Title: A new family of expander Cayley graphs
1A new family of expander Cayley graphs (?)
- Eyal Rozenman, Hebrew University
- Aner Shalev, Hebrew University
- Avi Wigderson, IAS
- You need the TexPoint PowerPoint add-in to view
this presentation properly - Download it from http//raw.cs.berkeley.edu/texpoi
nt/
2Definitions
Undirected, regular (multi)graphs. Definition.
The 2nd eigenvalue of a d-regular X ?(X) max
(AX /d) v v1 , v ? 1 ?(X) ?
0,1 Definition. Xn is an expander family if
?(Xn)? ?lt1 Equivalent RW on X converges in time
O(logX)
3Cayley graphs
- G a finite group
- U ½ G a (symmetric) set of generators.
- Definition Cayley graph C(G,U)
- Vertices elements of G
- Edges (g, gu) for all u?U.
- C(G,U) is regular with degree U.
- C(G,U) is connected , U generates G.
4Our Sequence of groups
- Gn Even symmetries of a (rooted) d-regular
tree. - degree (children) d (fixed). Depth n
5Our Sequence of groups
- Gn Even symmetries of a (rooted) d-regular
tree. - degree (children) d (fixed). Depth n
- Depth 1 tree symmetries alternating group Ad
6Our Sequence of groups
- Gn Even symmetries of a (rooted) d-regular
tree. - degree (children) d (fixed). Depth n
- Depth 1 tree symmetries alternating group Ad
- Theorem Under some assumption on Ad
- Every Gn has (const. ) expanding generators.
7Assumption on alternating gp. Ad
- Our construction is based on
- Assumption 9 U ½ Ad ( G1) such that
- U d1/30
- ?(Ad,U) 1/1000
- This is an Open problem (will discuss this more
later)
8Main theorem
- Thm RSW 9 Un ½ Gn such that
- Un d1/7 (constant - independent of n)
- C(Gn,Un) is a good expander (?(GnUn) 1/1000)
9Main theorem
- Thm RSW 9 Un ½ Gn such that
- Un d1/7 (constant - independent of n)
- C(Gn,Un) is a good expander (?(GnUn) 1/1000)
- Proof is by induction on n (base of induction is
the assumption on Ad)
10Inductive definition of groups
- G1 Ad, the alternating group.
- Gn1 Gno Ad (wreath product)
- Gn1 (?,x1,x2,,xd) ?2 Ad,xi 2 G
11Inductive definition of groups
- G1 Ad, the alternating group.
- Gn1 Gno Ad (wreath product)
- Gn1 (?,x1,x2,,xd) ?2 Ad,xi 2 G
- Multiplication rule
- (?,x1,x2,,xd) (?,y1,y2,,yd) (??, x?(1),y1,
x ?(2)y2,, x ?(d)yd)
12Inductive definition of groups
- G1 Ad, the alternating group.
- Gn1 Gno Ad (wreath product)
- Gn1 (?,x1,x2,,xd) ?2 Ad,xi 2 G
- Multiplication rule
- (?,x1,x2,,xd) (?,y1,y2,,yd) (??, x?(1),y1,
x ?(2)y2,, x ?(d)yd) - Example conjugation
- g(?,1,,1) , x (1,x1,,xd)
- g-1 x g(1,x?(1),,x?(d))
13Generating sets
- U1 ½ Ad small generating set
- Un ½ Gn small generating set
- Goal Un1 ½ Gn1 small generating set
14Generating sets
- U1 ½ Ad small generating set
- Un ½ Gn small generating set
- Goal Un1 ½ Gn1 small generating set
- Embed U1 ½ Gn (u,1,1,L,1) u 2 U1
- Generates all (s,1,1,L,1) s 2 Ad
U1 ½ Ad
1 2 Gn
1 2 Gn
1 2 Gn
1 2 Gn
15Generating sets
- Pick some x(1,x1,x2,L,xd) 2 Gn1
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
16Generating sets
- Pick some x(1,x1,x2,L,xd) 2 Gn1
- Conjugating x by (s,1,1,L,1) permutes x by s
- We get all (even) permutations of x.
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
17Generating sets
- Pick some x(1,x1,x2,L,xd) 2 Gn1
- Conjugating x by (s,1,1,L,1) permutes x by s
- We get all (even) permutations of x.
- If the permutations of x generate (Gn)d
- Then U1 x generates Gn1
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
18Generating sets
- Necessary condition x1,x2,L,xd generates Gn
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
19Generating sets
- Necessary condition x1,x2,L,xd generates Gn
- Un generates Gn. Unu1,u2,?,up. Suppose p
divides d - Put x (u1,u1,?,u1, u2,u2,?,u2,?, up,up,?,up)
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
20Generating sets
- Necessary condition x1,x2,L,xd generates Gn
- Un generates Gn. Unu1,u2,?,up. Suppose p
divides d - Put x (u1,u1,?,u1, u2,u2,?,u2,?, up,up,?,up)
- Define Un(d) the orbit of x under Ad.
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
21Generating sets
- Necessary condition x1,x2,L,xd generates Gn
- Un generates Gn. Unu1,u2,?,up. Suppose p
divides d - Put x (u1,u1,?,u1, u2,u2,?,u2,?, up,up,?,up)
- Define Un(d) the orbit of x under Ad.
- Does Un(d) generate (Gn)d ?
- Is it expanding ?
1
x1 2 Gn
x2 2 Gn
xd-1 2 Gn
xd 2 Gn
22Algebraic Zig-Zag theorem ALW
- If
- ?( (Gn)d, Un(d) ) 1/50
- ?(Ad,U1) 1/1000
23Algebraic Zig-Zag theorem ALW
- If
- ?( (Gn)d, Un(d) ) 1/50
- ?(Ad,U1) 1/1000
- Then
- 9 W½ Gn1 ( Gn o Ad)
- ?(Gn1, W) 1/50 1/1000
24Algebraic Zig-Zag theorem ALW
- If
- ?( (Gn)d, Un(d) ) 1/50
- ?(Ad,U1) 1/1000
- Then
- 9 W½ Gn1 ( Gn o Ad)
- ?(Gn1, W) 1/50 1/1000
- W U12 (Important! W is a func. of
U1const.)
25Proof of main theorem
- Induction assumption 9 Un ½ Gn , ?(Gn,Un)
1/1000
26Proof of main theorem
- Induction assumption 9 Un ½ Gn , ?(Gn,Un)
1/1000 - Main Lemma ?( (Gn)d, Un(d) ) 1/50 (for good
G,U)
27Proof of main theorem
- Induction assumption 9 Un ½ Gn , ?(Gn,Un)
1/1000 - Main Lemma ?( (Gn)d, Un(d) ) 1/50 (for good
G,U) - Zigzag thm RVW,ALW
- 9 W½ Gn1 ( Gn o Ad)
- ?(Gn1, W) 1/50 1/1000
- W U12
28Proof of main theorem
- Induction assumption 9 Un ½ Gn , ?(Gn,Un)
1/1000 - Main Lemma ?( (Gn)d, Un(d) ) 1/50 (for good
G,U) - Zigzag thm RVW,ALW
- 9 W½ Gn1 ( Gn o Ad)
- ?(Gn1, W) 1/50 1/1000
- W U12
- Define Un1 W2 all words of length 2 in W
- Un1 U14 d1/7
- ?(Gn1, Un1) (1/50 1/1000)2 1/50
29Main Lemma
- Main Lemma Given
- U ½ G
- ?(G, U) 1/1000
- then
- - ?( (G)d, U(d)) 1/50 (for good G,U)
30Main Lemma
- Main Lemma Given
- U ½ G
- ?(G, U) 1/1000
- then
- - ?( (G)d, U(d)) 1/50 (for good G,U)
- Example
- ?( (G)d, Ud ) 1/1000
- But Ud is NOT one Ad orbit
31Main Lemma
- Main Lemma Given
- U ½ G small enough
- ?(G, U) 1/1000
- then
- - ?( (G)d, U(d)) 1/50 (for good G,U)
- Example
- ?( (G)d, Ud ) 1/1000
- But Ud is NOT one Ad orbit
- The idea when U ltlt d
- A random element of Ud is more or less in U(d).
- So U(d) approximates Ud well.
32Main Lemma-reduction to G G
- When/Why is C( Gd, U(d)) connected ?
33Main Lemma-reduction to G G
- When/Why is C( Gd, U(d)) connected ?
- (u1 , u2 , u3 , L,ud) 2 U(d)
- (u2-1, u1 1 , u3-1 , L,ud-1) 2 U(d)
34Main Lemma-reduction to G G
- When/Why is C( Gd, U(d)) connected ?
- (u1 , u2 , u3 , L,ud) 2 U(d)
- (u2-1, u1 1 , u3-1 , L,ud-1) 2 U(d)
- multiply to get
- (u1, u2-1 , u2 u11 , 1 , L , 1)
35Main Lemma-reduction to G G
- When/Why is C( Gd, U(d)) connected ?
- (u1 , u2 , u3 , L,ud) 2 U(d)
- (u2-1, u1 1 , u3-1 , L,ud-1) 2 U(d)
- multiply to get
- (u1, u2-1 , u2 u11 , 1 , L , 1)
- Setting u2 1 (assume 1 2 U) we generate all
elements - (g , g-1 , 1 , 1, , 1) g 2 U
U1
36Main Lemma-reduction to G G
- When/Why is C( Gd, U(d)) connected ?
- (u1 , u2 , u3 , L,ud) 2 U(d)
- (u2-1, u1 1 , u3-1 , L,ud-1) 2 U(d)
- multiply to get
- (u1, u2-1 , u2 u11 , 1 , L , 1)
- Setting u2 1 (assume 1 2 U) we generate all
elements - (g , g-1 , 1 , 1, , 1) g 2 U
- If first two coordinates generate G G we are
done - Expansion also follows
U1
37Expansion on G G
- X ½ G
- Z ½ G G
- Z (x,x-1) x 2 X (completely correlated!)
38Expansion on G G
- X ½ G
- Z ½ G G
- Z (x,x-1) x 2 X (completely correlated!)
- Suppose ?(G,X) 1-?
- Is ?(G G, Z) 1-f(?) for some f?
39Expansion on G G
- X ½ G
- Z ½ G G
- Z (x,x-1) x 2 X (completely correlated!)
- Suppose ?(G,X) 1-?
- Is ?(G G, Z) 1-f(?) for some f?
- G abelian NO.
- C (G G, Z) is disconnected.
- Connected component of (1,1) is (g,g-1) g 2 G
40Expansion on G G
- X ½ G
- Z ½ G G
- Z (x,x-1) x 2 X (completely correlated!)
- Suppose ?(G,X) 1-?
- Is ?(G G, Z) 1-f(?) for some f?
- G abelian NO.
- C (G G, Z) is disconnected.
- Connected component of (1,1) is (g,g-1) g 2 G
- If Y (x,x) x 2 X then C(G G, Y) is
disconnected
41Expansion on G G decorrelating the gens.
- BUT If for every x 2 G, x ax,bx ax-1 bx-1
ax bx
42Expansion on G G decorrelating the gens.
- BUT If for every x 2 G, x ax,bx ax-1 bx-1
ax bx - Xc ax, bx, (ax-1bx-1) x 2 X (
inverses) - Z ½ G G
- Z (x,x-1) x 2 Xc
43Expansion on G G decorrelating the gens.
- BUT If for every x 2 G, x ax,bx ax-1 bx-1
ax bx - Xc ax, bx, (ax-1bx-1) x 2 X (
inverses) - Z ½ G G
- Z (x,x-1) x 2 Xc
- (ax-1bx-1 , bxax) 2 Z
- (ax , ax-1) 2 Z
- (bx , bx-1) 2 Z
44Expansion on G G decorrelating the gens.
- BUT If for every x 2 G, x ax,bx ax-1 bx-1
ax bx - Xc ax, bx, (ax-1bx-1) x 2 X (
inverses) - Z ½ G G
- Z (x,x-1) x 2 Xc
- (ax-1bx-1 , bxax) 2 Z
- (ax , ax-1) 2 Z
- (bx , bx-1) 2 Z
-
- (x , 1) is generated by Z
45Expansion on G G decorrelating the gens.
- BUT If for every x 2 G, x ax,bx ax-1 bx-1
ax bx - Xc ax, bx, (ax-1bx-1) x 2 X (
inverses) - Z ½ G G
- Z (x,x-1) x 2 Xc
- (ax-1bx-1 , bxax) 2 Z
- (ax , ax-1) 2 Z
- (bx , bx-1) 2 Z
-
- (x , 1) is generated by Z
- If ?(G,X) 1-? then ?(G G, Z) 1-(? /500X2)
46Commutator representation in Gn
- Def a group G has the commutator property (CP) if
every element in G is a commutator.
47Commutator representation in Gn
- Def a group G has the commutator property (CP) if
every element in G is a commutator. - Theorem ORR, 50s Ad has (CP)
48Commutator representation in Gn
- Def a group G has the commutator property (CP) if
every element in G is a commutator. - Theorem ORR, 50s Ad has (CP)
- Theorem N 03 If G has (CP) then G o Ad has
(CP)
49Commutator representation in Gn
- Def a group G has the commutator property (CP) if
every element in G is a commutator. - Theorem ORR, 50s Ad has (CP)
- Theorem N 03 If G has (CP) then G o Ad has
(CP) - Proof Reduce to the system of eqs
- x1 yh(1) xs(1)-1 yt(1) a1
- x2 yh(2) xs(2)-1 yt(2) a2
-
- xd yh(d) xs(d)-1 yt(d) ad
- 2d variables xi 2 G, yi 2 G, d constants ai 2
G. - 3 arbitrary permutations h,s,t 2 Sd
50Expansion on alternating gp. Ad
- Our construction was based on
- Assumption 9 U ½ Ad ( G1) such that
- U d1/30
- ?(Ad,U) 1/1000
51Expansion on alternating gp. Ad
- Our construction was based on
- Assumption 9 U ½ Ad ( G1) such that
- U d1/30
- ?(Ad,U) 1/1000
- For any group G, O(logG) elements suffice AR
- logG O(dlogd) in our case)
52Expansion on alternating gp. Ad
- Our construction was based on
- Assumption 9 U ½ Ad ( G1) such that
- U d1/30
- ?(Ad,U) 1/1000
- For any group G, O(logG) elements suffice AR
- logG O(dlogd) in our case)
- Abelian G needs ?(log G) generators
53Expansion on alternating gp. Ad
- Reasons to believe there are few expanding
generators
54Expansion on alternating gp. Ad
- Reasons to believe there are few expanding
generators - 2 random elements generate Ad
55Expansion on alternating gp. Ad
- Reasons to believe there are few expanding
generators - 2 random elements generate Ad D
- 9 7 elements in Ad such that Cayley graph has
diameter O( logAd ) BL?
56Expansion on alternating gp. Ad
- Reasons to believe there are few expanding
generators - 2 random elements generate Ad D
- 9 7 elements in Ad such that Cayley graph has
diameter O( logAd ) BKL - A conjecture (Aldous) implies O(d) transpositions
suffice for expansion (transpositions are weak!)
57Expansion on alternating gp. Ad
- Reasons to believe there are few expanding
generators - 2 random elements generate Ad D
- 9 7 elements in Ad such that Cayley graph has
diameter O( logAd ) BL? - A conjecture (Aldous) implies O(d) transpositions
suffice for expansion (transpositions are weak!) - A conjecture (Wigderson) implies O(d1/2)
permutations suffice for expansion.
58D