, is there an O(1) upper bound for the linear program for0-extensions when ∆ is an `1 semi-metric?
Ankur Moitra () Multicommodity October 23, 2009 115 / 116
Thanks!
Thanks!
Ankur Moitra () Multicommodity October 23, 2009 116 / 116