Vectorized Laplacians for dealing with high-dimensional data sets

Size: px
Start display at page:

Download "Vectorized Laplacians for dealing with high-dimensional data sets"

Transcription

1 Vectorized Laplacians for dealing with high-dimensional data sets Fan Chung Graham University of California, San Diego The Power Grid Graph drawn by Derrick Bezanson 2010

2 Vectorized Laplacians Graph gauge theory Earlier work (3 papers ) with Shlomo Sternberg, Bert Kostant on the vibrational spectra and spins of the molecules. New directions in organizing/analyzing high-dimensional data sets, with applications in - vector diffusion maps, - graphics, imaging, sensing - resource allocation with dynamic demands Singer + Wu 2011, C. + Zhao 2012

3 Cayley graphs G = (V,E) Γ V! a group H =A 5 u! v! u = gv for g" B, a chosen subset of H. B={(12345),(54321),(12)(34)}!A 5

4 Cayley graphs G = (V,E) Γ V! a group H u! v! u = gv for g" B, a chosen subset of H. adjacency matrix! 0 0 irreducible representations Frobenius 1897

5 Eigenvalues of the buckyball with weight 1 for single bonds and t for double bonds Roots of with multiplicity 5 (x 2 + x t 2 + t 1)(x 3 tx 2 x 2 t 2 x 3x + t 3 t 2 + t + 2) = 0 (x 2 + x t 2 1)(x 2 + x (t +1) 2 ) = 0 with multiplicity 4 (x 2 + (2t +1)x + t 2 + t 1)(x 4 3x 3 + ( 2t 2 + t 1)x 2 + (3t 2 4t + 8)x + t 4 t 3 + t 2 + 4t 4) = 0 with multiplicity 3 x t 2 = 0 with multiplicity = 60

6

7 The spectrum of a graph Discrete Laplace operator 1! f ( x) = #( f ( x) " f ( y)) d x y! x 1 " f ( x ) =! f ( x )! f ( x ) j j j 2 For In a path P n + 1 {( )!( f ( x )! f ( x ))} j j! 1 2! " f ( x )!!! x 2 j " f ( x ) " f ( x ) j+ 1 j! x! x { }

8 In a graph Γ = (V,E), Δ acts on In general, we consider F (V,R) = { f :V R} Δ acts on F (V, X) = { f :V X}. geometric data time based data statistical data Example 1: Vibrational spectra of molecules F (V,R 3 ). Δ = Δ Example 2: Vector Diffusion Maps F (V,R d ). Δ = Δ nd nd V = n

9 Connections: a classical framework for dealing with Vibrations and spins of the Buckyball Noise in the data sets, etc.

10 Vector bundle E = { E x : x V}

11 Vector bundle Connection T E = { E x : x V}

12 Vector bundle E = { E x : x V} Connection T T x,e : E y E x for x V, e =(x,y) E T x,e T y,e = id Example 1: Vibrational spectra of molecules Spins of the Buckyball

13 Example 1: Vibrational spectra of molecules u h(v) h(u) v F (V,R 3 ) = {h :V R 3 } the space of displacements

14 Example 1: Vibrational spectra of molecules Hooke s law: h :V R 3 Potential energy u,v u h(u) v h(v) ( ) = k u,v u + h(u) v h(v) u v u,v k u,v w u,v (h(u) h(v)) = k < h(u) h(v), A (h(u) h(v)) > u,v e u,v where w u,v = u v u v, 2 A = w w t e u,v u,v 2

15 The vibrational spectrum (infrared) of the Buckyball

16 Vector bundle E = { E x : x V} Connection T T x,e : E y E x for x V, Selection sece s(u) T x,e T y,e = id E u for s sece e =(x,y) E Suppose a group G acts on V as graph automorphisms. For a G and u V, a : E u E au a b = ab : E u E abu G acts as automorphisms on connections: (a T ) = at x,e a 1 x,a 1 e a 1 x,e Define Q (s) = (s(x) T s(y)) A (s(x) T s(y)) T,A x,e x,e x,e

17 Example 1: Vibrations and spins on the Buckyball Theorem: C.+Sternberg 1992 In a homogenous graph Γ = a G-invariant connection on G/H with edge generating set B, E = E(Γ, X) defines an isomorphism t b Auto(X), such that the operator Q T,A (s) = x,e can be represented as F = (s(x) T s(y)) A (s(x) T s(y)) x,e x,e x,e b B A v,be I b B A v,be t b r(b) For the Buckyball G = SL(2,5), PSL(2,5) A 5, G Isotropy subgroup Z 2

18 Examples of connection graphs Example 1: Vibrational spectra of molecules Spins of the buckyball Example 2: Vector Diffusion Maps Singer + Wu 2011 rotation O uv data point u data point v

19 Examples of connection Laplacians Example 2: Vector Diffusion Maps Singer + Wu 2011 At a node u, Data local PCA dimension reduction vector in R d In a network, rotation O uv data point u data point v

20 Consistency of connection Laplacians For a connected connection graph G of dimension d, For f :V R d (i, j ) E < f, L f >= f (u)o uv f (v) 2 wuv

21 Consistency of connection Laplacians For a connected connection graph G of dimension d, the following statements are equivalent: The 0 eigenvalue has multiplicity d. All eigenvalues have multiplicity d. For any two vertices u and v, any two paths P, P, O xy = O zw xy P zw P' For each vertex v, we have O v SO(d), such that O = O 1 O uv u v for all (u,v) E.

22 Consistency of connection Laplacians Theorem: For a connected connection graph G of dimension d, if the 0 eigenvalue has multiplicity d, then we can find O v SO(d), such that O = O 1 O uv u v for all (u,v) E. Method 1: For a fixed vertex u, choose a frame O u consisting of any d orthonormal vectors. i For any v, define O v = O vi v i+1 is any path joining u and v. where u = v 0,v 1,...,v t Method 2: Use eigenvectors. Singer+Spielman 2012

23 Dealing with noise in the data Several combinatorial approaches Random connection Laplacians Graph sparsification algorithms PageRank algorithms Graph limits and graphlets...

24 Dealing with noise in the data Several combinatorial approaches Random connection Laplacians Graph sparsification algorithms PageRank algorithms Graph limits and graphlets...

25 A random graph model for any given expected degree sequence Independent random indicator variables X ij = X ji,1 i j n. Pr(X ij = 1) = Pr(v i v j ) = p ij ( ) X = X ij v i v j `` Example: The Erdos-Renyi model G(n, p) with p ij all equal to p.

26 A matrix concentration inequality Theorem: C. + Radcliffe 2011 X i : independent random Ahlswede-Winter,Cristofides- Markstrom, Oliveira, Gross, Recht, Tropp,... n n Hermitian matrix Then with X = m i=1 X i E(X i ) M X i satisfies Pr( X E(X ) > a) 2n exp m a 2 2v 2 + 2Ma / 3 where v 2 = var(x ) i i=1

27 Eigenvalues of random graphs with general distributions Theorem: C. + Radcliffe 2011 G : a random graph, where Pr(v i v j ) = p ij. A : A : adjacency matrix of G expected adjacency matrix A ij = p ij With probability, eigenvalues of A and satisfy 1 ε A for all i, λ i (A) λ i (A) 4d max log(2n / ε) 1 i n,if the maximum degree satisfies d max > 4 9 log( 2n ε ).

28 Eigenvalues of the normalized Laplacian for random graphs with general distributions Theorem: C. + Radcliffe 2011 G : a random graph, where Pr(v i v j ) = p ij. A : A : D adjacency matrix of G expected adjacency matrix diagonal degree matrix of G A ij = p ij With probability 1 ε, eigenvalues of L = I D 1/2 AD 1/2 satisfy for all i, 3log(4n / ε) λ (L) λ (L) i i δ 1 i n,and if the minimum degree satisfies L = I D 1/2 AD 1/2, δ > 10logn.

29 Dealing with noise in the data Several combinatorial approaches Random connection Laplacians Graph sparsification algorithms PageRank algorithms Graph limits and graphlets...

30 Graph sparsification For a graph G=(V,E,W), we say H is an ε-sparsifier if for all have h G (S) h H (S) εh G (S). S V we Benczur and Karger sparsifier with size O(n log n) in running time O(n 2 ) Spielman and Sriastava using effective resistance running time O(n) Batson,Spielman and Sriastava running time O(n) O(nm) sparsifier

31 Graph sparsification Sparsification Sampling edges e The cut edges are more likely to be sampled in order to satisfy, for all S V, h G (S) h H (S) εh G (S) We need to rank edges.

32 Dealing with noise in the data Several combinatorial approaches Random connection Laplacians Graph sparsification algorithms PageRank algorithms for ranking edges Graph limits and graphlets...

33 PageRank for ranking vertices Definition of PageRank Two equivalent ways to define PageRank p=pr(!,s) (1) p =! s + (1 "!) pw seed jumping constant (2) p "!! = $ # t= 0 t (1 ) ( sw ) t s = (,,..., ) n n n the (original) PageRank s = some seed, e.g., (1,0,...,0) personalized PageRank

34 PageRank as generalized Green s function Discrete Green s function Chung+Yau 2000 Laplacian n 1 L = D 1/2 ΔD 1/2 = λ k P k k =1 Green s function G = LG = I n 1 k =1 The general Green s function G β = 1 λ k P k restricted to n 1 k =1 1 β + λ k P k (ker L ) PageRank pr(α, s) = βsd 1/2 G β D 1/2, β = 2α 1 α

35 Ranking pages using Green s function and electrical network theory G = (V,E,W ) i V :V : a weighted graph injected current function w u,v conductance i E : E induced current function 1 if v is e's head B(e,v) = 1 if v is e's tail 0 otherwise Kirchoff s current law: Ohm s current law: i V = i E B i E = f B T W L = B T WB 1 u voltage potantial function v 1

36 Ranking pages using Green s function and electrical network theory Kirchoff s current law: Ohm s current law: i V = f V B T W B = f V L f V = i V D 1/2 G D 1/2 R(u,v) = f V (χ u χ v ) T β β i V = i E B B' = D1/2 B i E = f B T W β β = i V D 1/2 G D 1/2 (χ u χ v ) T β W β = βi 0 0 W L β = βi + L L G β β = I = (χ u χ v ) D 1/2 G D 1/2 (χ u χ v ) T β pr α (s) = sd 1/2 G β D 1/2 where β = 2α 1 α

37 Ranking pages using Green s function and electrical network theory The Green value of (u,v) R α (u,v) = h α (u,v) + h α (v,u) = pr α,u(u) d u pr α,u(v) d v + pr α,v(v) d v pr α,v(u) d u

38 Vectorized Laplacian and vectorized PageRank A connection graph G = (V,E,O,w) A connection matrix w O if (u,v) E, A = uv uv 0 otherwise. d d D(v,v) = d u I d d L = D A (u,v) E f Lf T = w uv f (u)o uv f (v) 2

39 Vectorized Laplacian and vectorized PageRank A connection graph G = (V,E,O,w) The transition probability matrix The vectorized PageRank P = D 1 A Z = I + P 2 ˆp(α,v) = α ŝ + (1 α) ˆpZ

40 Ranking pages using Green s function and electrical network theory The Green value of (u,v) R α (u,v) = ˆp α,u (u) d u ˆp α,u (v) d v + ˆp α,v (v) d v ˆp α,v (u) d u which the sampling subroutine is based upon.

41 Dealing with noise in the data Several combinatorial approaches Random connection Laplacians Graph sparsification algorithms PageRank algorithms for ranking edges Graph limits and graphlets...

42 Examples of connection Laplacians Example 1: Vibrational spectra of molecules Spins of the buckyball Example 2: Electron microscopy Vector Diffusion Maps Example 3: Graphics, vision,... Managing millions of images, photos.

43 Millions of photos in a box! Navigate using connection graphs. - finding consistent subgraphs, paths...

44 Feature detection Detect features using SIFT [Lowe, IJCV 2004]

45 Feature matching Match features between each pair of images

46 (graph layout produced using the Graphviz toolkit: Part of an image connection graph Building Rome in a day Agarwal, Snavely, Simon, Seitz, Szeliski 2009

47 Vectorized Laplacians for dealing with high-dimensional data sets Fan Chung Graham University of California, San Diego The Power Grid Graph drawn by Derrick Bezanson 2010

48 Thank you.

49 (graph layout produced using the Graphviz toolkit: Part of an image connection graph Building Rome in a day Agarwal, Snavely, Simon, Seitz, Szeliski 2009

50 An example of connection graphs Building Rome in a day Agarwal, Snavely, Simon, Seitz, Szeliski 2009

Ranking and sparsifying a connection graph

Ranking and sparsifying a connection graph Ranking and sparsifying a connection graph Fan Chung 1, and Wenbo Zhao 1 Department of Mathematics Department of Computer Science and Engineering University of California, San Diego La Jolla, CA 9093,

More information

On the Spectra of General Random Graphs

On the Spectra of General Random Graphs On the Spectra of General Random Graphs Fan Chung Mary Radcliffe University of California, San Diego La Jolla, CA 92093 Abstract We consider random graphs such that each edge is determined by an independent

More information

Graphs, Vectors, and Matrices Daniel A. Spielman Yale University. AMS Josiah Willard Gibbs Lecture January 6, 2016

Graphs, Vectors, and Matrices Daniel A. Spielman Yale University. AMS Josiah Willard Gibbs Lecture January 6, 2016 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

More information

Four graph partitioning algorithms. Fan Chung University of California, San Diego

Four graph partitioning algorithms. Fan Chung University of California, San Diego Four graph partitioning algorithms Fan Chung University of California, San Diego History of graph partitioning NP-hard approximation algorithms Spectral method, Fiedler 73, Folklore Multicommunity flow,

More information

Spectral and Electrical Graph Theory. Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity

Spectral and Electrical Graph Theory. Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity Spectral and Electrical Graph Theory Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity Outline Spectral Graph Theory: Understand graphs through eigenvectors and

More information

Sparsification of Graphs and Matrices

Sparsification of Graphs and Matrices Sparsification of Graphs and Matrices Daniel A. Spielman Yale University joint work with Joshua Batson (MIT) Nikhil Srivastava (MSR) Shang- Hua Teng (USC) HUJI, May 21, 2014 Objective of Sparsification:

More information

Laplacians of Graphs, Spectra and Laplacian polynomials

Laplacians of Graphs, Spectra and Laplacian polynomials Laplacians of Graphs, Spectra and Laplacian polynomials Lector: Alexander Mednykh Sobolev Institute of Mathematics Novosibirsk State University Winter School in Harmonic Functions on Graphs and Combinatorial

More information

Probability & Computing

Probability & Computing Probability & Computing Stochastic Process time t {X t t 2 T } state space Ω X t 2 state x 2 discrete time: T is countable T = {0,, 2,...} discrete space: Ω is finite or countably infinite X 0,X,X 2,...

More information

Graph Sparsifiers. Smaller graph that (approximately) preserves the values of some set of graph parameters. Graph Sparsification

Graph Sparsifiers. Smaller graph that (approximately) preserves the values of some set of graph parameters. Graph Sparsification Graph Sparsifiers Smaller graph that (approximately) preserves the values of some set of graph parameters Graph Sparsifiers Spanners Emulators Small stretch spanning trees Vertex sparsifiers Spectral sparsifiers

More information

Graph Sparsifiers: A Survey

Graph Sparsifiers: A Survey Graph Sparsifiers: A Survey Nick Harvey UBC Based on work by: Batson, Benczur, de Carli Silva, Fung, Hariharan, Harvey, Karger, Panigrahi, Sato, Spielman, Srivastava and Teng Approximating Dense Objects

More information

Spectral Graph Theory and its Applications. Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity

Spectral Graph Theory and its Applications. Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity Spectral Graph Theory and its Applications Daniel A. Spielman Dept. of Computer Science Program in Applied Mathematics Yale Unviersity Outline Adjacency matrix and Laplacian Intuition, spectral graph drawing

More information

Groups and the Buckyball

Groups and the Buckyball Groups and the Buckyball Fan R. K. Chung, Bertram Kostant and Shlomo Sternberg 1 Introduction The purpose of this article is to collect a number of remarkable group theoretical facts having to do with

More information

1 T 1 = where 1 is the all-ones vector. For the upper bound, let v 1 be the eigenvector corresponding. u:(u,v) E v 1(u)

1 T 1 = where 1 is the all-ones vector. For the upper bound, let v 1 be the eigenvector corresponding. u:(u,v) E v 1(u) CME 305: Discrete Mathematics and Algorithms Instructor: Reza Zadeh (rezab@stanford.edu) Final Review Session 03/20/17 1. Let G = (V, E) be an unweighted, undirected graph. Let λ 1 be the maximum eigenvalue

More information

Spectral Graph Theory

Spectral Graph Theory Spectral raph Theory to appear in Handbook of Linear Algebra, second edition, CCR Press Steve Butler Fan Chung There are many different ways to associate a matrix with a graph (an introduction of which

More information

Laplacian and Random Walks on Graphs

Laplacian and Random Walks on Graphs Laplacian and Random Walks on Graphs Linyuan Lu University of South Carolina Selected Topics on Spectral Graph Theory (II) Nankai University, Tianjin, May 22, 2014 Five talks Selected Topics on Spectral

More information

Graph Partitioning Using Random Walks

Graph Partitioning Using Random Walks Graph Partitioning Using Random Walks A Convex Optimization Perspective Lorenzo Orecchia Computer Science Why Spectral Algorithms for Graph Problems in practice? Simple to implement Can exploit very efficient

More information

Graph Sparsification I : Effective Resistance Sampling

Graph Sparsification I : Effective Resistance Sampling Graph Sparsification I : Effective Resistance Sampling Nikhil Srivastava Microsoft Research India Simons Institute, August 26 2014 Graphs G G=(V,E,w) undirected V = n w: E R + Sparsification Approximate

More information

Laplacians of Graphs, Spectra and Laplacian polynomials

Laplacians of Graphs, Spectra and Laplacian polynomials Mednykh A. D. (Sobolev Institute of Math) Laplacian for Graphs 27 June - 03 July 2015 1 / 30 Laplacians of Graphs, Spectra and Laplacian polynomials Alexander Mednykh Sobolev Institute of Mathematics Novosibirsk

More information

Markov Chains, Random Walks on Graphs, and the Laplacian

Markov Chains, Random Walks on Graphs, and the Laplacian Markov Chains, Random Walks on Graphs, and the Laplacian CMPSCI 791BB: Advanced ML Sridhar Mahadevan Random Walks! There is significant interest in the problem of random walks! Markov chain analysis! Computer

More information

Using Laplacian Eigenvalues and Eigenvectors in the Analysis of Frequency Assignment Problems

Using Laplacian Eigenvalues and Eigenvectors in the Analysis of Frequency Assignment Problems Using Laplacian Eigenvalues and Eigenvectors in the Analysis of Frequency Assignment Problems Jan van den Heuvel and Snežana Pejić Department of Mathematics London School of Economics Houghton Street,

More information

Lecture 1: Graphs, Adjacency Matrices, Graph Laplacian

Lecture 1: Graphs, Adjacency Matrices, Graph Laplacian Lecture 1: Graphs, Adjacency Matrices, Graph Laplacian Radu Balan January 31, 2017 G = (V, E) An undirected graph G is given by two pieces of information: a set of vertices V and a set of edges E, G =

More information

Math 443/543 Graph Theory Notes 5: Graphs as matrices, spectral graph theory, and PageRank

Math 443/543 Graph Theory Notes 5: Graphs as matrices, spectral graph theory, and PageRank Math 443/543 Graph Theory Notes 5: Graphs as matrices, spectral graph theory, and PageRank David Glickenstein November 3, 4 Representing graphs as matrices It will sometimes be useful to represent graphs

More information

Data Mining and Analysis: Fundamental Concepts and Algorithms

Data Mining and Analysis: Fundamental Concepts and Algorithms Data Mining and Analysis: Fundamental Concepts and Algorithms dataminingbook.info Mohammed J. Zaki 1 Wagner Meira Jr. 2 1 Department of Computer Science Rensselaer Polytechnic Institute, Troy, NY, USA

More information

Spectral Graph Theory and its Applications

Spectral Graph Theory and its Applications Spectral Graph Theory and its Applications Yi-Hsuan Lin Abstract This notes were given in a series of lectures by Prof Fan Chung in National Taiwan University Introduction Basic notations Let G = (V, E)

More information

Solving Local Linear Systems with Boundary Conditions Using Heat Kernel Pagerank

Solving Local Linear Systems with Boundary Conditions Using Heat Kernel Pagerank olving Local Linear ystems with Boundary Conditions Using Heat Kernel Pagerank Fan Chung and Olivia impson Department of Computer cience and Engineering, University of California, an Diego La Jolla, CA

More information

Data Analysis and Manifold Learning Lecture 3: Graphs, Graph Matrices, and Graph Embeddings

Data Analysis and Manifold Learning Lecture 3: Graphs, Graph Matrices, and Graph Embeddings Data Analysis and Manifold Learning Lecture 3: Graphs, Graph Matrices, and Graph Embeddings Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline

More information

Learning from Sensor Data: Set II. Behnaam Aazhang J.S. Abercombie Professor Electrical and Computer Engineering Rice University

Learning from Sensor Data: Set II. Behnaam Aazhang J.S. Abercombie Professor Electrical and Computer Engineering Rice University Learning from Sensor Data: Set II Behnaam Aazhang J.S. Abercombie Professor Electrical and Computer Engineering Rice University 1 6. Data Representation The approach for learning from data Probabilistic

More information

Global Positioning from Local Distances

Global Positioning from Local Distances Global Positioning from Local Distances Amit Singer Yale University, Department of Mathematics, Program in Applied Mathematics SIAM Annual Meeting, July 2008 Amit Singer (Yale University) San Diego 1 /

More information

4 Unitary representation

4 Unitary representation 4 Unitary representation In this section, we consider the following problem: Question 4.1. Which irreducible representations in Theorem 3.7 are unitary? 4.1 Inner product Definition 4.. Let W be a vector

More information

Diffusion and random walks on graphs

Diffusion and random walks on graphs Diffusion and random walks on graphs Leonid E. Zhukov School of Data Analysis and Artificial Intelligence Department of Computer Science National Research University Higher School of Economics Structural

More information

Outline. The Distance Spectra of Cayley Graphs of Coxeter Groups. Finite Reflection Group. Root Systems. Reflection/Coxeter Groups.

Outline. The Distance Spectra of Cayley Graphs of Coxeter Groups. Finite Reflection Group. Root Systems. Reflection/Coxeter Groups. The Distance Spectra of of Coxeter Groups Paul Renteln 1 Department of Physics California State University San Bernardino 2 Department of Mathematics California Institute of Technology ECNU, Shanghai June

More information

2 Fundamentals. 2.1 Graph Theory. Ulrik Brandes and Thomas Erlebach

2 Fundamentals. 2.1 Graph Theory. Ulrik Brandes and Thomas Erlebach 2 Fundamentals Ulrik Brandes and Thomas Erlebach In this chapter we discuss basic terminology and notation for graphs, some fundamental algorithms, and a few other mathematical preliminaries. We denote

More information

Math 126 Lecture 8. Summary of results Application to the Laplacian of the cube

Math 126 Lecture 8. Summary of results Application to the Laplacian of the cube Math 126 Lecture 8 Summary of results Application to the Laplacian of the cube Summary of key results Every representation is determined by its character. If j is the character of a representation and

More information

Facebook Friends! and Matrix Functions

Facebook Friends! and Matrix Functions Facebook Friends! and Matrix Functions! Graduate Research Day Joint with David F. Gleich, (Purdue), supported by" NSF CAREER 1149756-CCF Kyle Kloster! Purdue University! Network Analysis Use linear algebra

More information

ECEN 689 Special Topics in Data Science for Communications Networks

ECEN 689 Special Topics in Data Science for Communications Networks ECEN 689 Special Topics in Data Science for Communications Networks Nick Duffield Department of Electrical & Computer Engineering Texas A&M University Lecture 8 Random Walks, Matrices and PageRank Graphs

More information

Link Analysis Ranking

Link Analysis Ranking Link Analysis Ranking How do search engines decide how to rank your query results? Guess why Google ranks the query results the way it does How would you do it? Naïve ranking of query results Given query

More information

EIGENVECTOR NORMS MATTER IN SPECTRAL GRAPH THEORY

EIGENVECTOR NORMS MATTER IN SPECTRAL GRAPH THEORY EIGENVECTOR NORMS MATTER IN SPECTRAL GRAPH THEORY Franklin H. J. Kenter Rice University ( United States Naval Academy 206) orange@rice.edu Connections in Discrete Mathematics, A Celebration of Ron Graham

More information

Detecting Sharp Drops in PageRank and a Simplified Local Partitioning Algorithm

Detecting Sharp Drops in PageRank and a Simplified Local Partitioning Algorithm Detecting Sharp Drops in PageRank and a Simplified Local Partitioning Algorithm Reid Andersen and Fan Chung University of California at San Diego, La Jolla CA 92093, USA, fan@ucsd.edu, http://www.math.ucsd.edu/~fan/,

More information

An Introduction to Spectral Graph Theory

An Introduction to Spectral Graph Theory An Introduction to Spectral Graph Theory Mackenzie Wheeler Supervisor: Dr. Gary MacGillivray University of Victoria WheelerM@uvic.ca Outline Outline 1. How many walks are there from vertices v i to v j

More information

Algebraic Representation of Networks

Algebraic Representation of Networks Algebraic Representation of Networks 0 1 2 1 1 0 0 1 2 0 0 1 1 1 1 1 Hiroki Sayama sayama@binghamton.edu Describing networks with matrices (1) Adjacency matrix A matrix with rows and columns labeled by

More information

Markov Chains and MCMC

Markov Chains and MCMC Markov Chains and MCMC Markov chains Let S = {1, 2,..., N} be a finite set consisting of N states. A Markov chain Y 0, Y 1, Y 2,... is a sequence of random variables, with Y t S for all points in time

More information

Laplacian eigenvalues and optimality: II. The Laplacian of a graph. R. A. Bailey and Peter Cameron

Laplacian eigenvalues and optimality: II. The Laplacian of a graph. R. A. Bailey and Peter Cameron Laplacian eigenvalues and optimality: II. The Laplacian of a graph R. A. Bailey and Peter Cameron London Taught Course Centre, June 2012 The Laplacian of a graph This lecture will be about the Laplacian

More information

Spectral Graph Theory and You: Matrix Tree Theorem and Centrality Metrics

Spectral Graph Theory and You: Matrix Tree Theorem and Centrality Metrics Spectral Graph Theory and You: and Centrality Metrics Jonathan Gootenberg March 11, 2013 1 / 19 Outline of Topics 1 Motivation Basics of Spectral Graph Theory Understanding the characteristic polynomial

More information

Data Analysis and Manifold Learning Lecture 7: Spectral Clustering

Data Analysis and Manifold Learning Lecture 7: Spectral Clustering Data Analysis and Manifold Learning Lecture 7: Spectral Clustering Radu Horaud INRIA Grenoble Rhone-Alpes, France Radu.Horaud@inrialpes.fr http://perception.inrialpes.fr/ Outline of Lecture 7 What is spectral

More information

A New Spectral Technique Using Normalized Adjacency Matrices for Graph Matching 1

A New Spectral Technique Using Normalized Adjacency Matrices for Graph Matching 1 CHAPTER-3 A New Spectral Technique Using Normalized Adjacency Matrices for Graph Matching Graph matching problem has found many applications in areas as diverse as chemical structure analysis, pattern

More information

Algorithms, Graph Theory, and the Solu7on of Laplacian Linear Equa7ons. Daniel A. Spielman Yale University

Algorithms, Graph Theory, and the Solu7on of Laplacian Linear Equa7ons. Daniel A. Spielman Yale University Algorithms, Graph Theory, and the Solu7on of Laplacian Linear Equa7ons Daniel A. Spielman Yale University Rutgers, Dec 6, 2011 Outline Linear Systems in Laplacian Matrices What? Why? Classic ways to solve

More information

The heat kernel as the pagerank of a graph

The heat kernel as the pagerank of a graph Classification: Physical Sciences, Mathematics The heat kernel as the pagerank of a graph by Fan Chung Department of Mathematics University of California at San Diego La Jolla, CA 9093-011 Corresponding

More information

A case study in Interaction cohomology Oliver Knill, 3/18/2016

A case study in Interaction cohomology Oliver Knill, 3/18/2016 A case study in Interaction cohomology Oliver Knill, 3/18/2016 Simplicial Cohomology Simplicial cohomology is defined by an exterior derivative df (x) = F (dx) on valuation forms F (x) on subgraphs x of

More information

Decomposition of random graphs into complete bipartite graphs

Decomposition of random graphs into complete bipartite graphs Decomposition of random graphs into complete bipartite graphs Fan Chung Xing Peng Abstract We consider the problem of partitioning the edge set of a graph G into the minimum number τg) of edge-disjoint

More information

Lecture 1 and 2: Random Spanning Trees

Lecture 1 and 2: Random Spanning Trees Recent Advances in Approximation Algorithms Spring 2015 Lecture 1 and 2: Random Spanning Trees Lecturer: Shayan Oveis Gharan March 31st Disclaimer: These notes have not been subjected to the usual scrutiny

More information

Algorithms, Graph Theory, and Linear Equa9ons in Laplacians. Daniel A. Spielman Yale University

Algorithms, Graph Theory, and Linear Equa9ons in Laplacians. Daniel A. Spielman Yale University Algorithms, Graph Theory, and Linear Equa9ons in Laplacians Daniel A. Spielman Yale University Shang- Hua Teng Carter Fussell TPS David Harbater UPenn Pavel Cur9s Todd Knoblock CTY Serge Lang 1927-2005

More information

Hyperlinked-Induced Topic Search (HITS) identifies. authorities as good content sources (~high indegree) HITS [Kleinberg 99] considers a web page

Hyperlinked-Induced Topic Search (HITS) identifies. authorities as good content sources (~high indegree) HITS [Kleinberg 99] considers a web page IV.3 HITS Hyperlinked-Induced Topic Search (HITS) identifies authorities as good content sources (~high indegree) hubs as good link sources (~high outdegree) HITS [Kleinberg 99] considers a web page a

More information

Data-dependent representations: Laplacian Eigenmaps

Data-dependent representations: Laplacian Eigenmaps Data-dependent representations: Laplacian Eigenmaps November 4, 2015 Data Organization and Manifold Learning There are many techniques for Data Organization and Manifold Learning, e.g., Principal Component

More information

Graph Sparsification III: Ramanujan Graphs, Lifts, and Interlacing Families

Graph Sparsification III: Ramanujan Graphs, Lifts, and Interlacing Families Graph Sparsification III: Ramanujan Graphs, Lifts, and Interlacing Families Nikhil Srivastava Microsoft Research India Simons Institute, August 27, 2014 The Last Two Lectures Lecture 1. Every weighted

More information

x 3y 2z = 6 1.2) 2x 4y 3z = 8 3x + 6y + 8z = 5 x + 3y 2z + 5t = 4 1.5) 2x + 8y z + 9t = 9 3x + 5y 12z + 17t = 7

x 3y 2z = 6 1.2) 2x 4y 3z = 8 3x + 6y + 8z = 5 x + 3y 2z + 5t = 4 1.5) 2x + 8y z + 9t = 9 3x + 5y 12z + 17t = 7 Linear Algebra and its Applications-Lab 1 1) Use Gaussian elimination to solve the following systems x 1 + x 2 2x 3 + 4x 4 = 5 1.1) 2x 1 + 2x 2 3x 3 + x 4 = 3 3x 1 + 3x 2 4x 3 2x 4 = 1 x + y + 2z = 4 1.4)

More information

LINK ANALYSIS. Dr. Gjergji Kasneci Introduction to Information Retrieval WS

LINK ANALYSIS. Dr. Gjergji Kasneci Introduction to Information Retrieval WS LINK ANALYSIS Dr. Gjergji Kasneci Introduction to Information Retrieval WS 2012-13 1 Outline Intro Basics of probability and information theory Retrieval models Retrieval evaluation Link analysis Models

More information

Constant Arboricity Spectral Sparsifiers

Constant Arboricity Spectral Sparsifiers Constant Arboricity Spectral Sparsifiers arxiv:1808.0566v1 [cs.ds] 16 Aug 018 Timothy Chu oogle timchu@google.com Jakub W. Pachocki Carnegie Mellon University pachocki@cs.cmu.edu August 0, 018 Abstract

More information

Beyond Scalar Affinities for Network Analysis or Vector Diffusion Maps and the Connection Laplacian

Beyond Scalar Affinities for Network Analysis or Vector Diffusion Maps and the Connection Laplacian Beyond Scalar Affinities for Network Analysis or Vector Diffusion Maps and the Connection Laplacian Amit Singer Princeton University Department of Mathematics and Program in Applied and Computational Mathematics

More information

Isoperimetric inequalities for cartesian products of graphs

Isoperimetric inequalities for cartesian products of graphs Isoperimetric inequalities for cartesian products of graphs F. R. K. Chung University of Pennsylvania Philadelphia 19104 Prasad Tetali School of Mathematics Georgia Inst. of Technology Atlanta GA 3033-0160

More information

Decomposition of random graphs into complete bipartite graphs

Decomposition of random graphs into complete bipartite graphs Decomposition of random graphs into complete bipartite graphs Fan Chung Xing Peng Abstract We consider the problem of partitioning the edge set of a graph G into the minimum number τg of edge-disjoint

More information

Linear algebra and applications to graphs Part 1

Linear algebra and applications to graphs Part 1 Linear algebra and applications to graphs Part 1 Written up by Mikhail Belkin and Moon Duchin Instructor: Laszlo Babai June 17, 2001 1 Basic Linear Algebra Exercise 1.1 Let V and W be linear subspaces

More information

The Interlace Polynomial of Graphs at 1

The Interlace Polynomial of Graphs at 1 The Interlace Polynomial of Graphs at 1 PN Balister B Bollobás J Cutler L Pebody July 3, 2002 Department of Mathematical Sciences, University of Memphis, Memphis, TN 38152 USA Abstract In this paper we

More information

Algorithmic Primitives for Network Analysis: Through the Lens of the Laplacian Paradigm

Algorithmic Primitives for Network Analysis: Through the Lens of the Laplacian Paradigm Algorithmic Primitives for Network Analysis: Through the Lens of the Laplacian Paradigm Shang-Hua Teng Computer Science, Viterbi School of Engineering USC Massive Data and Massive Graphs 500 billions web

More information

Expansion properties of random Cayley graphs and vertex transitive graphs via matrix martingales

Expansion properties of random Cayley graphs and vertex transitive graphs via matrix martingales Expansion properties of random Cayley graphs and vertex transitive graphs via matrix martingales Demetres Christofides Klas Markström Abstract The Alon-Roichman theorem states that for every ε > 0 there

More information

Topics in Graph Theory

Topics in Graph Theory Topics in Graph Theory September 4, 2018 1 Preliminaries A graph is a system G = (V, E) consisting of a set V of vertices and a set E (disjoint from V ) of edges, together with an incidence function End

More information

Spectra of Adjacency and Laplacian Matrices

Spectra of Adjacency and Laplacian Matrices Spectra of Adjacency and Laplacian Matrices Definition: University of Alicante (Spain) Matrix Computing (subject 3168 Degree in Maths) 30 hours (theory)) + 15 hours (practical assignment) Contents 1. Spectra

More information

Spectra of Digraph Transformations

Spectra of Digraph Transformations Spectra of Digraph Transformations Aiping Deng a,, Alexander Kelmans b,c a Department of Applied Mathematics, Donghua University, 201620 Shanghai, China arxiv:1707.00401v1 [math.co] 3 Jul 2017 b Department

More information

Large incidence-free sets in geometries

Large incidence-free sets in geometries Large incidence-free sets in geometries Stefaan De Winter Department of Mathematical Sciences Michigan Technological University Michigan, U.S.A. Jeroen Schillewaert sgdewint@mtu.edu Jacques Verstraete

More information

An Algorithmist s Toolkit September 10, Lecture 1

An Algorithmist s Toolkit September 10, Lecture 1 18.409 An Algorithmist s Toolkit September 10, 2009 Lecture 1 Lecturer: Jonathan Kelner Scribe: Jesse Geneson (2009) 1 Overview The class s goals, requirements, and policies were introduced, and topics

More information

Diameter of random spanning trees in a given graph

Diameter of random spanning trees in a given graph Diameter of random spanning trees in a given graph Fan Chung Paul Horn Linyuan Lu June 30, 008 Abstract We show that a random spanning tree formed in a general graph G (such as a power law graph) has diameter

More information

1.3 Vertex Degrees. Vertex Degree for Undirected Graphs: Let G be an undirected. Vertex Degree for Digraphs: Let D be a digraph and y V (D).

1.3 Vertex Degrees. Vertex Degree for Undirected Graphs: Let G be an undirected. Vertex Degree for Digraphs: Let D be a digraph and y V (D). 1.3. VERTEX DEGREES 11 1.3 Vertex Degrees Vertex Degree for Undirected Graphs: Let G be an undirected graph and x V (G). The degree d G (x) of x in G: the number of edges incident with x, each loop counting

More information

arxiv: v1 [math.co] 27 Jul 2012

arxiv: v1 [math.co] 27 Jul 2012 Harnack inequalities for graphs with non-negative Ricci curvature arxiv:1207.6612v1 [math.co] 27 Jul 2012 Fan Chung University of California, San Diego S.-T. Yau Harvard University May 2, 2014 Abstract

More information

U.C. Berkeley CS270: Algorithms Lecture 21 Professor Vazirani and Professor Rao Last revised. Lecture 21

U.C. Berkeley CS270: Algorithms Lecture 21 Professor Vazirani and Professor Rao Last revised. Lecture 21 U.C. Berkeley CS270: Algorithms Lecture 21 Professor Vazirani and Professor Rao Scribe: Anupam Last revised Lecture 21 1 Laplacian systems in nearly linear time Building upon the ideas introduced in the

More information

CMPSCI 791BB: Advanced ML: Laplacian Learning

CMPSCI 791BB: Advanced ML: Laplacian Learning CMPSCI 791BB: Advanced ML: Laplacian Learning Sridhar Mahadevan Outline! Spectral graph operators! Combinatorial graph Laplacian! Normalized graph Laplacian! Random walks! Machine learning on graphs! Clustering!

More information

arxiv: v1 [math.gr] 1 Jul 2018

arxiv: v1 [math.gr] 1 Jul 2018 Rhomboidal C 4 C 8 toris which are Cayley graphs F. Afshari, M. Maghasedi Department of Mathematics, Karaj Branch, Islamic Azad University, Karaj, Iran arxiv:87.78v [math.gr] Jul 8 Abstract A C 4 C 8 net

More information

Eigenvalues of graphs

Eigenvalues of graphs Eigenvalues of graphs F. R. K. Chung University of Pennsylvania Philadelphia PA 19104 September 11, 1996 1 Introduction The study of eigenvalues of graphs has a long history. From the early days, representation

More information

Eigenvalues of the Adjacency Tensor on Products of Hypergraphs

Eigenvalues of the Adjacency Tensor on Products of Hypergraphs Int. J. Contemp. Math. Sciences, Vol. 8, 201, no. 4, 151-158 HIKARI Ltd, www.m-hikari.com Eigenvalues of the Adjacency Tensor on Products of Hypergraphs Kelly J. Pearson Department of Mathematics and Statistics

More information

Quantum Physics II (8.05) Fall 2002 Assignment 3

Quantum Physics II (8.05) Fall 2002 Assignment 3 Quantum Physics II (8.05) Fall 00 Assignment Readings The readings below will take you through the material for Problem Sets and 4. Cohen-Tannoudji Ch. II, III. Shankar Ch. 1 continues to be helpful. Sakurai

More information

Simple eigenvalues of graphs and digraphs

Simple eigenvalues of graphs and digraphs Simple eigenvalues of graphs and digraphs by Krystal Guo M.Maths., University of Waterloo, 2010 B.Maths., University of Waterloo, 2009 Dissertation Submitted in Partial Fulfillment of the Requirements

More information

Petar Maymounkov Problem Set 1 MIT with Prof. Jonathan Kelner, Fall 07

Petar Maymounkov Problem Set 1 MIT with Prof. Jonathan Kelner, Fall 07 Petar Maymounkov Problem Set 1 MIT 18.409 with Prof. Jonathan Kelner, Fall 07 Problem 1.2. Part a): Let u 1,..., u n and v 1,..., v n be the eigenvalues of A and B respectively. Since A B is positive semi-definite,

More information

1.10 Matrix Representation of Graphs

1.10 Matrix Representation of Graphs 42 Basic Concepts of Graphs 1.10 Matrix Representation of Graphs Definitions: In this section, we introduce two kinds of matrix representations of a graph, that is, the adjacency matrix and incidence matrix

More information

A Little Beyond: Linear Algebra

A Little Beyond: Linear Algebra A Little Beyond: Linear Algebra Akshay Tiwary March 6, 2016 Any suggestions, questions and remarks are welcome! 1 A little extra Linear Algebra 1. Show that any set of non-zero polynomials in [x], no two

More information

Laplacian Integral Graphs with Maximum Degree 3

Laplacian Integral Graphs with Maximum Degree 3 Laplacian Integral Graphs with Maximum Degree Steve Kirkland Department of Mathematics and Statistics University of Regina Regina, Saskatchewan, Canada S4S 0A kirkland@math.uregina.ca Submitted: Nov 5,

More information

Stein s Method for Matrix Concentration

Stein s Method for Matrix Concentration Stein s Method for Matrix Concentration Lester Mackey Collaborators: Michael I. Jordan, Richard Y. Chen, Brendan Farrell, and Joel A. Tropp University of California, Berkeley California Institute of Technology

More information

Du Val Singularities

Du Val Singularities Du Val Singularities Igor Burban Introduction We consider quotient singularities C 2 /G, where G SL 2 (C) is a finite subgroup. Since we have a finite ring extension C[[x, y]] G C[[x, y]], the Krull dimension

More information

The Adjacency Matrix, Standard Laplacian, and Normalized Laplacian, and Some Eigenvalue Interlacing Results

The Adjacency Matrix, Standard Laplacian, and Normalized Laplacian, and Some Eigenvalue Interlacing Results The Adjacency Matrix, Standard Laplacian, and Normalized Laplacian, and Some Eigenvalue Interlacing Results Frank J. Hall Department of Mathematics and Statistics Georgia State University Atlanta, GA 30303

More information

HARMONIC MAPS INTO GRASSMANNIANS AND A GENERALIZATION OF DO CARMO-WALLACH THEOREM

HARMONIC MAPS INTO GRASSMANNIANS AND A GENERALIZATION OF DO CARMO-WALLACH THEOREM Proceedings of the 16th OCU International Academic Symposium 2008 OCAMI Studies Volume 3 2009, pp.41 52 HARMONIC MAPS INTO GRASSMANNIANS AND A GENERALIZATION OF DO CARMO-WALLACH THEOREM YASUYUKI NAGATOMO

More information

Solving systems of linear equations

Solving systems of linear equations Chapter 5 Solving systems of linear equations Let A R n n and b R n be given. We want to solve Ax = b. What is the fastest method you know? We usually learn how to solve systems of linear equations via

More information

Lecture 7: February 6

Lecture 7: February 6 CS271 Randomness & Computation Spring 2018 Instructor: Alistair Sinclair Lecture 7: February 6 Disclaimer: These notes have not been subjected to the usual scrutiny accorded to formal publications. They

More information

The Zeta Function of a Cyclic Language with Connections to Elliptic Curves and Chip-Firing Games

The Zeta Function of a Cyclic Language with Connections to Elliptic Curves and Chip-Firing Games The Zeta Function of a Cyclic Language with Connections to Elliptic Curves and Chip-Firing Games University of California, San Diego March 14, 2007 University of California, San Diego Slide 1 OUTLINE I.

More information

Extract useful building blocks: blobs. the same image like for the corners

Extract useful building blocks: blobs. the same image like for the corners Extract useful building blocks: blobs the same image like for the corners Here were the corners... Blob detection in 2D Laplacian of Gaussian: Circularly symmetric operator for blob detection in 2D 2 g=

More information

Molecular Symmetry 10/25/2018

Molecular Symmetry 10/25/2018 Symmetry helps us understand molecular structure, some chemical properties, and characteristics of physical properties (spectroscopy). Predict IR spectra or Interpret UV-Vis spectra Predict optical activity

More information

A Bound on the Number of Spanning Trees in Bipartite Graphs

A Bound on the Number of Spanning Trees in Bipartite Graphs A Bound on the Number of Spanning Trees in Bipartite Graphs Cheng Wai Koo Mohamed Omar, Advisor Nicholas Pippenger, Reader Department of Mathematics May, 2016 Copyright 2016 Cheng Wai Koo. The author grants

More information

Reductive group actions and some problems concerning their quotients

Reductive group actions and some problems concerning their quotients Reductive group actions and some problems concerning their quotients Brandeis University January 2014 Linear Algebraic Groups A complex linear algebraic group G is an affine variety such that the mappings

More information

Homework 1. Yuan Yao. September 18, 2011

Homework 1. Yuan Yao. September 18, 2011 Homework 1 Yuan Yao September 18, 2011 1. Singular Value Decomposition: The goal of this exercise is to refresh your memory about the singular value decomposition and matrix norms. A good reference to

More information

The giant component in a random subgraph of a given graph

The giant component in a random subgraph of a given graph The giant component in a random subgraph of a given graph Fan Chung, Paul Horn, and Linyuan Lu 2 University of California, San Diego 2 University of South Carolina Abstract We consider a random subgraph

More information

Notes on Linear Algebra and Matrix Theory

Notes on Linear Algebra and Matrix Theory Massimo Franceschet featuring Enrico Bozzo Scalar product The scalar product (a.k.a. dot product or inner product) of two real vectors x = (x 1,..., x n ) and y = (y 1,..., y n ) is not a vector but a

More information

Homogeneous structures

Homogeneous structures Homogeneous structures Robert Gray Centro de Álgebra da Universidade de Lisboa Lisboa, April 2010 Group theory A group is... a set G with a binary operation G G G, (x, y) xy, written multiplicatively,

More information

2.1 Laplacian Variants

2.1 Laplacian Variants -3 MS&E 337: Spectral Graph heory and Algorithmic Applications Spring 2015 Lecturer: Prof. Amin Saberi Lecture 2-3: 4/7/2015 Scribe: Simon Anastasiadis and Nolan Skochdopole Disclaimer: hese notes have

More information

Infinitesimal Einstein Deformations. Kähler Manifolds

Infinitesimal Einstein Deformations. Kähler Manifolds on Nearly Kähler Manifolds (joint work with P.-A. Nagy and U. Semmelmann) Gemeinsame Jahrestagung DMV GDM Berlin, March 30, 2007 Nearly Kähler manifolds Definition and first properties Examples of NK manifolds

More information