18.07.2014 Views

Discrete Geometry and Extremal Graph Theory - IMSA

Discrete Geometry and Extremal Graph Theory - IMSA

Discrete Geometry and Extremal Graph Theory - IMSA

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Discrete</strong> <strong>Geometry</strong> <strong>and</strong> <strong>Extremal</strong> <strong>Graph</strong> <strong>Theory</strong><br />

Tony Liu, Illinois Mathematics & Science Academy<br />

under the direction of<br />

Dr. László Babai, University of Chicago<br />

Abstract<br />

In 1946, Paul Erdős proposed the following problem: what is the maximum number of unit<br />

distances among n points in the plane? Erdős established an upper bound of cn 3/2 <strong>and</strong> a lower<br />

bound that grows slightly faster than n.<br />

A graph is a collection of “vertices” <strong>and</strong> “edges” connecting some pairs of vertices. For example,<br />

in an airline route chart, airports are the vertices <strong>and</strong> direct connections correspond to<br />

the edges. We examine graphs without a 4-cycle, <strong>and</strong> show the maximum number of edges<br />

to be at most cn 3/2 . Replacing the 4-cycle by the “Θ-graph,” we obtain an upper bound of<br />

c ′ n 3/2 on the number of edges. Finally, we use this seemingly unrelated graph theoretic result<br />

to establish Erdős’ upper bound on the number of unit distances.<br />

Although we present the optimal solution, within a constant factor, to the graph theory<br />

problem, Erdős’ question on unit distances remains wide open after sixty years.<br />

A cycle (also known as a circuit) is a subset of the edge set of a graph G that forms a<br />

path such that the first node of the path corresponds to the last. A 4-cycle is merely a cycle<br />

with 4 vertices, as shown below.<br />

1


We begin by proving the following.<br />

Theorem 1. A graph G on n vertices without a 4-cycle contains at most cn 3/2 edges for<br />

some constant c > 0.<br />

Proof of Theorem 1. Let d 1 , d 2 , . . . , d n be the degrees of the vertices v 1 , v 2 , . . . , v n of G.<br />

Let E be the number of edges of G, so 2E = ∑ n<br />

i=1 d i. For a pair of vertices v 1 <strong>and</strong> v 2 , note<br />

that there can be at most one other vertex v 3 connected to both v 1 <strong>and</strong> v 2 . Otherwise, for<br />

if v 4 was also joined to v 1 <strong>and</strong> v 2 , we would have a 4-cycle with vertices v 1 , v 3 , v 2 , v 4 . Thus,<br />

more generally, for a vertex v i connected to d i other vertices, we count every pair of these d i<br />

vertices. If we repeat this count for all the vertices v 1 , v 2 , . . . , v n , note that our total count<br />

must less than the total number of pairs of vertices, ( n<br />

2)<br />

. This is because every pair of some<br />

d i vertices connected to v i can be uniquely associated with v i . In other words, we have the<br />

following inequality.<br />

( )<br />

d1<br />

+<br />

2<br />

(<br />

d2<br />

) ( ) (<br />

dn n<br />

+ · · · ≤ .<br />

2 2 2)<br />

We claim that ( d 1<br />

) (<br />

2 +<br />

d2<br />

) (<br />

2 + · · ·<br />

dn<br />

)<br />

2 ≥<br />

2E 2<br />

− E, so it will follow that<br />

n<br />

( )<br />

2E 2 n<br />

n − E ≤ n(n − 1)<br />

=<br />

2 2<br />

⇒ 4E 2 − 2En ≤ n 2 (n − 1).<br />

Thus, ( )<br />

2E − n 2<br />

2 ≤ n 2 (n − 1) + n2 ≤ 4 n3 , so we have 2E ≤ n 3/2 + n . Finally, we conclude<br />

2<br />

that E ∼ < n 3/2 (<strong>and</strong> actually, c ≈ 1 will suffice for sufficiently large n).<br />

2<br />

Now, we give two proofs of the inequality ( d 1<br />

) (<br />

2 +<br />

d2<br />

) (<br />

2 + · · · +<br />

dn<br />

)<br />

2 ≥<br />

2E 2<br />

− E. n<br />

• First Proof. By the Quadratic-Arithmetic Mean Inequality (or Cauchy-Schwartz), we<br />

have<br />

( )<br />

d1<br />

+<br />

2<br />

( )<br />

d2<br />

+ · · · +<br />

2<br />

( )<br />

dn<br />

= 1 2 2 (d2 1 + d 2 2 + · · · + d 2 n) − 1 2 (d 1 + d 2 + · · · + d n )<br />

≥ 1<br />

2n (2E)2 − E.<br />

• Second Proof. As a function, f(x) = ( )<br />

x<br />

2 =<br />

x(x−1)<br />

is convex, so by Jensen’s inequality,<br />

2<br />

we have ( ) ( ) ( ) 2E<br />

)<br />

d1 d2<br />

dn<br />

+ + · · · + ≥ n(<br />

n<br />

= 2E2<br />

2 2<br />

2 2 n − E.<br />

This concludes our proof. <br />

The “Θ-graph” is the union of three internally disjoint (simple) paths that have the same<br />

two end vertices, as shown below.<br />

2


We prove the following result using a similar technique.<br />

Theorem 2. A graph G on n vertices without a “Θ-graph” contains at most c ′ n 3/2 edges<br />

for some constant c ′ > 0.<br />

Proof of Theorem 2 (sketch). As in the proof of Theorem 1, let d 1 , d 2 , . . . , d n be the degrees<br />

of the vertices v 1 , v 2 , . . . , v n . For a vertex v i , we count the pairs of the d i vertices<br />

connected to v i . As we continue this count for all the vertices of G, note that each pair is<br />

counted at most twice, or else G must have a “Θ-graph.” In other words, we arrive at the<br />

inequality<br />

( ) ( ) ( ) (<br />

Create PDF with GO2PDF for free, if you d1 wish to d2 remove this line, dn click here n to buy Virtual PDF Printer<br />

2)<br />

2<br />

+<br />

2<br />

+ · · ·<br />

Thus, as in the proof of Theorem 1, we see c ′ ≈ √ 2c will work for sufficiently large n. <br />

Finally, we prove our main result, showing an unexpected connection between Erdős’ problem<br />

on unit distances <strong>and</strong> graph theoretic results.<br />

Theorem 3. Let f(n) be the maximum number of unit distances among n points in the<br />

plane. Then, f(n) ≤ c ′ n 3/2 .<br />

Proof of Theorem 3. Consider the unit distance graph of the n points defined as follows: two<br />

points are joined by a segment if <strong>and</strong> only if they are unit distance apart. Now, note that<br />

no “Θ-graph” can appear in such a graph, for let us consider the two end vertices. The two<br />

unit circles centered at these vertices must each pass through three common points, but two<br />

circles can intersect in at most two points, contradiction. Thus, by Theorem 2, the result<br />

follows. <br />

2<br />

≤ 2<br />

.<br />

3


Other topics I have explored with my mentor include the following problems.<br />

1. Let F n be the Fibonacci sequence defined by F 1 = F 2 = 1 <strong>and</strong> F n+2 = F n+1 + F n for<br />

n ≥ 1. Show that gcd(F m , F n ) = F gcd(m,n) .<br />

Proof. First we note the following (which are easily proven by induction)<br />

• gcd(F n+1 , F n ) = 1<br />

• F m+n = F m+1 F n + F m F n−1 = F m F n+1 + F m−1 F n<br />

• n | m ⇒ F n | F m<br />

For instance, the first two can be done with straightforward induction while the third<br />

follows from induction <strong>and</strong> setting m = (k − 1)n, where m = kn. Now, without loss of<br />

generality, m ≥ n <strong>and</strong> let m = nq + r where 0 ≤ r < n. We have<br />

gcd(F m , F n ) = gcd(F nq+r , F n ) = gcd(F nq+1 F r + F nq F r+1 , F n ) = gcd(F r , F n )<br />

because F n | F nq <strong>and</strong> gcd(F nq+1 , F nq ) = 1. Now, we can write n = rq 1 + r 1 <strong>and</strong><br />

continue this procedure by writing r n = r n+1 q n+2 + r n+2 where r = r 0 , n = r −1 , <strong>and</strong><br />

0 ≤ r n+2 < r n+1 . Eventually, we will end up at r k = gcd(F m , F n ) <strong>and</strong> r k+1 = 0 for<br />

some k, at which point we have<br />

gcd(F m , F n ) = gcd(F rk , F rk+1 ) = gcd(F gcd(m,n) , 0) = F gcd(m,n)<br />

2. Show that ∏ p≤x p ≤ 4x where p is a prime.<br />

Proof. The base cases x ≤ 3 are easily verified. We may assume that x > 3 is<br />

an integer, as we will prove the result by induction. Note that for odd x, the even<br />

integer x + 1 clearly cannot be prime so we have<br />

∏<br />

p ≤ 4 x < 4 x+1<br />

p≤x+1<br />

p = ∏ p≤x<br />

so the result holds for x + 1 too. Now, for odd x = 2k + 1 where k > 1, note that<br />

primes k + 2 ≤ p ≤ 2k + 1 divide ( )<br />

2k+1<br />

k =<br />

(2k+1)!<br />

as they divide the numerator but<br />

k!(k+1)!<br />

not the denominator. Thus,<br />

∏<br />

( ) 2k + 1<br />

p ≤ = 1 (( ) ( ))<br />

2k + 1 2k + 1<br />

+<br />

< 1 k 2 k k + 1 2 (1 + 1)2k+1 = 4 k<br />

k+2≤p≤2k+1<br />

By the induction hypothesis, we have<br />

∏<br />

<strong>and</strong> multiplying these yields<br />

∏<br />

p≤k+1<br />

p<br />

p≤k+1<br />

∏<br />

k+2≤p≤2k+1<br />

p ≤ 4 k+1<br />

p =<br />

∏<br />

p≤2k+1<br />

p ≤ 4 2k+1<br />

4


3. Let R m (n) be the number of subsets of an n element set, with cardinality divisible by<br />

m. In other words, let<br />

⌊n/m⌋<br />

∑<br />

( n<br />

R m (n) =<br />

.<br />

mk)<br />

Show that R m (n) = 1 m<br />

k=0<br />

∑ m−1<br />

k=0 (1 + zk ) n where z is a primitive m-th root of unity.<br />

Proof. First, we note that 0 = z m − 1 = (z − 1)(z m−1 + z m−2 + · · · + 1) <strong>and</strong> z ≠ 1 so<br />

z m−1 + z m−2 + · · · + 1 = 0. Directly exp<strong>and</strong>ing, we have<br />

m−1<br />

1 ∑<br />

m<br />

k=0<br />

(1 + z k ) n = 1 m<br />

= 1 m<br />

m−1<br />

∑<br />

k=0<br />

n∑<br />

j=0<br />

n∑<br />

j=0<br />

( n<br />

j)<br />

z kj<br />

( m−1<br />

n ∑<br />

z<br />

j) kj<br />

k=0<br />

Now, note that ∑ m−1<br />

k=0 zkj equals m if m | j <strong>and</strong> 0 otherwise. Indeed, if m | j, then<br />

z j = 1 <strong>and</strong> the sum is clearly m. Otherwise, let g = gcd(m, j) <strong>and</strong> l = m g<br />

> 1 so<br />

m | lj. Then, z lj = 1 so z j is a primitive l-th root of unity. In particular, z lj − 1 = 0<br />

<strong>and</strong> z j ≠ 1 imply z j(l−1) + z j(l−2) + · · · + 1 = 0 <strong>and</strong> we have<br />

Thus, our sum becomes<br />

as claimed.<br />

R m (n) = 1 m<br />

4. Show that |R 3 (n) − 2n 3 | ≤ 2 3 .<br />

m−1<br />

∑<br />

k=0<br />

∑l−1<br />

z kj = gz kj = 0<br />

∑<br />

0≤j≤n,m|j<br />

k=0<br />

( n<br />

m =<br />

j)<br />

⌊n/m⌋<br />

∑<br />

k=0<br />

( n<br />

mk)<br />

Proof. Note that by the previous problem, we have R 3 (n) = 1 3 (2n +(1+ω) n +(1+ω 2 ) n )<br />

where ω 3 = 1 <strong>and</strong> ω ≠ 1. Thus, it suffices to prove that |(1 + ω) n + (1 + ω 2 ) n | ≤ 2.<br />

However, this follows immediately from the triangle inequality, as |1 + ω| = | − ω 2 | = 1<br />

<strong>and</strong> |1 + ω 2 | = | − ω| = 1, so<br />

|(1 + ω) n + (1 + ω 2 )| ≤ |(1 + ω) n | + |(1 + ω 2 ) n | = |(1 + ω)| n + |(1 + ω 2 )| n = 2.<br />

Comments. This can be generalized to a (much weaker) bound<br />

∣ R m(n) − 2n<br />

m ∣ ≤ (2 cos π m )n for m ≥ 4<br />

5


<strong>and</strong> the proof is pretty much analogous to the one above. By the formula in the previous<br />

problem, it suffices to prove | ∑ m−1<br />

k=1 (1 + zk ) n | ≤ (2 cos π m )n where z is a primitive<br />

m-th root of unity. Note that all the points 1 + z k must lie within the circle centered<br />

at the origin with radius 2 cos π . Thus, |(1 + m zk ) n | = |1 + z k | n ≤ 2 cos π m )n , <strong>and</strong> the<br />

result follows as before from the triangle inequality.<br />

5. A graph G is bipartite if <strong>and</strong> only if it contains no odd cycle.<br />

Proof: If G = G 1 ∪ G 2 is bipartite, where G 1 <strong>and</strong> G 2 are disjoint subgraphs of G,<br />

then it clearly contains no odd cycle. This is because two vertices are neighbors if <strong>and</strong><br />

only if they lie in different subsets, so a path must alternate between vertices in G 1<br />

<strong>and</strong> G 2 . If G has no odd cycles, let us pick a vertex v <strong>and</strong> color it red. We color all<br />

neighbors of v blue, <strong>and</strong> all uncolored neighbors of these blue vertices red, <strong>and</strong> so on.<br />

We might as well assume that G is connected, so all vertices have a color. We claim<br />

that setting G 1 <strong>and</strong> G 2 to contain all red <strong>and</strong> blue vertices, respectively, shows that<br />

G is bipartite. Assume otherwise, <strong>and</strong> say that v 1 , v 2 are neighbors of the same color.<br />

This means that there exist paths from v to v 1 <strong>and</strong> v 2 with lengths of the same parity.<br />

In other words, the path from v 1 to v 2 through v must have an even length. (If the<br />

paths from v 1 to v <strong>and</strong> v 2 to v first meet at v 3 , then we may replace v with v 3 <strong>and</strong> the<br />

proof continues.) Now, this path from v 1 to v 2 with the edge joining v 1 <strong>and</strong> v 2 creates<br />

an odd cycle, contradiction. Thus, G is bipartite if <strong>and</strong> only if it contains no odd cycle.<br />

6. Let G be a graph on n vertices without K k , a complete graph on k vertices. Then, G<br />

(k − 2)n2<br />

has at most edges. Moreover, there exists a graph with this number of edges.<br />

2(k − 1)<br />

Proof: I do admit to having seen this before, although I only vaguely remembered<br />

the ideas behind this proof of Turán’s Theorem.<br />

Let E(G) denote the number of edges of G. The proof proceeds by induction on<br />

n. Our base cases, for instance n = 1, 2, 3 are trivially true. Now, let G be a graph<br />

on n vertices without K k containing the maximal number of edges. Note that G must<br />

contain a K k−1 or else we could add edges, contradiction the maximality of E(G). Let<br />

H 1 be a K k−1 <strong>and</strong> let H 2 be G without H 1 . Note that E(H 1 ) = ( )<br />

k−1<br />

2 . By the induction<br />

hypothesis, we also have<br />

E(H 2 ) ≤<br />

(k − 2)(n − k + 1)2<br />

.<br />

2(k − 1)<br />

If v ∈ H 2 , then v is neighbors with at most k − 2 vertices in H 1 or else we have a K k .<br />

This implies that the number of edges with one vertex in each H 1 <strong>and</strong> H 2 is at most<br />

(n − k + 1)(k − 2), so<br />

( ) k − 1<br />

E(G) ≤ +<br />

2<br />

(k − 2)(n − k + 1)2<br />

2(k − 1)<br />

+ (n − k + 1)(k − 2) =<br />

(k − 2)n2<br />

2(k − 1) .<br />

Now, let n = (k −1)q +r where 0 ≤ r < k −1. Consider a partition G = G 1 ∪· · ·∪G k−1<br />

where r of the subgraphs have q + 1 vertices <strong>and</strong> the rest have q vertices. We join two<br />

6


vertices if <strong>and</strong> only if they lie in different G i . After a bit of computation, this shows<br />

that the bound above is essentially sharp.<br />

7. We prove a few results regarding even <strong>and</strong> odd permutations.<br />

Lemma 1: Every permutation σ ∈ S n can be written as a product of transpositions.<br />

Proof: Write σ = c 1 c 2 · · · c m where the c k are (possibly trivial) disjoint cycles. It<br />

suffices to show that any cycle can be written as a product of transpositions. Let<br />

c = (i 1 i 2 · · · i k ) be a cycle. We show that c can be written as a product of transpositions<br />

by induction on k. For k = 1, 2, the result is trivial. Now, note that<br />

c = (i 1 i 2 · · · i k ) = (i 1 i 2 · · · i k−1 )(i k−1 i k ) <strong>and</strong> (i 1 i 2 · · · i k−1 ) can be written as a product<br />

of transpositions by the induction hypothesis. The result follows.<br />

Lemma 2: Let τ 1 , τ 2 , . . . , τ m ∈ S n be transpositions. Then, the number of cycles<br />

in τ 1 τ 2 · · · τ m has the same parity as m + n.<br />

Proof: We first prove the following: if σ ∈ S n is a permutation <strong>and</strong> τ ∈ S n is a<br />

transposition, then the number of disjoint cycles in σ <strong>and</strong> στ have different parity.<br />

As before, write σ = c 1 c 2 · · · c m where the c k are (possibly trivial) disjoint cycles <strong>and</strong><br />

suppose τ = (ij). First we deal with the case where i <strong>and</strong> j are in the same cycle, say<br />

c t . If σ(i) = j (or vice versa), then c t = (ij) or (iji 1 · · · i s ). If c t = (ij), then we may<br />

replace c t with (i)(j) in στ. If c t = (iji 1 · · · i s ), then we replace c t with (ii 1 · · · i s )(j) In<br />

either case, the parity of the number of disjoint cycles changes.<br />

If i <strong>and</strong> j are in different cycles, say c t <strong>and</strong> c u respectively, we argue as follows. First,<br />

if c t = (i) <strong>and</strong> c u = (j), then στ is the product of all cycles c i <strong>and</strong> (ij) except for c t<br />

<strong>and</strong> c u . If c t = (i) <strong>and</strong> c u = (jj 1 · · · j v )) (or vice versa), then in στ, replace c u with<br />

(jij 1 · · · j v ) <strong>and</strong> remove (ij). If c t = (ii 1 · · · i s ) <strong>and</strong> c u = (jj 1 · · · j v ), then στ can be<br />

rewriten with c ′ t,u = (ij 1 · · · j v ji 1 · · · i s ) <strong>and</strong> (ij), c t , c u removed. Again, in either case,<br />

the parity changes.<br />

Now we prove Lemma 2 by induction on m, <strong>and</strong> our base cases are easily verified.<br />

We have τ 1 τ 2 · · · τ m+1 = στ m+1 , where we set σ = τ 1 τ 2 · · · τ m . The number of disjoint<br />

cycles in σ <strong>and</strong> στ m+1 have different parity, so this completes our induction.<br />

Comment: It immediately follows that a permutation σ can be written as a product<br />

of an even number of transpositions or an odd number of transpositions, but not<br />

both.<br />

7


References<br />

[1] I. Anderson, Combinatorics of Finite Sets, Oxford University Press, Oxford, Engl<strong>and</strong>,<br />

1989.<br />

[2] G. E. Andrews, Number <strong>Theory</strong>, W. B. Saunders Company, Philadelphia, 1971.<br />

[3] D. Reinhard, <strong>Graph</strong> <strong>Theory</strong>, Springer-Verlag, Heidelberg, July 2005.<br />

[4] L. Lovasz, Combinatorial Problems <strong>and</strong> Exercises, Akademiai Kiado, Budapest, Hungary,<br />

1993.<br />

8

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!