Improved bounds for the sunflower lemma Kewen Wu Peking U UC - - PowerPoint PPT Presentation

improved bounds for the sunflower lemma
SMART_READER_LITE
LIVE PREVIEW

Improved bounds for the sunflower lemma Kewen Wu Peking U UC - - PowerPoint PPT Presentation

Main result Applications Proof overview Open problems Thanks Improved bounds for the sunflower lemma Kewen Wu Peking U UC Berkeley Joint work with Ryan Alweiss Shachar Lovett Jiapeng Zhang Princeton UCSD Harvard USC 1 / 18 Main


slide-1
SLIDE 1

Main result Applications Proof overview Open problems Thanks

Improved bounds for the sunflower lemma

Kewen Wu

Peking U → UC Berkeley

Joint work with Ryan Alweiss Shachar Lovett Jiapeng Zhang Princeton UCSD Harvard → USC

1 / 18

slide-2
SLIDE 2

Main result Applications Proof overview Open problems Thanks

Definitions

Definition (w-set system and r-sunflower) A w-set system is a family of sets of size at most w.

2 / 18

slide-3
SLIDE 3

Main result Applications Proof overview Open problems Thanks

Definitions

Definition (w-set system and r-sunflower) A w-set system is a family of sets of size at most w. An r-sunflower is r sets S1, . . . , Sr where Kernel: Y = S1 ∩ · · · ∩ Sr; Petals: S1 \ Y, . . . , Sr \ Y are pairwise disjoint.

2 / 18

slide-4
SLIDE 4

Main result Applications Proof overview Open problems Thanks

Definitions

Definition (w-set system and r-sunflower) A w-set system is a family of sets of size at most w. An r-sunflower is r sets S1, . . . , Sr where Kernel: Y = S1 ∩ · · · ∩ Sr; Petals: S1 \ Y, . . . , Sr \ Y are pairwise disjoint. Example {{1, 2} , {1, 3, 4, 6} , {1, 5} , {2, 3}} is a 4-set system of size 4. It has a 3-sunflower {{1, 2} , {1, 3, 4, 6} , {1, 5}} with kernel {1} and petals {2} , {3, 4, 6} , {5}.

2 / 18

slide-5
SLIDE 5

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower.

3 / 18

slide-6
SLIDE 6

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower. Let’s focus on r = 3.

3 / 18

slide-7
SLIDE 7

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower. Let’s focus on r = 3. Erd˝

  • s and Rado 1960: s = w! · 2w ≈ ww.

3 / 18

slide-8
SLIDE 8

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower. Let’s focus on r = 3. Erd˝

  • s and Rado 1960: s = w! · 2w ≈ ww.

Kostochka 2000: s ≈ (w log log log w/ log log w)w.

3 / 18

slide-9
SLIDE 9

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower. Let’s focus on r = 3. Erd˝

  • s and Rado 1960: s = w! · 2w ≈ ww.

Kostochka 2000: s ≈ (w log log log w/ log log w)w. Fukuyama 2018: s ≈ w0.75w.

3 / 18

slide-10
SLIDE 10

Main result Applications Proof overview Open problems Thanks

Main result

Theorem (Erd˝

  • s-Rado sunflower)

Any w-set system of size s has an r-sunflower. Let’s focus on r = 3. Erd˝

  • s and Rado 1960: s = w! · 2w ≈ ww.

Kostochka 2000: s ≈ (w log log log w/ log log w)w. Fukuyama 2018: s ≈ w0.75w. Now: s ≈ (log w)w and this is tight for our approach.

3 / 18

slide-11
SLIDE 11

Main result Applications Proof overview Open problems Thanks

Actual bound and further refinement

Theorem (Improved sunflower lemma) For some constant C, any w-set system of size s has an r-sunflower, where s =

  • Cr2 ·
  • log w log log w + (log r)2w .

4 / 18

slide-12
SLIDE 12

Main result Applications Proof overview Open problems Thanks

Actual bound and further refinement

Theorem (Improved sunflower lemma) For some constant C, any w-set system of size s has an r-sunflower, where s =

  • Cr2 ·
  • log w log log w + (log r)2w .

Recently, Anup Rao improved it to s = (Cr(log w + log r)))w .

4 / 18

slide-13
SLIDE 13

Main result Applications Proof overview Open problems Thanks

Applications – Theoretical computer science

Circuit lower bounds Data structure lower bounds Matrix multiplication Pseudorandomness Cryptography Property testing Fixed parameter complexity Communication complexity ...

5 / 18

slide-14
SLIDE 14

Main result Applications Proof overview Open problems Thanks

Applications – Combinatorics

Erd˝

  • s-Szemer´

edi sunflower lemma Intersecting set systems Packing Kneser graphs Alon-Jaeger-Tarsi nowhere-zero conjecture Thersholds in random graphs ...

6 / 18

slide-15
SLIDE 15

Main result Applications Proof overview Open problems Thanks

Section 3 Proof overview

7 / 18

slide-16
SLIDE 16

Main result Applications Proof overview Open problems Thanks

Make it robust

Assume F = {S1, . . . , Sm} is a w-set system.

8 / 18

slide-17
SLIDE 17

Main result Applications Proof overview Open problems Thanks

Make it robust

Assume F = {S1, . . . , Sm} is a w-set system. Define a width-w DNF fF as fF = m

i=1

  • j∈Si xj.

Example If F = {{1, 2} , {1, 3, 4, 6} , {1, 5} , {2, 3}}, then fF = (x1 ∧ x2) ∨ (x1 ∧ x3 ∧ x4 ∧ x6) ∨ (x1 ∧ x5) ∨ (x2 ∧ x3).

8 / 18

slide-18
SLIDE 18

Main result Applications Proof overview Open problems Thanks

Make it robust

Assume F = {S1, . . . , Sm} is a w-set system. Define a width-w DNF fF as fF = m

i=1

  • j∈Si xj.

Example If F = {{1, 2} , {1, 3, 4, 6} , {1, 5} , {2, 3}}, then fF = (x1 ∧ x2) ∨ (x1 ∧ x3 ∧ x4 ∧ x6) ∨ (x1 ∧ x5) ∨ (x2 ∧ x3). Definition (Satisfying system) F is satisfying if Pr [fF(x) = 0] < 1/3 with Pr [xi = 1] = 1/3,

8 / 18

slide-19
SLIDE 19

Main result Applications Proof overview Open problems Thanks

Make it robust

Assume F = {S1, . . . , Sm} is a w-set system. Define a width-w DNF fF as fF = m

i=1

  • j∈Si xj.

Example If F = {{1, 2} , {1, 3, 4, 6} , {1, 5} , {2, 3}}, then fF = (x1 ∧ x2) ∨ (x1 ∧ x3 ∧ x4 ∧ x6) ∨ (x1 ∧ x5) ∨ (x2 ∧ x3). Definition (Satisfying system) F is satisfying if Pr [fF(x) = 0] < 1/3 with Pr [xi = 1] = 1/3, i.e., Pr [∀i ∈ [m], Si ⊂ S] < 1/3 with Pr [xi ∈ S] = 1/3.

8 / 18

slide-20
SLIDE 20

Main result Applications Proof overview Open problems Thanks

Satisfyingness implies sunflower

Assume F is a set system on ground set {1, . . . , n}. Lemma If F is satisfying, then it has 3 pairwise disjoint sets.

9 / 18

slide-21
SLIDE 21

Main result Applications Proof overview Open problems Thanks

Satisfyingness implies sunflower

Assume F is a set system on ground set {1, . . . , n}. Lemma If F is satisfying, then it has 3 pairwise disjoint sets. 3 pairwise disjoint sets is a 3-sunflower with empty kernel.

9 / 18

slide-22
SLIDE 22

Main result Applications Proof overview Open problems Thanks

Satisfyingness implies sunflower

Assume F is a set system on ground set {1, . . . , n}. Lemma If F is satisfying, then it has 3 pairwise disjoint sets. 3 pairwise disjoint sets is a 3-sunflower with empty kernel. Proof. Color x1, . . . , xn to red, green, blue uniformly and independenty.

9 / 18

slide-23
SLIDE 23

Main result Applications Proof overview Open problems Thanks

Satisfyingness implies sunflower

Assume F is a set system on ground set {1, . . . , n}. Lemma If F is satisfying, then it has 3 pairwise disjoint sets. 3 pairwise disjoint sets is a 3-sunflower with empty kernel. Proof. Color x1, . . . , xn to red, green, blue uniformly and independenty. By definition, F contains a purely red (green/blue) set w.p > 2/3.

9 / 18

slide-24
SLIDE 24

Main result Applications Proof overview Open problems Thanks

Satisfyingness implies sunflower

Assume F is a set system on ground set {1, . . . , n}. Lemma If F is satisfying, then it has 3 pairwise disjoint sets. 3 pairwise disjoint sets is a 3-sunflower with empty kernel. Proof. Color x1, . . . , xn to red, green, blue uniformly and independenty. By definition, F contains a purely red (green/blue) set w.p > 2/3. By union bound, F contains one purely red set, one purely green set, and one purely blue set w.p > 0.

9 / 18

slide-25
SLIDE 25

Main result Applications Proof overview Open problems Thanks

Structure vs pseudorandomness

Assume F = {S1, . . . , Sm} , m > κw is a w-set system. Define link FY = {Si\Y | Y ⊂ Si}, which is a (w − |Y |)-set system. Example If F = {{1, 2} , {1, 3, 4} , {1, 5} , {2, 3}}, then F{2} = {{1} , {3}}.

10 / 18

slide-26
SLIDE 26

Main result Applications Proof overview Open problems Thanks

Structure vs pseudorandomness

Assume F = {S1, . . . , Sm} , m > κw is a w-set system. Define link FY = {Si\Y | Y ⊂ Si}, which is a (w − |Y |)-set system. Example If F = {{1, 2} , {1, 3, 4} , {1, 5} , {2, 3}}, then F{2} = {{1} , {3}}. If there exists Y such that |FY | ≥ m/κ|Y | > κw−|Y |, then we can apply induction and find an 3-sunflower in FY .

10 / 18

slide-27
SLIDE 27

Main result Applications Proof overview Open problems Thanks

Structure vs pseudorandomness

Assume F = {S1, . . . , Sm} , m > κw is a w-set system. Define link FY = {Si\Y | Y ⊂ Si}, which is a (w − |Y |)-set system. Example If F = {{1, 2} , {1, 3, 4} , {1, 5} , {2, 3}}, then F{2} = {{1} , {3}}. If there exists Y such that |FY | ≥ m/κ|Y | > κw−|Y |, then we can apply induction and find an 3-sunflower in FY . So induction starts at such F, that |FY |<m/κ|Y | holds for any Y . Lemma Let κ ≥ (log w)O(1). If |FY | < m/κ|Y | holds for any Y , then F is satisfying, which means F has 3 pairwise disjoint sets.

10 / 18

slide-28
SLIDE 28

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system.

11 / 18

slide-29
SLIDE 29

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system. Assume |FY | < m/κ|Y | holds for any Y . ⇐ F is pseudorandom

11 / 18

slide-30
SLIDE 30

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system. Assume |FY | < m/κ|Y | holds for any Y . ⇐ F is pseudorandom Take ≈ 1/√κ-fraction of the ground set as W, and construct a w/2-(multi-)set system F′ from each Si:

11 / 18

slide-31
SLIDE 31

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system. Assume |FY | < m/κ|Y | holds for any Y . ⇐ F is pseudorandom Take ≈ 1/√κ-fraction of the ground set as W, and construct a w/2-(multi-)set system F′ from each Si: Good: If there exists |Sj\W| ≤ w/2 and Sj\W ⊂ Si\W, then put Sj\W into F′; (j may equal i)

11 / 18

slide-32
SLIDE 32

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system. Assume |FY | < m/κ|Y | holds for any Y . ⇐ F is pseudorandom Take ≈ 1/√κ-fraction of the ground set as W, and construct a w/2-(multi-)set system F′ from each Si: Good: If there exists |Sj\W| ≤ w/2 and Sj\W ⊂ Si\W, then put Sj\W into F′; (j may equal i) E.g., Sj\W = {1} , Si\W = {1, 2, 3, 4, 5}.

11 / 18

slide-33
SLIDE 33

Main result Applications Proof overview Open problems Thanks

Randomness preserves pseudorandomness

Let F = {S1, . . . , Sm} be a w-(multi-)set system. Assume |FY | < m/κ|Y | holds for any Y . ⇐ F is pseudorandom Take ≈ 1/√κ-fraction of the ground set as W, and construct a w/2-(multi-)set system F′ from each Si: Good: If there exists |Sj\W| ≤ w/2 and Sj\W ⊂ Si\W, then put Sj\W into F′; (j may equal i) E.g., Sj\W = {1} , Si\W = {1, 2, 3, 4, 5}. Bad: otherwise, we do nothing for Si. Example If F = {{1, 2} , {1, 3} , {2, 3, 4} , {4, 5, 6, 7}} and w = 4, W = {1}, then F′ = {{1,2} , {1,3} , {2, 3, 4} , {4, 5, 6, 7}}.

11 / 18

slide-34
SLIDE 34

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

12 / 18

slide-35
SLIDE 35

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W.

12 / 18

slide-36
SLIDE 36

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i):

12 / 18

slide-37
SLIDE 37

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7}

12 / 18

slide-38
SLIDE 38

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′

12 / 18

slide-39
SLIDE 39

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s

12 / 18

slide-40
SLIDE 40

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s we know Sj ∩ Si = {4, 5, 6}

12 / 18

slide-41
SLIDE 41

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s we know Sj ∩ Si = {4, 5, 6} k = 2

12 / 18

slide-42
SLIDE 42

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s we know Sj ∩ Si = {4, 5, 6} k = 2 Si ranks 2 in F{4,5,6}, we recover i = 4

12 / 18

slide-43
SLIDE 43

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s we know Sj ∩ Si = {4, 5, 6} k = 2 Si ranks 2 in F{4,5,6}, we recover i = 4 aux2 = $$$$

12 / 18

slide-44
SLIDE 44

Main result Applications Proof overview Open problems Thanks

One reduction step

Then |F′

Y | ≤ |FY | and |F′|≈|F|. ⇐ F′ is also pseudorandom

Prove by encoding bad (W, i) → (W ′ = W ∪ Si, aux1, k, aux2), where Si ranks k < |F|/κw/2 in FSj∩Si for the first j ≤ i that Sj\W ⊂ Si\W. Example F′ = {{1,2} , {2, 3, 4} , {1, 4, 5, 6}, {4, 5, 6, 7}} , W = {1} , i = 4. Encode/decode bad pair (W, i): W ′ = W ∪ Si = {1, 4, 5, 6, 7} we find j = 3 with Sj ⊂ W ′ aux1 = ∗$$$ with at least w/2 $s we know Sj ∩ Si = {4, 5, 6} k = 2 Si ranks 2 in F{4,5,6}, we recover i = 4 aux2 = $$$$ we recover W = W ′\ {4, 5, 6, 7}

12 / 18

slide-45
SLIDE 45

Main result Applications Proof overview Open problems Thanks

Reductions

Let F = {S1, . . . , Sm} be a w-(multi-)set system on {1, . . . , n}. Assume |FY | < m/κ|Y | holds for any Y , and κ ≈ (log w)2.

13 / 18

slide-46
SLIDE 46

Main result Applications Proof overview Open problems Thanks

Reductions

Let F = {S1, . . . , Sm} be a w-(multi-)set system on {1, . . . , n}. Assume |FY | < m/κ|Y | holds for any Y , and κ ≈ (log w)2. It suffices to prove F is satisfying ⇐ ⇒ w.h.p S covers some set of F, where Pr [xi ∈ S] = 1/3.

13 / 18

slide-47
SLIDE 47

Main result Applications Proof overview Open problems Thanks

Reductions

Let F = {S1, . . . , Sm} be a w-(multi-)set system on {1, . . . , n}. Assume |FY | < m/κ|Y | holds for any Y , and κ ≈ (log w)2. It suffices to prove F is satisfying ⇐ ⇒ w.h.p S covers some set of F, where Pr [xi ∈ S] = 1/3. Split S into several parts, Pr [xi ∈ S] = 1/3 ≈ take 1/3-fraction of the ground set as S ≈ view S as W1, W2, . . . , Wlog w, each of ≈ 1/√κ-fraction

13 / 18

slide-48
SLIDE 48

Main result Applications Proof overview Open problems Thanks

Reductions

Let F = {S1, . . . , Sm} be a w-(multi-)set system on {1, . . . , n}. Assume |FY | < m/κ|Y | holds for any Y , and κ ≈ (log w)2. It suffices to prove F is satisfying ⇐ ⇒ w.h.p S covers some set of F, where Pr [xi ∈ S] = 1/3. Split S into several parts, Pr [xi ∈ S] = 1/3 ≈ take 1/3-fraction of the ground set as S ≈ view S as W1, W2, . . . , Wlog w, each of ≈ 1/√κ-fraction Then we iteratively apply (pseudorandom-preserving) reductions, F

W1

− − → F′

W2

− − → F′′

W3

− − → · · ·

Wlog w

− − − − → Flast.

13 / 18

slide-49
SLIDE 49

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

14 / 18

slide-50
SLIDE 50

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

14 / 18

slide-51
SLIDE 51

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

W2

− − → F′′

  • width-w/4

14 / 18

slide-52
SLIDE 52

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

W2

− − → F′′

  • width-w/4

W3

− − → · · ·

Wlog w

− − − − → Flast

  • width-0

.

14 / 18

slide-53
SLIDE 53

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

W2

− − → F′′

  • width-w/4

W3

− − → · · ·

Wlog w

− − − − → Flast

  • width-0

. either we stop at Wi when some set is contained in

j<i Wj,

⇒ S contains some set of F

14 / 18

slide-54
SLIDE 54

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

W2

− − → F′′

  • width-w/4

W3

− − → · · ·

Wlog w

− − − − → Flast

  • width-0

. either we stop at Wi when some set is contained in

j<i Wj,

⇒ S contains some set of F

  • r we don’t stop.

Then, Flast is a width-0 (multi-)set system of size ≈ m > κw, and

  • Flast

Y

  • Flast

/κ|Y | still holds for any Y . ⇒ Impossible

14 / 18

slide-55
SLIDE 55

Main result Applications Proof overview Open problems Thanks

Final step

Recall S = W1 ∪ · · · ∪ Wlog w and F

  • width-w

W1

− − → F′

  • width-w/2

W2

− − → F′′

  • width-w/4

W3

− − → · · ·

Wlog w

− − − − → Flast

  • width-0

. either we stop at Wi when some set is contained in

j<i Wj,

⇒ S contains some set of F

  • r we don’t stop.

Then, Flast is a width-0 (multi-)set system of size ≈ m > κw, and

  • Flast

Y

  • Flast

/κ|Y | still holds for any Y . ⇒ Impossible Thus, (informally) we proved such F is satisfying, which means F has an 3-sunflower (3 pairwise disjoint sets).

14 / 18

slide-56
SLIDE 56

Main result Applications Proof overview Open problems Thanks

Section 4 Open problems

15 / 18

slide-57
SLIDE 57

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Rado sunflower

Problem (Erd˝

  • s-Rado sunflower conjecture)

Any w-set system of size Or(1)w has an r-sunflower.

16 / 18

slide-58
SLIDE 58

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Rado sunflower

Problem (Erd˝

  • s-Rado sunflower conjecture)

Any w-set system of size Or(1)w has an r-sunflower. Our approach cannot go beyond (log w)(1−o(1))w.

16 / 18

slide-59
SLIDE 59

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Rado sunflower

Problem (Erd˝

  • s-Rado sunflower conjecture)

Any w-set system of size Or(1)w has an r-sunflower. Our approach cannot go beyond (log w)(1−o(1))w. Lift the sunflower size? r = 3 = ⇒ r = 4.

16 / 18

slide-60
SLIDE 60

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Rado sunflower

Problem (Erd˝

  • s-Rado sunflower conjecture)

Any w-set system of size Or(1)w has an r-sunflower. Our approach cannot go beyond (log w)(1−o(1))w. Lift the sunflower size? r = 3 = ⇒ r = 4. Is (log w)(1−o(1))w actually tight? Counterexamples?

16 / 18

slide-61
SLIDE 61

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Szemer´

edi sunflower

Assume F = {S1, . . . , Sm} and Si ⊂ {1, 2. . . . , n}. Problem (Erd˝

  • s-Szemer´

edi sunflower conjecture) There exists function ε = ε(r) > 0, such that, if m > 2n(1−ε), then F has an r-sunflower.

17 / 18

slide-62
SLIDE 62

Main result Applications Proof overview Open problems Thanks

Erd˝

  • s-Szemer´

edi sunflower

Assume F = {S1, . . . , Sm} and Si ⊂ {1, 2. . . . , n}. Problem (Erd˝

  • s-Szemer´

edi sunflower conjecture) There exists function ε = ε(r) > 0, such that, if m > 2n(1−ε), then F has an r-sunflower. ER sunflower conjecture = ⇒ ES sunflower conjecture. Now:

general r: ε = Or (1/ log n) from ER sunflower. r = 3: Naslund proved it using polynomial method.

17 / 18

slide-63
SLIDE 63

Main result Applications Proof overview Open problems Thanks

Section 5 Thanks

18 / 18