Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal...
Transcript of Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal...
![Page 1: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/1.jpg)
Probabilistic Methods inCombinatorics
Lecture 1
Linyuan Lu
University of South Carolina
Mathematical Sciences Center at Tsinghua University
November 16, 2011 – December 30, 2011
![Page 2: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/2.jpg)
Course information
2 / 29
Textbook: The probabilistic method (3rd edition) by NogaAlon and Joel H. Spencer, publisher: Wiley, 2008.
![Page 3: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/3.jpg)
Course information
2 / 29
Textbook: The probabilistic method (3rd edition) by NogaAlon and Joel H. Spencer, publisher: Wiley, 2008.
Goal: We shall cover some selected topics in the textbookand some supplemental material.
![Page 4: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/4.jpg)
Course information
2 / 29
Textbook: The probabilistic method (3rd edition) by NogaAlon and Joel H. Spencer, publisher: Wiley, 2008.
Goal: We shall cover some selected topics in the textbookand some supplemental material.
Lecture time: (November 16-December 30) Wednesday,Friday, 1:00pm-2:45pm
![Page 5: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/5.jpg)
Course information
2 / 29
Textbook: The probabilistic method (3rd edition) by NogaAlon and Joel H. Spencer, publisher: Wiley, 2008.
Goal: We shall cover some selected topics in the textbookand some supplemental material.
Lecture time: (November 16-December 30) Wednesday,Friday, 1:00pm-2:45pm
Location: Conference room 3, floor 2.
![Page 6: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/6.jpg)
History
3 / 29
Paul Erdos: 1913–19961525 papers511 coauthors
![Page 7: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/7.jpg)
History
3 / 29
Paul Erdos: 1913–19961525 papers511 coauthors
Main contributions:
Ramsey theory Probabilistic method Extremal combinatorics Additive number theory
![Page 8: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/8.jpg)
Notation
4 / 29
A graph G consists of two sets V and E.
- V is the set of vertices (or nodes).- E is the set of edges, where each edge is a pair of
vertices.
Complete graphs Kn:
K 3 K 4 K 5 K 6
![Page 9: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/9.jpg)
Ramsey number R(k, k)
5 / 29
Ramsey number R(k, l): the smallest integer n such thatin any two-coloring of the edges of a complete graph on nvertices Kn by red and blue, either there is a red Kk or ablue Kl.
![Page 10: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/10.jpg)
Ramsey number R(k, k)
5 / 29
Ramsey number R(k, l): the smallest integer n such thatin any two-coloring of the edges of a complete graph on nvertices Kn by red and blue, either there is a red Kk or ablue Kl.
Major question: How large is R(k, k)?
![Page 11: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/11.jpg)
Ramsey number R(k, k)
5 / 29
Ramsey number R(k, l): the smallest integer n such thatin any two-coloring of the edges of a complete graph on nvertices Kn by red and blue, either there is a red Kk or ablue Kl.
Major question: How large is R(k, k)?
Proposition (by Erdos): If(
n2
)
21−(k
2) < 1, thenR(k, k) > n. Thus
R(k, k) >k
e√
22k/2.
![Page 12: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/12.jpg)
Ramsey number R(3, 3) = 6
6 / 29
If edges of K6 are 2-colored then there exists amonochromatic triangle.
![Page 13: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/13.jpg)
Ramsey number R(3, 3) = 6
6 / 29
If edges of K6 are 2-colored then there exists amonochromatic triangle.
There exists a 2-coloring of edges of K5 with nomonochromatic triangle.
![Page 14: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/14.jpg)
Erdos’ idea
7 / 29
To prove R(k, k) > n, we need construct a 2-coloring of Kn
so that it contains no red Kk or blue Kn.
![Page 15: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/15.jpg)
Erdos’ idea
7 / 29
To prove R(k, k) > n, we need construct a 2-coloring of Kn
so that it contains no red Kk or blue Kn.
Make the set of all 2-colorings of Kn into a probabilityspace, then show the event “ no red Kk or blue Kn” withpositive probability.
![Page 16: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/16.jpg)
Probability space
8 / 29
Finite probability space (Ω, P ):
Ω := s1, s2, . . . , sn: a set of n elements.
![Page 17: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/17.jpg)
Probability space
8 / 29
Finite probability space (Ω, P ):
Ω := s1, s2, . . . , sn: a set of n elements.
P : Ω → [0, 1]: a probability measure. View P as avector (p1, p2, . . . , pn), where 0 ≤ pi ≤ 1 and∑n
i=1 pi = 1.
![Page 18: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/18.jpg)
Probability space
8 / 29
Finite probability space (Ω, P ):
Ω := s1, s2, . . . , sn: a set of n elements.
P : Ω → [0, 1]: a probability measure. View P as avector (p1, p2, . . . , pn), where 0 ≤ pi ≤ 1 and∑n
i=1 pi = 1.
An event A: a subset of Ω.
![Page 19: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/19.jpg)
Probability space
8 / 29
Finite probability space (Ω, P ):
Ω := s1, s2, . . . , sn: a set of n elements.
P : Ω → [0, 1]: a probability measure. View P as avector (p1, p2, . . . , pn), where 0 ≤ pi ≤ 1 and∑n
i=1 pi = 1.
An event A: a subset of Ω.
Probability of A: Pr(A) =∑
si∈A pi.
![Page 20: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/20.jpg)
Probability space
8 / 29
Finite probability space (Ω, P ):
Ω := s1, s2, . . . , sn: a set of n elements.
P : Ω → [0, 1]: a probability measure. View P as avector (p1, p2, . . . , pn), where 0 ≤ pi ≤ 1 and∑n
i=1 pi = 1.
An event A: a subset of Ω.
Probability of A: Pr(A) =∑
si∈A pi.
Two events A and B are independent if
Pr(AB) = Pr(A)Pr(B).
![Page 21: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/21.jpg)
Proof of Proposition 1:
9 / 29
Color every edge of Kn independently either red or blue,where each color is equally likely.
![Page 22: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/22.jpg)
Proof of Proposition 1:
9 / 29
Color every edge of Kn independently either red or blue,where each color is equally likely.
For any fixed set R of k vertices, let AR be the event thatall pairs with both ends in R are either all red or all blue.
![Page 23: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/23.jpg)
Proof of Proposition 1:
9 / 29
Color every edge of Kn independently either red or blue,where each color is equally likely.
For any fixed set R of k vertices, let AR be the event thatall pairs with both ends in R are either all red or all blue.
Pr(AR) = 21−(k
2).
![Page 24: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/24.jpg)
Proof of Proposition 1:
9 / 29
Color every edge of Kn independently either red or blue,where each color is equally likely.
For any fixed set R of k vertices, let AR be the event thatall pairs with both ends in R are either all red or all blue.
Pr(AR) = 21−(k
2).
Pr(∨RAR) ≤∑
R
Pr(AR) =
(
n
k
)
21−(k
2) < 1.
![Page 25: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/25.jpg)
Proof of Proposition 1:
9 / 29
Color every edge of Kn independently either red or blue,where each color is equally likely.
For any fixed set R of k vertices, let AR be the event thatall pairs with both ends in R are either all red or all blue.
Pr(AR) = 21−(k
2).
Pr(∨RAR) ≤∑
R
Pr(AR) =
(
n
k
)
21−(k
2) < 1.
Hence Pr(∧RAR) = 1 − Pr(∨RAR) > 0.
![Page 26: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/26.jpg)
Estimation of n
10 / 29
Since(
nk
)
≤ 1e
(
enk
)kfor all n ≥ k ≥ 1, we have
(
n
k
)
21−(k
2) ≤ 1
e
(en
k
)k
21−(k
2)
≤ 2
e
( en
k2(k−1)/2
)k
< 1
provided n ≤ ke√
22k/2.
![Page 27: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/27.jpg)
Estimation of n
10 / 29
Since(
nk
)
≤ 1e
(
enk
)kfor all n ≥ k ≥ 1, we have
(
n
k
)
21−(k
2) ≤ 1
e
(en
k
)k
21−(k
2)
≤ 2
e
( en
k2(k−1)/2
)k
< 1
provided n ≤ ke√
22k/2.
Hence,
R(k, k) >k
e√
22k/2.
![Page 28: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/28.jpg)
How good is the bound?
11 / 29
Erdos [1947]:
R(k, k) > (1 + o(1))1
e√
2k2k/2.
![Page 29: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/29.jpg)
How good is the bound?
11 / 29
Erdos [1947]:
R(k, k) > (1 + o(1))1
e√
2k2k/2.
Spencer [1990]:
R(k, k) > (1 + o(1))1
ek2k/2.
![Page 30: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/30.jpg)
How good is the bound?
11 / 29
Erdos [1947]:
R(k, k) > (1 + o(1))1
e√
2k2k/2.
Spencer [1990]:
R(k, k) > (1 + o(1))1
ek2k/2.
Spencer [1975] (using Lovasz Local Lemma)
R(k, k) > (1 + o(1))
√2
ek2k/2.
![Page 31: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/31.jpg)
Upper bound of R(k, k)
12 / 29
A trivial bound:
R(k, k) ≤(
2k − 2
k − 1
)
.
![Page 32: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/32.jpg)
Upper bound of R(k, k)
12 / 29
A trivial bound:
R(k, k) ≤(
2k − 2
k − 1
)
.
Thomason [1988]:
R(k, k) ≤ k−1/2+c/√
log k
(
2k − 2
k − 1
)
.
![Page 33: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/33.jpg)
Upper bound of R(k, k)
12 / 29
A trivial bound:
R(k, k) ≤(
2k − 2
k − 1
)
.
Thomason [1988]:
R(k, k) ≤ k−1/2+c/√
log k
(
2k − 2
k − 1
)
.
Colon [2009]:
R(k, k) ≤ k−C log k
log log k
(
2k − 2
k − 1
)
.
![Page 34: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/34.jpg)
Diagonal Ramsey Problem
13 / 29
Erdos problems:
$100 for proving the limit limk→∞ R(k, k)1/k exists.
![Page 35: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/35.jpg)
Diagonal Ramsey Problem
13 / 29
Erdos problems:
$100 for proving the limit limk→∞ R(k, k)1/k exists.
$250 for determining the value of limk→∞ R(k, k)1/k if itexists.
![Page 36: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/36.jpg)
Diagonal Ramsey Problem
13 / 29
Erdos problems:
$100 for proving the limit limk→∞ R(k, k)1/k exists.
$250 for determining the value of limk→∞ R(k, k)1/k if itexists.
If limk→∞ R(k, k)1/k exists, then it is between√
2 to 4.
![Page 37: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/37.jpg)
Tournament
14 / 29
V : a set of n players.
![Page 38: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/38.jpg)
Tournament
14 / 29
V : a set of n players.
(x, y) means player x beats y.
![Page 39: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/39.jpg)
Tournament
14 / 29
V : a set of n players.
(x, y) means player x beats y.
Tournament on V : an orientation T = (V, E) ofcomplete graphs on V . For each pair of plays x and y,either (x, y) or (y, x) is in E.
We say T has property Sk if for every set of k players thereis one beats all.
![Page 40: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/40.jpg)
A question
15 / 29
Question (by Schutte): Is there a tournament satisfyingthe property Sk?
![Page 41: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/41.jpg)
A question
15 / 29
Question (by Schutte): Is there a tournament satisfyingthe property Sk?
Theorem (Erdos [1963]) If(
nk
)
(1 − 2−k)n−k < 1, thenthere is a tournament on n vertices that has the property Sk.
![Page 42: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/42.jpg)
A question
15 / 29
Question (by Schutte): Is there a tournament satisfyingthe property Sk?
Theorem (Erdos [1963]) If(
nk
)
(1 − 2−k)n−k < 1, thenthere is a tournament on n vertices that has the property Sk.
Proof: Consider a random tournament on V . For each pairx and y, the choice of (x, y) and (y, x) is equally likely.
![Page 43: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/43.jpg)
A question
15 / 29
Question (by Schutte): Is there a tournament satisfyingthe property Sk?
Theorem (Erdos [1963]) If(
nk
)
(1 − 2−k)n−k < 1, thenthere is a tournament on n vertices that has the property Sk.
Proof: Consider a random tournament on V . For each pairx and y, the choice of (x, y) and (y, x) is equally likely.
K: a fixed subset of size k of V . AK: the event that there is no vertex that beats all the
members of K.
![Page 44: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/44.jpg)
A question
15 / 29
Question (by Schutte): Is there a tournament satisfyingthe property Sk?
Theorem (Erdos [1963]) If(
nk
)
(1 − 2−k)n−k < 1, thenthere is a tournament on n vertices that has the property Sk.
Proof: Consider a random tournament on V . For each pairx and y, the choice of (x, y) and (y, x) is equally likely.
K: a fixed subset of size k of V . AK: the event that there is no vertex that beats all the
members of K.
Pr(AK) = (1 − 2−k)n−k.
![Page 45: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/45.jpg)
Proof continues
16 / 29
Pr(
∨K∈(V
k)AK
)
≤∑
K∈(V
k)
Pr(AK)
=
(
n
k
)
(1 − 2−k)n−k < 1.
Therefore, with positive probability, no event AK occurs;that is, there is a tournament on n vertices that has theproperty Sk.
![Page 46: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/46.jpg)
Estimation of n
17 / 29
Let f(k) denote the minimum possible number of vertices ofa tournament that has the property Sk.On one hand, since
(
nk
)
< (en/k)k and
(1 − 2−k)n−k < 2(n−k)/2k
, we have
f(k) ≤ (1 + o(1)) ln 2 · k2 · 2k.
On the other hand, Szekeres proved
f(k) ≥ c1k2k.
![Page 47: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/47.jpg)
Random variable
18 / 29
(Ω, P ): a probability space.
![Page 48: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/48.jpg)
Random variable
18 / 29
(Ω, P ): a probability space.
X : Ω → R: a random variable.
![Page 49: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/49.jpg)
Random variable
18 / 29
(Ω, P ): a probability space.
X : Ω → R: a random variable.
The expectation of X, denoted by E(X), is defined as
E(X) =∑
v∈Ω
X(v)pv.
![Page 50: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/50.jpg)
Random variable
18 / 29
(Ω, P ): a probability space.
X : Ω → R: a random variable.
The expectation of X, denoted by E(X), is defined as
E(X) =∑
v∈Ω
X(v)pv.
Linearity of expectation:
E(X + Y ) = E(X) + E(Y ).
![Page 51: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/51.jpg)
Dominating set
19 / 29
A dominating set of a graph G = (V, E) is a set U ⊆ Vsuch that vertex v ∈ V − U has at least one neighbor in U .
![Page 52: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/52.jpg)
Dominating set
19 / 29
A dominating set of a graph G = (V, E) is a set U ⊆ Vsuch that vertex v ∈ V − U has at least one neighbor in U .
Theorem: Let G = (V, E) be a graph on n vertices, withminimum degree δ > 1. Then G has a dominating set of at
most 1+ln(δ+1)δ+1 n.
![Page 53: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/53.jpg)
Proof
20 / 29
p ∈ [0, 1]: a probability chosen later.
![Page 54: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/54.jpg)
Proof
20 / 29
p ∈ [0, 1]: a probability chosen later.
X: a random set, whose vertex is picked randomly andindependently with probability p.
![Page 55: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/55.jpg)
Proof
20 / 29
p ∈ [0, 1]: a probability chosen later.
X: a random set, whose vertex is picked randomly andindependently with probability p.
Y := YX : the set of vertices in V − X that do not haveany neighbor in X.
![Page 56: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/56.jpg)
Proof
20 / 29
p ∈ [0, 1]: a probability chosen later.
X: a random set, whose vertex is picked randomly andindependently with probability p.
Y := YX : the set of vertices in V − X that do not haveany neighbor in X.
E(|X|) =∑
v
Pr(v ∈ X) = np.
![Page 57: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/57.jpg)
Proof
20 / 29
p ∈ [0, 1]: a probability chosen later.
X: a random set, whose vertex is picked randomly andindependently with probability p.
Y := YX : the set of vertices in V − X that do not haveany neighbor in X.
E(|X|) =∑
v
Pr(v ∈ X) = np.
E(|Y |) =∑
v
Pr(v ∈ Y )
≤ n(1 − p)δ+1.
![Page 58: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/58.jpg)
continue
21 / 29
Let U = X ∪ YX . The set U is clearly a dominating set.
![Page 59: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/59.jpg)
continue
21 / 29
Let U = X ∪ YX . The set U is clearly a dominating set. Wehave
E(|U |) = E(X) + E(Y )
≤ np + n(1 − p)δ+1
≤ n(p + e−p(δ+1)).
Choose p = ln(δ+1)δ+1 to minimize the upper bound. There is a
dominating set of size at most
1 + ln(δ + 1)
δ + 1n.
![Page 60: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/60.jpg)
Hypergraphs
22 / 29
H = (V, E) is an r-uniform hypergraph (r-graph, for short).
V : the set of vertices E: the set of edges, each edge has cardinality r.
![Page 61: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/61.jpg)
Hypergraphs
22 / 29
H = (V, E) is an r-uniform hypergraph (r-graph, for short).
V : the set of vertices E: the set of edges, each edge has cardinality r.
A 3-uniform loose cycle A 3-uniform tight cycle
![Page 62: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/62.jpg)
Property B problem
23 / 29
We say a r-uniform hypergraph H has property B if thereis a two-coloring of V such that no edge is monochromatic.
![Page 63: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/63.jpg)
Property B problem
23 / 29
We say a r-uniform hypergraph H has property B if thereis a two-coloring of V such that no edge is monochromatic.
Let m(r) denote the minimum possible number of edges ofan r-uniform hypergraph that does not have property B.
![Page 64: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/64.jpg)
Property B problem
23 / 29
We say a r-uniform hypergraph H has property B if thereis a two-coloring of V such that no edge is monochromatic.
Let m(r) denote the minimum possible number of edges ofan r-uniform hypergraph that does not have property B.
Proposition [Erdos (1963)] Every r-uniform hypergraphwith less than 2r−1 edges has property B. Thereforem(r) ≥ 2r−1.
![Page 65: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/65.jpg)
Proof
24 / 29
Let H be an r-uniform hypergraph with less than 2r−1
edges. Color V randomly by two colors. For each edgee ∈ E, let Ae be the event that e is monochromatic.
Pr(Ae) = 21−r.
![Page 66: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/66.jpg)
Proof
24 / 29
Let H be an r-uniform hypergraph with less than 2r−1
edges. Color V randomly by two colors. For each edgee ∈ E, let Ae be the event that e is monochromatic.
Pr(Ae) = 21−r.
Therefore,
Pr (∨e∈EAe) ≤∑
e∈E
Pr(Ae) < 1.
There is a two-coloring without monochromatic edges.
![Page 67: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/67.jpg)
Upper bound
25 / 29
Theorem (Erdos [1964]): m(r) < (1 + o(1))e ln 24 r22r.
![Page 68: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/68.jpg)
Upper bound
25 / 29
Theorem (Erdos [1964]): m(r) < (1 + o(1))e ln 24 r22r.
Proof: Fix V with n points. Let χ be a coloring of V witha points in one color, b = n − a points in the other. LetS ⊂ V be a uniformly selected r-set.
![Page 69: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/69.jpg)
Upper bound
25 / 29
Theorem (Erdos [1964]): m(r) < (1 + o(1))e ln 24 r22r.
Proof: Fix V with n points. Let χ be a coloring of V witha points in one color, b = n − a points in the other. LetS ⊂ V be a uniformly selected r-set. Then
Pr(S is monochromatic under χ) =
(
ar
)
+(
br
)
(
nr
) .
![Page 70: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/70.jpg)
Upper bound
25 / 29
Theorem (Erdos [1964]): m(r) < (1 + o(1))e ln 24 r22r.
Proof: Fix V with n points. Let χ be a coloring of V witha points in one color, b = n − a points in the other. LetS ⊂ V be a uniformly selected r-set. Then
Pr(S is monochromatic under χ) =
(
ar
)
+(
br
)
(
nr
) .
Assume n = 2k is even. Then(
ar
)
+(
br
)
reaches theminimum when a = b = k. Thus
Pr(S is monochromatic under χ) ≥ 2(
kr
)
(
nr
) .
![Page 71: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/71.jpg)
continue
26 / 29
Let p :=2(k
r)(n
r).
![Page 72: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/72.jpg)
continue
26 / 29
Let p :=2(k
r)(n
r).
Pick m r-edges S1, . . . , Sm uniformly and independentlyfrom
(
Vr
)
.
![Page 73: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/73.jpg)
continue
26 / 29
Let p :=2(k
r)(n
r).
Pick m r-edges S1, . . . , Sm uniformly and independentlyfrom
(
Vr
)
.
Let H = (V, E) where E = S1, . . . , Sm.
![Page 74: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/74.jpg)
continue
26 / 29
Let p :=2(k
r)(n
r).
Pick m r-edges S1, . . . , Sm uniformly and independentlyfrom
(
Vr
)
.
Let H = (V, E) where E = S1, . . . , Sm.
For each coloring χ, let Aχ be the event that none of Si aremonochromatic.
Pr(Aχ) ≤ (1 − p)m.
![Page 75: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/75.jpg)
continue
26 / 29
Let p :=2(k
r)(n
r).
Pick m r-edges S1, . . . , Sm uniformly and independentlyfrom
(
Vr
)
.
Let H = (V, E) where E = S1, . . . , Sm.
For each coloring χ, let Aχ be the event that none of Si aremonochromatic.
Pr(Aχ) ≤ (1 − p)m.
Pr(∨χAχ) ≤∑
χ
Pr(Aχ) ≤ 2n(1 − p)m.
![Page 76: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/76.jpg)
continue
27 / 29
Choose m = ⌈n ln 2p ⌉. Then 2n(1 − p)m < 1. There is a
positive probability that H does not have property B.
![Page 77: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/77.jpg)
continue
27 / 29
Choose m = ⌈n ln 2p ⌉. Then 2n(1 − p)m < 1. There is a
positive probability that H does not have property B. Hence,
m(r) ≤ ⌈n ln 2
p⌉.
![Page 78: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/78.jpg)
continue
27 / 29
Choose m = ⌈n ln 2p ⌉. Then 2n(1 − p)m < 1. There is a
positive probability that H does not have property B. Hence,
m(r) ≤ ⌈n ln 2
p⌉.
p =2(
kr
)
(
nr
)
= 21−rr−1∏
i=0
n − 2i
n − i
≈ 21−re−r2/2n.
![Page 79: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/79.jpg)
Optimization
28 / 29
Choose n = r2
2 to minimize n/p.
![Page 80: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/80.jpg)
Optimization
28 / 29
Choose n = r2
2 to minimize n/p. We get
m = ⌈n ln 2
p⌉
≈ (ln 2)2r−1ner2/2n
≈ e ln 2
4r22r.
![Page 81: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/81.jpg)
Optimization
28 / 29
Choose n = r2
2 to minimize n/p. We get
m = ⌈n ln 2
p⌉
≈ (ln 2)2r−1ner2/2n
≈ e ln 2
4r22r.
Hence m(r) < (1 + o(1))e ln 24 r22r.
![Page 82: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/82.jpg)
Property B problem
29 / 29
Beck [1978]:
m(r) ≥ r1/3−ǫ2r.
![Page 83: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/83.jpg)
Property B problem
29 / 29
Beck [1978]:
m(r) ≥ r1/3−ǫ2r.
Radhakrishnan-Srinivasan [2000]: (best lower bound)
m(r) ≥ Ω
(
( r
ln r
)1/2
2r
)
.
![Page 84: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/84.jpg)
Property B problem
29 / 29
Beck [1978]:
m(r) ≥ r1/3−ǫ2r.
Radhakrishnan-Srinivasan [2000]: (best lower bound)
m(r) ≥ Ω
(
( r
ln r
)1/2
2r
)
.
Theorem (Erdos [1964]): (best upper bound)
m(r) < (1 + o(1))e ln 2
4r22r.
![Page 85: Probabilistic Methods in Combinatoricspeople.math.sc.edu/lu/probmethod/probmethod1.pdfExtremal combinatorics Additive number theory. Notation 4 / 29 A graph G consists of two sets](https://reader033.fdocuments.in/reader033/viewer/2022053013/5f1090617e708231d449ba8c/html5/thumbnails/85.jpg)
Property B problem
29 / 29
Beck [1978]:
m(r) ≥ r1/3−ǫ2r.
Radhakrishnan-Srinivasan [2000]: (best lower bound)
m(r) ≥ Ω
(
( r
ln r
)1/2
2r
)
.
Theorem (Erdos [1964]): (best upper bound)
m(r) < (1 + o(1))e ln 2
4r22r.
m(2) = 3, m(3) = 7, 20 ≤ m(4) ≤ 23.