Transcript Document

Polynomial
Partitioning
By Or Yarnitzky
1
Introduction
β€’ Based on β€œSimple Proofs of Classical Theorem in Discrete
Geometry” by Kaplan, Matousek and Sharir, arXiv:
1102.5391, 2011.
2
Introduction
Distinct Distances Problem:
β€’ Given a set 𝑃 of 𝑛 points in the plane, how many distinct
distances are there between them?
β€’ We have seen the bound Ξ©(𝑛6 7 )
β€’ We will see next week Ξ©(𝑛 log(𝑛))
3
Introduction
β€’ In 2010, Guth and Katz obtained a nearly complete
solution to this problem using the Polynomial
Partitioning Theorem, which allowed them to obtain a
partition of the set P with favorable properties.
β€’ Today we will prove the theorem and use it to prove
various bounds.
4
Incidences
β€’ Given a line 𝑙 and a point 𝑝 in the plane ℝ2 , we say they
are incident to each other if the point 𝑝 lies on 𝑙.
β€’ Given a set of points 𝑃 and a set of lines (or more
generally any type of curves) 𝐿, denote the number of
incidences between them by 𝐼 𝑃, 𝐿 .
5
Theorem 1 - Szemerédi-Trotter
β€’ For sets 𝑃 of π‘š points and 𝐿 of 𝑛 lines we proved the
following tight upper bound on 𝐼 𝑃, 𝐿 :
𝐼 𝑃, 𝐿 = 𝑂 π‘š2 3 𝑛2
3
+π‘š+𝑛
β€’ First proof was based on the crossing lemma for simple
graphs.
β€’ Second proof used cuttings and the KST (KΕ‘vari-SósTurán) bound on 𝐼 π‘š, 𝑛
6
Theorem 2
β€’ Let 𝑏, π‘˜, 𝐢 be constants. Given a set 𝑃 of π‘š points and Ξ“
a set of 𝑛 algebraic curves (defined later) such that:
β€’ Every 𝛾 ∈ Ξ“ is of a degree at most 𝑏
β€’ For every π‘˜ different points in 𝑃 there exist at most 𝐢 distinct
curves in Ξ“ passing through all of them
β€’ Then 𝐼 𝑃, Ξ“ = 𝑂(π‘šπ‘˜ (2π‘˜βˆ’1) 𝑛(2π‘˜βˆ’2) (2π‘˜βˆ’1) + π‘š + 𝑛)
with constant proportional to 𝑏, π‘˜, 𝐢.
β€’ Special case of β€œπ‘˜ degrees of freedom” presented in
previous lectures.
7
Geometric Graphs
β€’ Let 𝑃 be a finite set of points. A geometric graph on 𝑃 is
a graph 𝐺 whose vertex set is 𝑃 and edges are straight
segments in the plane.
β€’ Crossing number of 𝐺 is the maximal number of edges a
single line not passing through 𝑃 and not coinciding with
edges of 𝐺 intersects.
β€’ This is not the crossing number we have seen.
8
Theorem 3
β€’ Every set of 𝑛 points in ℝ2 has a geometric spanning tree
with crossing number 𝑂( 𝑛).
𝑛
𝑛
9
Lemmas about Polynomials
β€’ Bivariate polynomials: 𝑓 =
𝑖,𝑗 π‘Žπ‘–,𝑗 π‘₯
𝑖𝑦𝑗
∈ ℝ[π‘₯, 𝑦]
β€’ Denote:
𝑍 𝑓 = π‘₯, 𝑦 ∈ ℝ2 𝑓 π‘₯, 𝑦 = 0}
deg 𝑓 = max 𝑖 + 𝑗 π‘Žπ‘–,𝑗 β‰  0}
β€’ The analysis can be extended to 𝑑-variate polynomials.
10
Lemma 1
β€’ A nonzero bivariate polynomial (with at least one
nonzero coefficient) does not vanish on all ℝ2
Proof:
β€’ Let 𝑓 ∈ ℝ[π‘₯, 𝑦] be a nonzero polynomial. 𝑓 is also a
polynomial in 𝑦 with coefficients from ℝ π‘₯ (rational
functions in π‘₯)
β€’ ℝ π‘₯ is a infinite field, therefore there is a constant 𝑏
such that 𝑓 π‘₯, 𝑏 β‰’ 0 (here 0 is the zero function)
β€’ 𝑓(π‘₯, 𝑏) is a nonzero polynomial over π‘₯ and therefore
there is a constant π‘Ž such that 𝑓 π‘Ž, 𝑏 β‰  0
11
Lemma 2
β€’ If 𝑙 is a line and 𝑓 ∈ ℝ π‘₯, 𝑦 is of degree 𝐷, then:
β€’ 𝑙 βŠ† 𝑍(𝑓)
or
β€’ 𝑍 𝑓 βˆ©π‘™ ≀𝐷
Proof:
𝑙 = 𝑣 + 𝑑𝑒 𝑑 ∈ ℝ
𝑍(𝑓)
𝑒
𝑣
12
Lemma 2
β€’ If 𝑙 is a line and 𝑓 ∈ ℝ π‘₯, 𝑦 is of degree 𝐷, then:
β€’ 𝑙 βŠ† 𝑍(𝑓)
𝑒
or
𝑣
β€’ 𝑍 𝑓 βˆ©π‘™ ≀𝐷
Proof:
β€’ Write 𝑙 in a parametric form 𝑣 + 𝑑𝑒 𝑑 ∈ ℝ , 𝑣, 𝑒 ∈ ℝ2 .
β€’ 𝑍 𝑓 ∩ 𝑙 is the set of roots of the univariate polynomial
𝑔 𝑑 ≔ 𝑓 𝑣 + 𝑑𝑒 . Hence:
β€’ 𝑔 ≑ 0 β‡’ 𝑙 βŠ† 𝑍(𝑓)
or
β€’ 𝑍 𝑓 ∩ 𝑙 = 𝑍 𝑔 ≀ deg 𝑔 ≀ deg 𝑓 = 𝐷
13
Lemma 3
β€’ If 𝑓 ∈ ℝ π‘₯, 𝑦 is nonzero and of degree 𝐷, then 𝑍(𝑓)
contains at most 𝐷 different lines
Proof: By Lemma 1, there is a 𝑝 ∈ ℝ2 βˆ– 𝑍(𝑓).
β€’ Suppose 𝑍(𝑓) contains π‘˜ distinct lines 𝑙1 , … , π‘™π‘˜ .
β€’ Let 𝑙 be a line passing through 𝑝 which is not parallel to
any 𝑙𝑖 and does not pass through any 𝑙𝑖 ∩ 𝑙𝑗
β€’ There is such 𝑙 because only a finite number of directions needed
to be avoided
β€’ 𝑙 has exactly π‘˜ intersections with
π‘˜
𝑖=1 𝑙𝑖
βŠ†π‘ 𝑓 .
β€’ 𝑙 ⊈ 𝑍 𝑓 . By Lemma 2 it has at most 𝐷 intersections with
𝑍(𝑓), therefore π‘˜ ≀ 𝐷
14
Harnack’s Curve Theorem
β€’ For the proof of Theorem 3 (Geometric Spanning Trees
with 𝑂( 𝑛) crossing number), we will need the following
result:
β€’ Let 𝑓 ∈ ℝ π‘₯, 𝑦 , deg 𝑓 = 𝐷.
β€’ Then the number of path-connected components of
𝑍(𝑓) is 𝑂 𝐷2 .
β€’ The tight bound is 1 +
Example:
𝑛
𝑖=1
π‘₯ βˆ’ π‘₯𝑖
2
+
𝑛
𝑖=1
π·βˆ’1
2
𝑦 βˆ’ 𝑦𝑖
deg 𝑓 = 2𝑛,
components
𝑛2
2
(we will not prove it).
βˆ’ πœ€2 = 0
connected
15
(Weak) Bézout's Theorem
β€’ To prove Harnack’s Theorem, we will use without proof
the following result:
Let 𝑓, 𝑔 ∈ ℝ π‘₯, 𝑦 be polynomials of degrees 𝐷𝑓 , 𝐷𝑔 . Then:
β€’ If the system 𝑓 = 𝑔 = 0 has a finite number of solutions
π‘˜, β‡’ π‘˜ ≀ 𝐷𝑓 𝐷𝑔
β€’ If the system 𝑓 = 𝑔 = 0 has a infinitely many solutions
β‡’ 𝑓 and 𝑔 have a nontrivial common factor (not a
constant).
β€’ Bézout's Theorem is a generalization to Lemma 2
β€’ Note that for two polynomials 𝑓|𝑔 ⇔ 𝑍 𝑓 βŠ† 𝑍(𝑔)
16
Lemma 4
β€’ Let 𝑓 ∈ 𝑅[π‘₯, 𝑦] be a square-free polynomial, then 𝑓, 𝑓𝑦 =
πœ•π‘“
πœ•π‘¦
have no common nontrivial factor (assuming 𝑓𝑦 β‰’ 0)
Proof:
β€’ Proof by induction on the degree 𝐷 of 𝑓. Assume that
𝑓 = β„Ž β‹… 𝑔 and 𝑓𝑦 = β„Ž β‹… π‘˜ for polynomials β„Ž, 𝑔, π‘˜ where
β„Ž, 𝑔, π‘˜ are not trivial.
β€’ 𝑓𝑦 = β„Žπ‘¦ β‹… 𝑔 + β„Ž β‹… 𝑔𝑦 = β„Ž β‹… π‘˜, so β„Ž divides β„Žπ‘¦ β‹… 𝑔. By
induction β„Ž, β„Žπ‘¦ have no common factors so β„Ž divides g, a
contradiction because 𝑓 is square free
17
Harnack’s Theorem Proof
β€’ We can assume that 𝑓 is square free, otherwise we
remove repeating factors without changing 𝑍(𝑓).
β€’ For every bounded component assign the rightmost
point. We can assume it exists and unique.
β€’ Such points satisfy the equation 𝑓 = 𝑓𝑦 = 0 (Implicit
Function Theorem), but because of Lemma 4 and
Bézout's Theorem this equation has at most 𝐷(𝐷 βˆ’ 1)
different solutions.
β€’ Therefore, there are at most 𝐷 𝐷 βˆ’ 1 bounded
components of 𝑍(𝑓) 𝑦
18
π‘₯
Harnack’s Theorem Proof Continued
β€’ There are at most 2𝐷 unbounded components.
β€’ Assume by contradiction there are more.
β€’ A circle 𝐢 (which is 𝑍 π‘₯ 2 + 𝑦 2 βˆ’ π‘Ÿ ) with arbitrarily large
radius will not be a subset of 𝑍(𝑓) but will intersect all
components.
β€’ Therefore the system 𝑓 = π‘₯ 2 + 𝑦 2 βˆ’ π‘Ÿ = 0 will have
more than 2𝐷 solutions, a contradiction.
β€’ In total we get a bound of 𝐷 𝐷 βˆ’ 1 + 2𝐷 = 𝑂(𝐷2 ), as
required
19
Ham-Sandwich Theorem
(Stone-Tukey)
β€’ We say a finite set 𝐴 is bisected by a hyperplane β„Ž if the
halfspaces created by β„Ž have at most
each.
𝐴
2
points of 𝐴
We assume without proof the following theorem:
β€’ For every 𝑑 finite sets in ℝ𝑑 there is a hyperplane β„Ž
simultaneously bisecting all them.
20
Polynomial Ham-Sandwich
Theorem
β€’ Let 𝐴1 , … 𝐴𝑠 βŠ‚ ℝ2 be finite sets, and let 𝐷 ∈ β„• such that
𝐷+2
βˆ’ 1 β‰₯ 𝑠,
2
β€’ Then there is a nonzero bivariate polynomial of degree at
most 𝐷 that simultaneously bisects each 𝐴𝑖
β€’ Here β€œbisects 𝑨” means that 𝑓 > 0 on at most
and 𝑓 < 0 on at most
𝐴
2
𝐴
2
points
points.
β€’ This definition is equivalent to the previous if 𝑓 is linear
(then 𝑍(𝑓) is a hyperplane)
β€’ There is a generalized result
21
Polynomial Ham-Sandwich
Theorem Proof
β€’ Note 𝐷+2
is the number of monomials in a bivariate
2
polynomial of degree 𝐷, which is the number of
solutions to 𝑖 + 𝑗 ≀ 𝐷 where 𝑖, 𝑗 ∈ β„•
β€’ Set π‘˜ ≔
𝐷+2
2
βˆ’ 1, Number of non-constant monomials
β€’ Denote Ξ¦: ℝ2 β†’ β„π‘˜ the Veronese map:
Ξ¦ π‘₯, 𝑦 ≔ π‘₯ 𝑖 𝑦 𝑗
𝑖,𝑗 |1≀𝑖+𝑗≀𝐷
∈ β„π‘˜
β€’ The coordinates in β„π‘˜ are denoted by pairs (𝑖, 𝑗) where
1≀𝑖+𝑗 ≀𝐷
β€’ Example 𝐷 = 2: Ξ¦ π‘₯, 𝑦 = (π‘₯, 𝑦, π‘₯ 2 , π‘₯𝑦, 𝑦 2 )
22
Polynomial Ham-Sandwich
Theorem Proof - Continued
β€’ We assume that 𝑠 = π‘˜, and we denote 𝐴′𝑖 = Ξ¦ 𝐴𝑖 .
β€’ 𝐴′𝑖
π‘˜
𝑖=1
are π‘˜ sets in β„π‘˜
β€’ Let β„Ž be a hyperplane in β„π‘˜ simultaneously bisecting 𝐴′𝑖 .
β€’ β„Ž has equation of the form: β„Ž 𝑧 = π‘Ž00 + 𝑖𝑗 π‘Žπ‘–π‘— 𝑧𝑖𝑗 = 0
where 𝑧𝑖𝑗 1≀𝑖+𝑗≀𝐷 are the coordinates in β„π‘˜ .
β€’ By the definition of Ξ¦, it is easy to see that the
polynomial 𝑓 π‘₯, 𝑦 = 𝑖,𝑗 π‘Žπ‘–π‘— π‘₯ 𝑖 𝑦 𝑗 (we include π‘Ž00 ) is the
desired polynomial, and deg 𝑓 ≀ 𝐷
β€’ Indeed, 𝑓 π‘₯, 𝑦 = β„Ž(Ξ¦ π‘₯, 𝑦 )
23
General Polynomial HamSandwich Theorem
β€’ Let 𝐴1 , … 𝐴𝑠 βŠ‚ ℝ𝑑 be finite sets, and let 𝐷 ∈ β„• such that
𝐷+𝑑
βˆ’ 1 β‰₯ 𝑠,
𝑑
β€’ Then there is a nonzero 𝑑-variate polynomial of degree
at most 𝐷 that simultaneously bisects each 𝐴𝑖
24
Definition
β€’ Let 𝑃 βŠ‚ ℝ2 be a set of π‘š points, let π‘Ÿ ∈ β„•, 1 ≀ π‘Ÿ ≀ π‘š
be a parameter.
β€’ We say that 𝑓 ∈ ℝ[π‘₯, 𝑦] is an r-partitioning polynomial
for 𝑃 if every connected component of ℝ2 βˆ– 𝑍(𝑓)
(β€œCell”) contains at most π‘š π‘Ÿ points of 𝑃. Note every cell
is open
β€’ No bound on 𝑃 ∩ 𝑍 𝑓
25
Polynomial Partitioning
Theorem
β€’ For every set of π‘š points 𝑃 βŠ‚ ℝ2 and 1 ≀ π‘Ÿ ≀ π‘š there is
an r-partitioning polynomial of degree at most 𝑂 π‘Ÿ
β€’ The number of cells is 𝑂(π‘Ÿ) (we will not prove or use it)
Proof Structure:
26
Polynomial Partitioning
Theorem Proof
β€’ Construct collections 𝒫0 , 𝒫1 , … inductively, each consists
disjoint subsets of 𝑃 such that 𝒫𝑗 ≀ 2𝑗 .
β€’ Start with 𝒫0 ≔ {𝑃}. Assume 𝒫𝑗 was constructed.
β€’ By the Polynomial Ham-Sandwich Theorem there is a
polynomial 𝑓𝑗 bisecting all sets in 𝒫𝑗 , deg 𝑓𝑗 ≀ 2 β‹… 2𝑗 .
β€’ We can use the theorem because
2β‹…2𝑗 +2
2
βˆ’ 1 β‰₯ 2𝑗 β‰₯ |𝒫𝑗 |.
β€’ For every 𝑄 ∈ 𝒫𝑗 subset of 𝑃 denote
𝑄+ = {𝑝 ∈ 𝑄|𝑓𝑗 𝑝 > 0} and π‘„βˆ’ = {𝑝 ∈ 𝑄|𝑓𝑗 𝑝 < 0}
β€’ Set 𝒫𝑗+1 ≔
π‘„βˆˆπ’«π‘— {𝑄
+, π‘„βˆ’}
β€’ By induction βˆ€π‘„ ∈ 𝒫𝑗 , 𝑄 ≀ |𝑃|
2𝑗 .
27
Polynomial Partitioning
Theorem Proof - Continued
β€’ Let 𝑑 = log 2 π‘Ÿ . Then βˆ€π‘„ ∈ 𝒫𝑑 𝑄 ≀ |𝑃| π‘Ÿ.
β€’ Set 𝑓 ≔ 𝑓1 𝑓2 … 𝑓𝑑
β€’ No cell contains points of two different sets in 𝒫𝑑 . Otherwise,
for points 𝑝, π‘ž in different sets there is 𝑓𝑗 such that 𝑓𝑗 𝑝 > 0,
𝑓𝑗 π‘ž < 0.
β€’ There is a path between the two points, so somewhere on
that path 𝑓𝑗 𝑐 = 0, so it crosses 𝑍(𝑓) which is a contradiction
β€’ Bounding the degree of 𝑓:
𝑑
deg 𝑓 =
𝑑
2𝑗 2
deg(𝑓𝑗 ) ≀ 2
𝑗=1
𝑗=1
≀
2
2βˆ’1
2𝑑
2
= 𝑂( π‘Ÿ)
28
Cuttings - Reminder
β€’ Given 𝑛 lines, 𝒕-cutting is dividing the plane into
(generalized) triangles such that each cell is intersected
at most by 𝑛 𝑑 lines
β€’ Cutting Lemma: For every 𝑛 and a parameter 1 ≀ 𝑑 ≀ 𝑛
there is a 𝑑-cutting with 𝑂(𝑑 2 ) triangles
29
Credit to
Peleg
Partitioning compared to
Cuttings
Both partitioning and cuttings divide the plane into cells:
𝒓-Partitioning
Points in a cell
|𝑃| π‘Ÿ
Cells a line intersects
𝑂( π‘Ÿ)
Lines intersecting a
cell
Average of 𝑂(|𝐿|
𝒓-Cutting
Average of 𝑂(|𝑃| π‘Ÿ)
𝑂
π‘Ÿ)
π‘Ÿ (? )
|𝐿|
π‘Ÿ
30
Theorem 1 Reminder
β€’ For sets 𝑃 of π‘š points and 𝐿 of 𝑛 lines:
𝐼 𝑃, 𝐿 = 𝑂 π‘š2 3 𝑛2
3
+π‘š+𝑛
31
Theorem 1 Lemma
β€’ Lemma: 𝐼 𝑃, 𝐿 ≀ 𝑛 + π‘š2
Proof:
β€’ Denote by 𝐿′ the lines incident to at most 1 point of 𝑃,
and 𝐿′′ the lines incident to 2 points or more. 𝐼 𝑃, 𝐿 =
𝐼 𝑃, 𝐿′ + 𝐼(𝑃, 𝐿′′ )
β€’ 𝐼 𝑃, 𝐿′ ≀ 𝐿′ ≀ 𝑛.
β€’ For any 𝑝 ∈ 𝑃, there are at most π‘š βˆ’ 1 lines passing
through 𝑝 and another point of 𝑃.
β€’ Hence 𝐼 𝑃, 𝐿′′ ≀ π‘š π‘š βˆ’ 1 < π‘š2
32
Theorem 1 Proof
β€’ We can now assume 𝑛 ≀ π‘š.
β€’ Also assume π‘š ≀ 𝑛, by considering the dual problem:
π‘Ž, 𝑏 ⟼ 𝑦 = π‘Žπ‘₯ βˆ’ 𝑏, 𝑦 = π‘šπ‘₯ + 𝑛 ⟼ π‘š, βˆ’π‘› .
β€’ 𝑏 = π‘šπ‘Ž + 𝑛 ⟺ βˆ’π‘› = π‘Žπ‘š βˆ’ 𝑏
β€’ Denote π‘Ÿ = π‘š4
3
𝑛2 3 .
β€’ By assumptions 1 ≀ π‘Ÿ ≀ π‘š.
β€’ Let 𝑓 be an r-partitioning polynomial of 𝑃.
𝐷 ≔ deg 𝑓 = 𝑂
π‘Ÿ = 𝑂(π‘š2
3
𝑛1 3 )
33
Theorem 1 Proof - Continued
β€’ Let 𝑍 = 𝑍(𝑓),
β€’ 𝐢1 , … , 𝐢𝑠 - Cells of ℝ2 βˆ– 𝑍
β€’ 𝑃0 ≔ 𝑃 ∩ 𝑍 – Points in 𝑍
β€’ 𝑃𝑖 ≔ 𝑃 ∩ 𝐢𝑖 - Points in the i-th cell
β€’ 𝐿0 = 𝑙 ∈ 𝐿 𝑙 βŠ‚ 𝑍 - Lines in 𝑍
β€’ 𝐿𝑖 = 𝑙 ∈ 𝐿 𝑙 ∩ 𝐢𝑖 β‰  πœ™ - Lines intersecting the i-th cell
β€’ We know βˆ€π‘– β‰₯ 1 ,
𝑃𝑖 ≀ π‘š π‘Ÿ = 𝑛2
3
π‘š1
β€’ Now 𝐼 𝑃, 𝐿 = 𝐼 𝑃0 , 𝐿0 + 𝐼 𝑃0 , 𝐿 βˆ– 𝐿0 +
3
𝑠
𝑖=1 𝐼(𝑃𝑖 , 𝐿𝑖 )
34
π‘Ÿ = π‘š4
3
𝑛2
3
Theorem 1 Proof - Continued
𝐷 = 𝑂(π‘š2
π‘š π‘Ÿ = 𝑛2
β€’ By Lemma 3 𝐿0 ≀ 𝐷, hence:
3
3
𝑛1 3 )
π‘š1 3
𝐼 𝑃0 , 𝐿0 ≀ 𝑃0 𝐿0 ≀ π‘šπ· ≀ 𝑛𝐷 = 𝑂(π‘š2 3 𝑛2 3 )
β€’ Every 𝑙 ∈ 𝐿 βˆ– 𝐿0 intersects 𝑍 at most 𝐷 times (Lemma 2):
𝐼 𝑃0 , 𝐿 βˆ– 𝐿0 ≀ 𝐷 𝐿\L0 = 𝑂(π‘š2 3 𝑛2 3 )
β€’ By the lemma: 𝑠𝑖=1 𝐼 𝑃𝑖 , 𝐿𝑖 ≀ 𝑠𝑖=1( 𝐿𝑖 + 𝑃𝑖 2 )
β€’ Because every line passes through 𝑍 at most 𝐷 times, it
intersects at most 𝐷 + 1 cells. Therefore:
𝑠
𝑖=1 |𝐿𝑖 | = 𝑂 𝐷 +
𝑠
2
≀ max 𝑃𝑖
𝑖=1 𝑃𝑖
𝑖
1 𝑛 = 𝑂(π‘š2 3 𝑛2
β‹…
𝑠
𝑖=1
𝑃𝑖 ≀
π‘š
π‘Ÿ
3
+ 𝑛)).
β‹… π‘š = 𝑂(π‘š2 3 𝑛2 3 ).
35
Theorem 2
β€’ An algebraic curve 𝜸 of degree 𝒃 is defined to be 𝑍(𝑓)
for some 𝑓 ∈ ℝ π‘₯, 𝑦 , deg 𝑓 = 𝑏.
β€’ Given a set 𝑃 of π‘š points and Ξ“ a set of 𝑛 algebraic
curves such that:
β€’ Every 𝛾 ∈ Ξ“ is of a degree at most 𝑏
β€’ For every π‘˜ different points in 𝑃 there exist at most 𝐢
distinct curves in Ξ“ passing through all of them
β€’ Then 𝐼 𝑃, Ξ“ = 𝑂(π‘šπ‘˜ (2π‘˜βˆ’1) 𝑛(2π‘˜βˆ’2)
with constant proportional to 𝑏, π‘˜, 𝐢
(2π‘˜βˆ’1)
+ π‘š + 𝑛)
β€’ In the proof we will assume every 𝛾 ∈ Ξ“ is irreducible
36
Theorem 2 Lemma
Lemma:
β€’ 𝐼 𝑃, Ξ“ = 𝑂 𝑛 + π‘šπ‘˜
β€’ 𝐼 𝑃, Ξ“ = 𝑂 π‘š + 𝑛2
Proof:
β€’ Separate between curves with fewer than π‘˜ incidences,
which have 𝑂(𝑛) incidences, and the other curves.
π‘˜βˆ’1
β€’ For each 𝑝 ∈ 𝑃 there are at most 𝐢 π‘šβˆ’1
=
𝑂(π‘š
)
π‘˜βˆ’1
curves incident to it and to π‘˜ βˆ’ 1 other points. Indeed,
for each choice of π‘˜ βˆ’ 1 points there are at most 𝐢
curves incident to them and 𝑝. In total 𝑂(π‘šπ‘˜ ) incidences
37
Theorem 2 Lemma-continued
Lemma:
β€’ 𝐼 𝑃, Ξ“ = 𝑂 𝑛 + π‘šπ‘˜
β€’ 𝐼 𝑃, Ξ“ = 𝑂 π‘š + 𝑛2
Proof:
β€’ By Bézout the curves have at most 𝑏 2 intersections.
Points lying on at most 1 curve generate 𝑂 π‘š
incidences
β€’ Remaining points lie on at least 2 curves. Each 𝛾
contributes at most 𝑏 2 𝑛 βˆ’ 1 = 𝑂(𝑛) incidences
with such points
β€’ In total 𝑂 π‘š + 𝑛𝑂 𝑛 = 𝑂(π‘š
+ 𝑛2 )
38
Theorem 2 Proof
β€’ Assume π‘š < 𝑛2 and 𝑛 ≀ π‘šπ‘˜ , otherwise by the lemma
𝐼 𝑃, Ξ“ = 𝑂(π‘š + 𝑛)
β€’ Set π‘Ÿ = π‘š2π‘˜ (2π‘˜βˆ’1) 𝑛2 (2π‘˜βˆ’1) , by assumptions 1 ≀ π‘Ÿ ≀
π‘š. Let 𝑓 be an π‘Ÿ-partitioning polynomial for 𝑃 such that
deg 𝑓 = 𝑂 π‘Ÿ = 𝑂(π‘šπ‘˜ 2π‘˜βˆ’1 𝑛1 2π‘˜βˆ’1 )
β€’ Similarly to Theorem 1, denote 𝑍, 𝐢𝑖 , 𝑃0 , 𝑃𝑖 , Ξ“0 , Γ𝑖 .
β€’ Again: 𝐼 𝑃, Ξ“ = 𝐼 𝑃0 , Ξ“0 + 𝐼 𝑃0 , Ξ“ βˆ– Ξ“0 +
𝑠
𝑖=1 𝐼(𝑃𝑖 , Γ𝑖 )
β€’ Assume 𝛾 ∈ Ξ“0 is infinite, otherwise include it in Ξ“ βˆ– Ξ“0
39
Theorem 2 Proof-continued
β€’ Note every 𝛾 ∈ Ξ“0 is irreducible and infinite.
β€’ 𝛾 is 𝑍(𝑔) for some polynomial irreducible 𝑔. Therefore
by Bézout's Theorem 𝑔 is a factor of 𝑓, so Ξ“0 ≀ deg 𝑓
= 𝑂( π‘Ÿ).
β€’ By the lemma 𝐼 𝑃0 , Ξ“0 = 𝑂 π‘š + Ξ“0
= 𝑂(π‘š)
2
=𝑂 π‘š+π‘Ÿ
40
Theorem 2 Proof - Continued
π‘Ÿ = π‘šπ‘˜
2π‘˜βˆ’1
𝑛1
2π‘˜βˆ’1
β€’ Applying Bézout's Theorem to 𝛾 ∈ Ξ“ βˆ– Ξ“0 and 𝑓,
𝛾 ∩ 𝑍 ≀ 𝑏 β‹… deg 𝑓 = 𝑂 π‘Ÿ , hence:
𝐼 𝑃0 , Ξ“ βˆ– Ξ“0 = 𝑂 𝑛 π‘Ÿ = 𝑂(π‘šπ‘˜
(2π‘˜βˆ’1) (2π‘˜βˆ’2) (2π‘˜βˆ’1)
𝑛
)
β€’ Because each 𝛾 ∈ Ξ“ βˆ– Ξ“0 crosses 𝑍 at most 𝑂(deg 𝑓 )
times, it passes through 𝑂(deg 𝑓 ) sets 𝐢𝑖 , therefore
𝑠
=𝑂 𝑛 π‘Ÿ
𝑖=1 |Γ𝑖 | = 𝑂 𝑛 β‹… deg 𝑓
𝑠
𝑖=1 𝐼
β€’ By lemma:
𝑂 𝑛 π‘Ÿ + max 𝑃𝑖
𝑖
𝑠
𝑖=1
𝑃𝑖 , Γ𝑖 = 𝑂
π‘˜βˆ’1
𝑂
𝑠
𝑖=1
𝑃𝑖
Γ𝑖 + 𝑃𝑖
π‘˜
≀
=𝑂 𝑛 π‘Ÿ+
41
Theorem 2 Applications
β€’ Lines: 𝐼 π‘š, 𝑛 = 𝑂 π‘š2 3 𝑛2
3
+π‘š+𝑛
β€’ Unit circles: 𝐼 π‘š, 𝑛 = 𝑂 π‘š2 3 𝑛2
3
+π‘š+𝑛
β€’ General circles: 𝐼 π‘š, 𝑛 = 𝑂 π‘š3 5 𝑛4
5
+π‘š+𝑛
42
Theorem 3 - Reminder
β€’ Every set of 𝑛 points in ℝ2 has a geometric spanning tree
with crossing number 𝑂( 𝑛)
β€’ We will actually prove there is a geometric spanning path
with crossing number 𝑂( 𝑛)
43
Theorem 3 – Motivation
β€’ Range Searching: Given a set 𝑃 of 𝑛 points in ℝ2 and a
collection of subsets of ℝ2 (β€œRanges”), create an efficient
algorithm that receives a range and outputs the points in
the range. We will focus on half-plane range searching
(the ranges are half-planes)
44
Theorem 3 – Motivation-continued
β€’ Approach: Preprocess the points.
β€’ Given a geometric spanning tree with a low crossing
number, there are algorithms which use it to store the
points efficiently and answer half-plane queries in a
relatively short time
45
Theorem 3 – Lower Bound
𝑛 points
𝑛 βˆ’ 1 lines
𝑛 points
𝑛 βˆ’ 1 lines
β‡’ There is a line with
intersections
π‘›βˆ’1
2 π‘›βˆ’2
= Ξ©( 𝑛)
46
Definition
β€’ Let 𝑋 βŠ† ℝ2 , we say 𝑋 has crossing number at most π‘˜ if
each line (with possibly finitely many exceptions)
intersects 𝑋 in at most π‘˜ points.
𝑝5
𝑝4
𝑝1
𝑝2
𝑝6
𝑝7
𝑝3
𝑝8
47
Lemma A
β€’ Let 𝑃 be a set of 𝑛 points, let 𝑋 βŠ† ℝ2 be a pathconnected set containing 𝑃 with crossing number at
most π‘˜.
β€’ Then exists a geometric spanning tree of 𝑃 with crossing
number at most 2π‘˜
β€’ Given this lemma, to prove Theorem 3 we only need to
construct a path-connected set 𝑋 containing 𝑃 with
𝑝5
crossing number at most 𝑂( 𝑛)
𝑝4
𝑝5
𝑝4
𝑝1
𝑝2
𝑝3
𝑝1
𝑝6
𝑝7
𝑝8
𝑝6
𝑝2
𝑝3
48
𝑝7
𝑝8
Lemma A Stage 1 Drawing
𝑝5
𝑝5
𝑝4
𝑝4
π‘ž4
𝑝1 /π‘ž1
𝑝1
𝑝6
𝑝2
𝑆
𝑝2 /π‘ž2 /π‘ž3
π‘ž5
𝑝6 /π‘ž6
π‘ž7
𝑝7
𝑝7
𝑝3
𝑝3
𝑝8
𝑝8
𝑝1 /π‘ž1
𝑝1
𝑝1 /π‘ž1
𝑝2 /π‘ž2
𝑝2
𝑝3
49
Lemma A Proof Stage 1
β€’ Stage 1: We build a geometric tree S which spans 𝑃 =
{𝑝1 , … , 𝑝𝑛 } and some other points Q, whose edges are
arcs (not necessarily straight segments) contained in 𝑋.
β€’ We set 𝑆1 = 𝑝1 . Assuming we built such a tree 𝑆𝑖 for
points {𝑝1 , … , 𝑝𝑖 }, we choose an arc 𝛼𝑖 that connects
𝑝𝑖+1 to some point π‘žπ‘– ∈ 𝑆𝑖 such that 𝑆𝑖 ∩ 𝛼𝑖 = {π‘žπ‘– }.
β€’ We set 𝑆𝑖+1 ≔ 𝑆𝑖 βˆͺ 𝛼𝑖 and 𝑆 ≔ 𝑆𝑛 . Because 𝑆 βŠ† 𝑋 and
𝑋 has crossing number at most π‘˜, 𝑆 has a crossing
number at most π‘˜ as well.
50
Lemma A Proof - Stage 2
β€’ Stage 2: We replace the arcs of 𝑆 by straight segments.
β€’ This yields a tree 𝑇 spanning 𝑃 and 𝑄 = {π‘ž1 , … , π‘žπ‘›βˆ’1 },
which has a crossing number of at most π‘˜.
β€’ Indeed, if a line crosses 𝑑 edges of 𝑇 it will cross at least 𝑑
arcs of 𝑆, but the crossing number of 𝑆 is at most π‘˜
51
Lemma A – Stage 3 Drawing
𝑝1 , 𝑝2 , 𝑝3 , π‘ž4 , 𝑝5 , π‘ž5 , 𝑝4 , 𝑝6 , π‘ž7 , 𝑝8 , 𝑝7
𝑝5
𝑝4
π‘ž4
π‘ž5
𝑝1
𝑝6
𝑝2
π‘ž7
𝑝3
𝑝8
𝑝7
52
Lemma A – Stage 3 Drawing
𝑝1 , 𝑝2 , 𝑝3 , π‘ž4 , 𝑝5 , π‘ž5 , 𝑝4 , 𝑝6 , π‘ž7 , 𝑝8 , 𝑝7
𝑝5
𝑝4
π‘ž4
π‘ž5
𝑝1
𝑝6
𝑝2
π‘ž7
𝑝3
𝑝8
𝑝7
53
Lemma A Proof - Stage 3
β€’ Stage 3: We preform a DFS run on the tree starting from
an arbitrary root, creating a list of the points in 𝑃 and 𝑄.
We delete the points of 𝑄 from the list, and connect each
pair of consecutive points by a straight segment.
β€’ This is a geometric spanning tree of 𝑃, which has a
crossing number at most 2π‘˜
𝑝5
𝑝4
π‘ž4
54
π‘ž5
Lemma A –Stage 3 Explanation
β€’ Assume a line crosses 𝑑 new edges. For each new edge
crossed, the same line crosses an edge from the tree 𝑇,
we match between new edges to their edges in 𝑇,
resulting 𝑑 edges of 𝑇.
β€’ Each edge of 𝑇 appears in this list at most twice,
therefore the number of unique edges in the list is at
least 𝑑 2, and the number of unique edges from 𝑇 the
line crosses is at most π‘˜.
β€’ Hence 𝑑 2 ≀ π‘˜, or 𝑑 ≀ 2π‘˜ as required.
𝑝5
𝑝4
π‘ž4
π‘ž5
55
Lemma B
β€’ Let 𝑃 be a set of 𝑛 points, then there exist a set 𝑋 βŠ† ℝ2 that
contains 𝑃 has at most 𝑛 2 connected components and
crossing number 𝑂( 𝑛)
Proof:
β€’ Let 𝑓 be an π‘Ÿ-partitioning polynomial, π‘Ÿ specified later.
β€’ By Harnack’s Theorem the number of connected components
of 𝑍 ≔ 𝑍(𝑓) is at most π‘˜ β‹… deg 𝑓 2 for some constant π‘˜, and
deg 𝑓 ≀ π‘š π‘Ÿ for some constant π‘š.
β€’ We choose π‘Ÿ = 𝑛 𝑐 where 𝑐 = 2π‘š2 π‘˜ is a constant. The
number of connected components of 𝑍 is at most 𝑛 2
β€’ Note that in order to use polynomial partitioning we require
1 ≀ π‘Ÿ ≀ 𝑛 ⇔ 1 ≀ 𝑐 ≀ 𝑛, but we can choose such 𝑐 and
assume 𝑛 is arbitrarily large
56
Lemma B Proof
β€’ For every 𝑝 ∈ 𝑃 βˆ– Z, define πœŽπ‘ a straight segment connecting
𝑝 to 𝑍, avoiding 𝑍 otherwise.
β€’ Define 𝑋 ≔ 𝑍 βˆͺ
components.
π‘βˆˆπ‘ƒβˆ–Z πœŽπ‘ .
𝑋 also has at most 𝑛 2 connected
β€’ Let 𝑙 be a line not contained in 𝑍 and not containing any πœŽπ‘
(By Lemma 3 only finitely many lines are avoided).
β€’ 𝑙 intersects 𝑍 in at most deg 𝑓 = 𝑂( 𝑛) points. 𝑙 crosses at
most deg 𝑓 + 1 components of ℝ2 βˆ– 𝑍, each contains at
most 𝑐 points of 𝑃 (Because 𝑓 is 𝑛 𝑐-partitioning)
β€’ Therefore 𝑙 crosses at most 𝑐 1 + deg 𝑓 = 𝑂 𝑛
segments πœŽπ‘ , hence 𝑙 crosses 𝑋 at most 𝑂 𝑛 times
57
Lemma B Drawing
58
Theorem 3 Proof
β€’ By Lemma A, we only need to construct a connected set
𝑋 containing 𝑃 with crossing number at most 𝑂( 𝑛)
β€’ To create 𝑋, apply Lemma B iteratively. We construct a
sequence of sets 𝐡0 , 𝐡1 , … such that each 𝐡𝑖 contains 𝑃
and has at most 𝑛 2𝑖 connected components
β€’ We begin with 𝐡0 ≔ 𝑃, and having constructed 𝐡𝑖 , we
choose a point in each of its connected components
resulting a set 𝑅𝑖 , 𝑅𝑖 ≀ 𝑛 2𝑖 .
β€’ Lemma B provides us with a set 𝑋𝑖 , 𝑅𝑖 βŠ† 𝑋𝑖 with crossing
number 𝑂( 𝑛 2𝑖 ) and at most 𝑛 2𝑖+1 connected
components.
59
Theorem 3 Proof - Continued
β€’ We define 𝐡𝑖+1 ≔ 𝐡𝑖 βˆͺ 𝑋𝑖
β€’ 𝐡𝑖+1 has at most 𝑛 2𝑖+1 connected components
because each point is reachable from one of the
components of 𝑋𝑖
β€’ For some 𝑖0 we reach a connected 𝐡𝑖0 , which will be 𝑋.
β€’ By definition 𝑋 =
is at most
𝑂( 𝑛)
𝑖≀𝑖0 𝑂(
𝑖≀𝑖0 𝑋𝑖 , therefore its crossing number
βˆ’π‘– 2
𝑛 2𝑖 ) ≀ 𝑂 𝑛 β‹… ∞
𝑂
2
=
𝑖=0
60
Summary
β€’ We’ve proved the Polynomial Partitioning Theorem
β€’ Used it to prove Szemerédi-Trotter theorem
β€’ Used it to generalize the theorem to algebraic curves
β€’ Proved the existence of geometric spanning trees with
low crossing number
β€’ Next lecture: (We) will prove the lower bound of the
distinct distances problem
Questions?
61
Thank You!
62