Graphs Vectors and Matrices Daniel A Spielman Yale
- Slides: 105
Graphs, Vectors, and Matrices Daniel A. Spielman Yale University AMS Josiah Willard Gibbs Lecture January 6, 2016
From Applied to Pure Mathematics Algebraic and Spectral Graph Theory Sparsification: approximating graphs by graphs with fewer edg The Kadison-Singer problem
A Social Network Graph
A Social Network Graph
A Social Network Graph “vertex” or “node” “edge” = pair of nodes
A Social Network Graph “vertex” or “node” “edge” = pair of nodes
A Big Social Network Graph
A Graph = vertices, = edges, pairs of vertices 1 8 2 3 5 4 6 9 12 7 10 11
The Graph of a Mesh
Examples of Graphs 1 8 2 3 5 4 6 9 12 7 10 11
Examples of Graphs 1 8 2 3 5 4 6 9 12 7 10 11
How to understand large-scale structure Draw the graph Identify communities and hierarchical structure Use physical metaphors Edges as resistors or rubber bands Examine processes Diffusion of gas / Random Walks
The Laplacian quadratic form of
The Laplacian quadratic form of 1 0. 5 1. 5 0 1 0. 5
The Laplacian quadratic form of 1 0. 5 1. 5 0 1 0. 5
The Laplacian matrix of
Graphs as Resistor Networks View edges as resistors connecting vertices Apply voltages at some vertices. Measure induced voltages and current flow. 1 V 0 V
Graphs as Resistor Networks Induced voltages minimize subject to constraints. , 1 V 0 V
Graphs as Resistor Networks Induced voltages minimize subject to constraints. , 1 V 1 V 0. 5 V 0 V 0 V 0. 5 V 0. 375 V 0. 625 V
Graphs as Resistor Networks Induced voltages minimize subject to constraints. , 1 V 1 V 0. 5 V 0 V 0 V 0. 5 V 0. 375 V 0. 625 V
Graphs as Resistor Networks Induced voltages minimize subject to constraints. , Effective conductance = current flow with one vo 1 V 1 V 0. 5 V 0 V 0 V 0. 5 V 0. 375 V 0. 625 V
Weighted Graphs Edge assigned a non-negative real weigh measuring strength of connection 1/resistance
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Edges are drawn as curves for visibility.
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize to fix scale, require
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize to fix scale, require
Courant-Fischer Theorem Where is the smallest eigenvalue of and is the corresponding eigenvector.
Courant-Fischer Theorem Where is the smallest eigenvalue of and is the corresponding eigenvector. For and is a constant vector
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize Such that and
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize Such that and Courant-Fischer Theorem: solution is , the eigenvector of the second-smallest eigenvalue ,
Spectral Graph Drawing (Hall ’ 70) = area under blue curves
Spectral Graph Drawing (Hall ’ 70) = area under blue curves
Space the points evenly
And, move them to the circle
Finish by putting me back in the center
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize Such that and
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize Such that and
Spectral Graph Drawing (Hall ’ 70) Want to map with most edges short Minimize Such that and and , to prevent
Spectral Graph Drawing (Hall ’ 70) Minimize Such that and Courant-Fischer Theorem: solution is , up to rotation
Spectral Graph Drawing (Hall ’ 70) 9 1 2 3 5 5 4 1 6 8 7 4 6 8 9 Arbitrary Drawing 3 2 7 Spectral Drawing
Spectral Graph Drawing (Hall ’ 70) Original Drawing Spectral Drawing
Spectral Graph Drawing (Hall ’ 70) Original Drawing Spectral Drawing
Dodecahedron Best embedded by first three eigenvectors
Spectral drawing of Erdos graph: edge between co-authors of papers
When there is a “nice” drawing: Most edges are short Vertices are spread out and don’t clump too mu is close to 0 When is big, say there is no nice picture of the graph
Expanders: when is big Formally: infinite families of graphs of constant degree d and large Examples: random d-regular graphs Ramanujan graphs Have no communities or clusters. Incredibly useful in Computer Science: Act like random graphs (pseudo-random) Used in many important theorems and algorit
Good Expander Graphs -regular graphs with Courant-Fischer: for all
Good Expander Graphs -regular graphs with Courant-Fischer: for all For , the complete graph on , so for vertices
Good Expander Graphs
(S-Teng ‘ 04) Sparse Approximations of Graphs A graph is a sparse approximation of if has few edges and few: the number of edges in is or , where if for all
(S-Teng ‘ 04) Sparse Approximations of Graphs A graph is a sparse approximation of if has few edges and few: the number of edges in is or , where if Where for all if for all
(S-Teng ‘ 04) Sparse Approximations of Graphs A graph is a sparse approximation of if has few edges and few: the number of edges in is or , where if Where for all if for all
(S-Teng ‘ 04) Sparse Approximations of Graphs The number of edges in is or , where Where if for all
Why we sparsify graphs To save memory when storing graphs. To speed up algorithms: flow problems in graphs (Benczur-Karger ‘ 96) linear equations in Laplacians (S-Teng ‘ 04)
Graph Sparsification Theorems For every , there is a and (Batson-S-Srivastava ‘ 0
Graph Sparsification Theorems For every , there is a and (Batson-S-Srivastava ‘ 0 By careful random sampling, can quickly get (S-Srivastava ‘ 08)
Laplacian Matrices
Laplacian Matrices
Laplacian Matrices
Matrix Sparsification
Matrix Sparsification subset of vectors, scaled up
Matrix Sparsification subset of vectors, scaled up
Matrix Sparsification
Simplification of Matrix Sparsification is equivalent to
Simplification of Matrix Sparsification Set We need
Simplification of Matrix Sparsification Set “Decomposition of the identity” “Parseval frame” “Isotropic Position”
Matrix Sparsification by Sampling (Rudelson ‘ 99, Ahlswede-Winter ‘ 02, Tropp ’ 11) For Choose If choose with probability , set
Matrix Sparsification by Sampling (Rudelson ‘ 99, Ahlswede-Winter ‘ 02, Tropp ’ 11) For Choose If choose with probability , set (effective conductance
Matrix Sparsification by Sampling (Rudelson ‘ 99, Ahlswede-Winter ‘ 02, Tropp ’ 11) For Choose If choose with probability , set
Matrix Sparsification by Sampling (Rudelson ‘ 99, Ahlswede-Winter ‘ 02, Tropp ’ 11) For Choose If choose with probability , set With high probability, choose vectors an d
Optimal (? ) Matrix Sparsification (Batson-S-Srivastava ‘ 09 For with Can choose vectors and nonzero values for the so that
Optimal (? ) Matrix Sparsification (Batson-S-Srivastava ‘ 09 For with Can choose vectors and nonzero values for the so that What are the !?
Optimal (? ) Matrix Sparsification (Batson-S-Srivastava ‘ 09 For with Can choose vectors and nonzero values for the so that !?
The Kadison-Singer Problem ‘ 59 Equivalent to: Anderson’s Paving Conjectures (‘ 79, ‘ 81) Bourgain-Tzafriri Conjecture (‘ 91) Feichtinger Conjecture (‘ 05) Many others Implied by: Weaver’s KS 2 conjecture (‘ 04)
Weaver’s Conjecture: Isotropic vectors for every unit vector
Partition into approximately ½-Isotropic S
Partition into approximately ½-Isotropic S
Partition into approximately ½-Isotropic S
Partition into approximately ½-Isotropic S because
Big vectors make this difficult
Big vectors make this difficult
Weaver’s Conjecture KS 2 There exist positive constants if all and so that and then exists a partition into S 1 and S 2 with
Theorem (Marcus-S-Srivastava ‘ 15) For all if all and then exists a partition into S 1 and S 2 with
We want
We want
We want Consider expected polynomial of a random partiti
Proof Outline 1. Prove expected characteristic polynomial has real roots 2. Prove its largest root is at most 3. Prove is an interlacing family, so exists a partition whose polynomial has largest root at most
Interlacing Polynomial interlaces if Example:
Common Interlacing and have a common interlacing if can partition the line into intervals so that each contains one root from each polynomial )( )( ) ( )(
Common Interlacing If p 1 and p 2 have a common interlacing, for some i. Largest root of average )( )( ) ( )(
Common Interlacing If p 1 and p 2 have a common interlacing, for some i. Largest root of average )( )( ) ( )(
Without a common interlacing
Without a common interlacing
Without a common interlacing
Without a common interlacing
Common Interlacing If p 1 and p 2 have a common interlacing, for some i. Largest root of average )( )( ) ( )(
Common Interlacing and )( have a common interlacing iff is real rooted for all )( ) ( )(
Interlacing Family of Polynomials is an interlacing family if its members can be placed on the leaves of a tree so when every node is labeled with the average of leaves siblings have common interlacings
Interlacing Family of Polynomials is an interlacing family if its members can be placed on the leaves of a tree so when every node is labeled with the average of leaves siblings have common interlacings have a common interlacing
Interlacing Family of Polynomials is an interlacing family if its members can be placed on the leaves of a tree so when every node is labeled with the average of leaves siblings have common interlacings have a common interlacing
Interlacing Family of Polynomials Theorem: There is a so that
Interlacing Family of Polynomials Theorem: There is a so that have a common interlacing
Interlacing Family of Polynomials Theorem: There is a so that have a common interlacing
Our family is interlacing Form other polynomials in the tree by fixing the choices of where some vectors
Summary 1. Prove expected characteristic polynomial has real roots 2. Prove its largest root is at most 3. Prove is an interlacing family, so exists a partition whose polynomial has largest root at most
To learn more about Laplacians, see My class notes from “Spectral Graph Theory” and “Graphs and Networks” My web page on Laplacian linear equations, sparsification, etc. To learn more about Kadison-Singer Papers in Annals of Mathematics and survey from ICM. Available on ar. Xiv and my web page
- Daniel spielman spectral graph theory
- Simulink matrix multiplication
- Ingrid spielman
- Stephanie spielman osu
- Spectral graph theory spielman
- Draw the graphs with the following adjacency matrices.
- Testability tips in software testing
- Graphs that enlighten and graphs that deceive
- Speed and velocity
- End behavior of polynomials
- Color vectors and viewpoint in still and moving images
- Ivory no.1 retainer uses
- Aims and objectives of matrices
- Applications of matrices and determinants
- Pam scoring
- Homogeneous system of linear equations matrix
- Node reduction algorithm in software testing
- Blosum62
- Identity matrix
- Condition for matrix multiplication
- Determine whether each matrix product is defined
- Matrix all formula
- Persuasive communication theories
- Yale program for recovery and community health
- Partitioning a line segment formula
- A storm system moves 5000 km due east
- What are i and j in vectors
- Difference between vector and scalar
- A vector has both location and direction.
- Optr is a trapezium
- 2d motion equations
- Scalar vs vector
- Scalars and vectors
- Linear algebra chapter 4 solutions
- Linearly dependent and independent vectors
- Vectors and the geometry of space
- Linear dependency
- Dot product
- Vectors form 3
- Chapter 12 vectors and the geometry of space
- Properties of vector
- Vectors trigonometry
- The diagram shows two vectors that point west and north.
- Resolution and composition of vectors
- Collinear vectors
- Chapter 12 vectors and the geometry of space solutions
- Physical quantities of vectors
- Si unit of pressure
- Antiparallel vectors
- A student adds two vectors with magnitudes of 200 and 40
- Scalar triple product properties
- Technology applications teks
- Free fall
- You are adding vectors of length 20 and 40 units
- Networks and graphs: circuits, paths, and graph structures
- Tomer owns a daycare
- Matrici progresive raven standard raspunsuri
- Inversa de la matriz identidad
- Find the determinant of the following matrix:
- Can matrices be divided
- Linear independence matlab
- Wj-iv sentence writing fluency scoring
- Basic perinatal matrices
- Matrices identidad
- ç
- Matrix equation
- 2*2 matrix
- What is i in matrix
- Matriz epistemologica
- Linear algebra for dummies
- Multiplication of determinants
- Foote cone belding matrix
- Clasificacion de matrices
- Matrix multiplication dimensions
- Inverse of a 3×3 matrix
- Homogenous transformation matrices
- Subtract matrices calculator
- Division de matrices 3x3
- Matriz fila concepto
- Concatenating vectors matlab
- Chapter 4 matrices answers
- Augmented matrix
- Cuadro comparativo de matrices
- Aplicacion de estructuras de datos vectores y matrices
- Matrix multiplication associative property
- Transformations using matrices
- Order of matrices
- Matrices
- Matrices equivalentes
- 3-6 solving systems using matrices
- Leslie matrix
- Dimension u orden de una matriz
- Matrices y determinantes
- Matrices in google docs
- Det(a)
- Subtraction of matrices definition
- Partioned matrix
- Irreducible matrix
- Shumezimi i dy matricave
- Unit 1 algebra basics homework 6 matrices
- Solving equations calculator
- Matrix algebra for dummies
- Matrices for dummies
- Con las matrices n
- Matrices
- Matrices