Generalizing the degree sequence problem



Similar documents
Luby s Alg. for Maximal Independent Sets using Pairwise Independence

The Greedy Method. Introduction. 0/1 Knapsack Problem

1 Example 1: Axis-aligned rectangles

BERNSTEIN POLYNOMIALS

v a 1 b 1 i, a 2 b 2 i,..., a n b n i.

Recurrence. 1 Definitions and main statements

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

Rate Monotonic (RM) Disadvantages of cyclic. TDDB47 Real Time Systems. Lecture 2: RM & EDF. Priority-based scheduling. States of a process

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

n + d + q = 24 and.05n +.1d +.25q = 2 { n + d + q = 24 (3) n + 2d + 5q = 40 (2)

Extending Probabilistic Dynamic Epistemic Logic

Ring structure of splines on triangulations

Support Vector Machines

General Auction Mechanism for Search Advertising

This circuit than can be reduced to a planar circuit

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

On Leonid Gurvits s proof for permanents

On Lockett pairs and Lockett conjecture for π-soluble Fitting classes

Embedding lattices in the Kleene degrees

What is Candidate Sampling

PERRON FROBENIUS THEOREM

NMT EE 589 & UNM ME 482/582 ROBOT ENGINEERING. Dr. Stephen Bruder NMT EE 589 & UNM ME 482/582

REGULAR MULTILINEAR OPERATORS ON C(K) SPACES

Finite Math Chapter 10: Study Guide and Solution to Problems

NON-CONSTANT SUM RED-AND-BLACK GAMES WITH BET-DEPENDENT WIN PROBABILITY FUNCTION LAURA PONTIGGIA, University of the Sciences in Philadelphia

Level Annuities with Payments Less Frequent than Each Interest Period

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

Production. 2. Y is closed A set is closed if it contains its boundary. We need this for the solution existence in the profit maximization problem.

The OC Curve of Attribute Acceptance Plans

Efficient Project Portfolio as a tool for Enterprise Risk Management

Forecasting the Direction and Strength of Stock Market Movement

An Alternative Way to Measure Private Equity Performance

Product-Form Stationary Distributions for Deficiency Zero Chemical Reaction Networks

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

Project Networks With Mixed-Time Constraints

Compiling for Parallelism & Locality. Dependence Testing in General. Algorithms for Solving the Dependence Problem. Dependence Testing

8 Algorithm for Binary Searching in Trees

Portfolio Loss Distribution

NPAR TESTS. One-Sample Chi-Square Test. Cell Specification. Observed Frequencies 1O i 6. Expected Frequencies 1EXP i 6

Logical Development Of Vogel s Approximation Method (LD-VAM): An Approach To Find Basic Feasible Solution Of Transportation Problem

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt.

Availability-Based Path Selection and Network Vulnerability Assessment

DEFINING %COMPLETE IN MICROSOFT PROJECT

J. Parallel Distrib. Comput.

Fast degree elevation and knot insertion for B-spline curves

Complete Fairness in Secure Two-Party Computation

COLLOQUIUM MATHEMATICUM

Loop Parallelization

The Full-Wave Rectifier

When Network Effect Meets Congestion Effect: Leveraging Social Services for Wireless Services

Nordea G10 Alpha Carry Index

Minimal Coding Network With Combinatorial Structure For Instantaneous Recovery From Edge Failures

Every tree contains a large induced subgraph with all degrees odd

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM

Feasibility of Using Discriminate Pricing Schemes for Energy Trading in Smart Grid

8.4. Annuities: Future Value. INVESTIGATE the Math Annuities: Future Value

7.5. Present Value of an Annuity. Investigate

F-Rational Rings and the Integral Closures of Ideals

Mean Molecular Weight

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits

Calculation of Sampling Weights

FINANCIAL MATHEMATICS. A Practical Guide for Actuaries. and other Business Professionals

The Noether Theorems: from Noether to Ševera

The Distribution of Eigenvalues of Covariance Matrices of Residuals in Analysis of Variance

Ad-Hoc Games and Packet Forwardng Networks

Sngle Snk Buy at Bulk Problem and the Access Network

Connectivity and cuts

RESEARCH ON DUAL-SHAKER SINE VIBRATION CONTROL. Yaoqi FENG 1, Hanping QIU 1. China Academy of Space Technology (CAST)

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT


Mean Value Coordinates for Closed Triangular Meshes

Figure 1. Inventory Level vs. Time - EOQ Problem

To Fill or not to Fill: The Gas Station Problem

Simple Interest Loans (Section 5.1) :

AN EFFECTIVE MATRIX GEOMETRIC MEAN SATISFYING THE ANDO LI MATHIAS PROPERTIES

From Selective to Full Security: Semi-Generic Transformations in the Standard Model

The program for the Bachelor degrees shall extend over three years of full-time study or the parttime equivalent.

Code_Aster ( ) D Charter for the realization of the mathematical formulas in the documentation of the Code_Aster ( )

FINITE HILBERT STABILITY OF (BI)CANONICAL CURVES

Uniform topologies on types

Natural hp-bem for the electric field integral equation with singular solutions

An Overview of Financial Mathematics

How To Assemble The Tangent Spaces Of A Manfold Nto A Coherent Whole

Transport-Problem-Based Algorithm fordynamicload Balancing in Distributed LogicSimulation

Transcription:

Mddlebury College March 2009 Arzona State Unversty Dscrete Mathematcs Semnar

The degree sequence problem Problem: Gven an nteger sequence d = (d 1,...,d n ) determne f there exsts a graph G wth d as ts sequence of degrees. If such a G exsts then d s sad to be graphc, and G s called a realzaton.

An example Is d = (3,3,3,3,3,3) graphc?

An example Is d = (3,3,3,3,3,3) graphc? Havel (1955) and Hakm (1962) gave an algorthm to decde. (3,3,3,3,3,3) (2,2,2,3,3) = (3,3,2,2,2) (2,1,1,2) = (2,2,1,1) (1,0,1) = (1,1,0) (0,0) As (0,0) s graphc, so s the gven.

An example Is d = (3,3,3,3,3,3) graphc? Havel (1955) and Hakm (1962) gave an algorthm to decde. (3,3,3,3,3,3) (2,2,2,3,3) = (3,3,2,2,2) (2,1,1,2) = (2,2,1,1) (1,0,1) = (1,1,0) (0,0) As (0,0) s graphc, so s the gven. To construct a realzaton, work backwards usng smple edge augmentatons.

Erdős-Galla crteron Theorem [Erdős, Galla (1960)] A nonncreasng sequence of nonnegatve ntegers d = (d 1,...,d n ) (n 2) s graphc f, and only f, n =1 d s even and for each nteger k, 1 k n 1, k d k(k 1) + =1 n =k+1 mn{k,d }.

Erdős-Galla crteron Theorem [Erdős, Galla (1960)] A nonncreasng sequence of nonnegatve ntegers d = (d 1,...,d n ) (n 2) s graphc f, and only f, n =1 d s even and for each nteger k, 1 k n 1, k d k(k 1) + =1 n =k+1 mn{k,d }. The degrees of the frst k vertces are absorbed wthn k-subset and the degrees of remanng vertces. A necessary condton whch s also suffcent!

Theorem (Erdős, Galla) For a graphc d, n =1 d 2(n 1) f and only f there exsts a connected G realzng d.

Theorem (Erdős, Galla) For a graphc d, n =1 d 2(n 1) f and only f there exsts a connected G realzng d. Proof: (Suffcency) If there exsts a connected realzaton then G contans a spannng tree. Thus G has n 1 edges and so n =1 d 2(n 1).

Theorem (Erdős, Galla) For a graphc d, n =1 d 2(n 1) f and only f there exsts a connected G realzng d. Proof: (Suffcency) If there exsts a connected realzaton then G contans a spannng tree. Thus G has n 1 edges and so n =1 d 2(n 1). (Necessty) Pck the realzaton of d wth the fewest number of components. If ths number s 1, then we are done. Otherwse one of the components contans a cycle. Performng a smple edge-exchange allows us to move to a realzaton wth fewer components.

For a subgraph F, d s sad to be potentally F-graphc f there exsts a realzaton of d contanng F.

For a subgraph F, d s sad to be potentally F-graphc f there exsts a realzaton of d contanng F. (2,2,2,2,2,2) s potentally K 3 -graphc.

For a subgraph F, d s sad to be potentally F-graphc f there exsts a realzaton of d contanng F. (2,2,2,2,2,2) s potentally K 3 -graphc.

For a subgraph F, d s sad to be potentally F-graphc f there exsts a realzaton of d contanng F. (2,2,2,2,2,2) s potentally K 3 -graphc.

Problem Gven a subgraph F, determne the least even nteger m s.t. Σd m d s potentally F-graphc. Denote m by σ(f, n).

Erdős, Jacobson, Lehel Conjecture Conjecture (EJL - 1991) For n suffcently large, σ(k t,n) = (t 2)(2n t + 1) + 2. Lower bound arses from consderng: + ÃØ ¾ d = ((n 1) t 2,(t 2) n t+2 ) ÃÒ Ø ¾

Erdős, Jacobson, Lehel Conjecture Conjecture settled: t = 3 Erdős, Jacobson, & Lehel(1991), t = 4 Gould, Jacobson, & Lehel(1999), L & Song(1998), t = 5 L & Song(1998), t 6 L, Song, & Luo(1998) t 3 S.(2005), Ferrara, Gould, S. (2009+) - purely graph-theoretc proof. Theorem For n suffcently large, σ(k t,n) = (t 2)(2n t + 1) + 2.

Sketch of our proof Uses nducton on t. Erdős-Galla guarantees enough vertces of hgh degree. Uses noton of an edge-exhange. Edge-exchange allows us to place desred subgraph on vertces of hghest degree and buld K t from smaller clque guaranteed by nductve hypothess.

Extendng the EJL-conjecture to an arbtrary graph F Let F be a forbdden subgraph. Let α(f) denote the ndependence number of F and defne: and u := u(f) = V (F) α(f) 1, s := s(f) = mn{ (H) : H F, H = α(f) + 1}. Consder the followng sequence, π(f,n) = ((n 1) u,(u + s 1) n u ).

A General Lower Bound If F s a subgraph of F then σ(f,n) σ(f,n) for every n. Let σ(π) denote the sum of the terms of π. Proposton (Ferrara, S. - 09) Gven a graph F and n suffcently large then, σ(f,n) max{σ(π(f,n)) + 2 F F } (1) = max{n(2u(f ) + s(f ) 1) F F } (2)

Proof of Lower Bound Proof: Let F F be the subgraph whch acheves the max. Consder, + K u(f ) + An s(f ) regular graph on n u(f ) vertces. u(f ) = V (F ) α(f ) 1 s(f ) = mn{ (H) : H F, H = α(f ) + 1}

A Stronger Lower Bound Let v (H) be the number of vertces of degree n H. Let M (H) denote the set of nduced subgraphs on α + 1 vertces wth v (H) > 0. For all, s α 1 defne: m = mn M (H){vertces of degree at least } n s = m s 1 and n = mn{m 1,n 1 } Fnally, set δ α 1 = n α 1 and for all, s α 2 defne δ = n n +1 and π (F,n) = ((n 1) u,(u + α 1) δ α 1,(u + α 2) δ α 2,... (u + s) δs,(u + s 1) n u Σδ ).

An Example Let F = K 6,6. Then u(k 6,6 ) = 12 6 1 = 5 and s(k 6,6 ) = 4. m 4 = 3 and m 5 = 2 n 4 = m 4 1 = 2 and n 5 = mn{m 5 1,n 4 } = 1 δ 5 = n 5 = 1 and δ 4 = n 5 n 4 = 1 Thus, π (K 6,6,n) = ((n 1) 5,10,9,8 n 7 )

A Stronger Lower Bound Theorem (Ferrara, S. - 09) Gven a graph F and n suffcently large then, σ(f,n) max{σ(π (F,n)) + 2 F F }

When Does Equalty Hold? clques complete bpartte graphs Chen, L, Yn 04; Gould, Jacobson, Lehel 99; L, Yn 02 complete multpartte graphs Chen, Yn 08; G. Chen, Ferrara, Gould, S. 08; Ferrara, Gould, S. 08 matchngs Gould, Jacobson, Lehel 99 cycles La 04 (generalzed) frendshp graph Ferrara, Gould, S. 06, (Chen, S., Yn 08) clque mnus an edge La 01; L, Mao, Yn 05 dsjont unon of clques Ferrara 08

Conjecture Gven a graph F and n suffcently large then, σ(f,n) = max{σ(π (F,n)) + 2 F F } Conjecture (weaker verson) Gven a graph F, let ǫ > 0. Then there exsts an n 0 = n 0 (ǫ,f) such that for any n > n 0 σ(f,n) max{(n(2u(f ) + d(f ) 1 + ǫ) F F }.

Conjecture Gven a graph F and n suffcently large then, σ(f,n) = max{σ(π (F,n)) + 2 F F } Conjecture (weaker verson) Gven a graph F, let ǫ > 0. Then there exsts an n 0 = n 0 (ǫ,f) such that for any n > n 0 σ(f,n) max{(n(2u(f ) + d(f ) 1 + ǫ) F F }. Conjecture (strong form) holds for graphs wth ndependence number 2 (Ferrara, S. - 09)

An example of the generalzed problem Is the followng graphc? < V,d,D >=< {V 1,V 2 },(5 4,3 8 ), [ 6 8 8 8 ] >

An example of the generalzed problem Is the followng graphc? < V,d,D >=< {V 1,V 2 },(5 4,3 8 ), [ 6 8 8 8 ] >

An example of the generalzed problem Is the followng graphc? < V,d,D >=< {V 1,V 2 },(5 4,3 8 ), [ 6 8 8 8 ] > V 1 V 2

An example of the generalzed problem Is the followng graphc? < V,d,D >=< {V 1,V 2 },(5 4,3 8 ), [ 6 8 8 8 ] > V 1 V 2

An example of the generalzed problem Is the followng graphc? < V,d,D >=< {V 1,V 2 },(5 4,3 8 ), [ 6 8 8 8 ] > V 1 V 2

Let d = (d v 1 1,dv 2 2,...,dv k k ) where v = V and so V s the set of vertces of degree d. Let V = {V 1,...,V k }. Let D = (d j ) be a k k matrx, wth d j denotng the number of edges between V and V j ; d s the number of edges contaned entrely wthn V.

Let d = (d v 1 1,dv 2 2,...,dv k k ) where v = V and so V s the set of vertces of degree d. Let V = {V 1,...,V k }. Let D = (d j ) be a k k matrx, wth d j denotng the number of edges between V and V j ; d s the number of edges contaned entrely wthn V. Jont degree-matrx graphc realzaton problem Gven < V,d,D >, decde whether a smple graph G exsts such that, for all, each vertex n V has degree d, and, for j, there are exactly d j edges between V and V j, whle, for all, there are exactly d edges contaned n V.

Amanatds, Green and Mhal (AGM) have shown that the followng natural necessary condtons for a realzaton to exst are also suffcent. The condtons are: Degree feasblty: 2d + Σ j [k],j d j = v d, for all 1 k, and Matrx feasblty: D s a symmetrc matrx wth non-negatve ntegral entres, d j v v j, for all 1 k, and d ( v 2), for all 1 k.

AGM s algorthmc proof Algorthm rests on a balanced degree nvarant. It starts wth the empty graph and adds one edge at a tme whle keepng the dfference between any two vertex degrees n a gven V to at most 1. Whle there exsts some,j such that d j s not satsfed the algorthm adds an edge between V and V j.

AGM algorthm G p M M j N N j V V j

AGM algorthm G p M M j N u v N j V V j

AGM algorthm G p+1 M M j N u v N j V V j

AGM algorthm G p M v M j N u N j V V j

AGM algorthm G p M v M j N u N j V V j

AGM algorthm G p+1 M N u V v V j M j N j

AGM algorthm G p M u v M j N N j V V j

AGM algorthm G p M u v M j N N j V V j

AGM algorthm G p+1 M u v M j N x y N j V V j x may equal y

AGM algorthm G p M N u v V

AGM algorthm G p+1 M N u v V

AGM algorthm G p M N v u V

AGM algorthm G p+1 M N v u If N =1 V

AGM algorthm G p M N v u V If N > 1

AGM algorthm G p+1 M N v u V If N > 1

AGM algorthm G p M N v u V

AGM algorthm G p M N v u V If N =1

AGM algorthm G p+1 M N v u V If N =1

AGM algorthm G p M N v u If N >1 V

AGM algorthm G p+1 M v u If N >1 N V

Theorem (Jont Degree-Matrx Realzaton Theorem - AGM) Gven < V,d,D >, f degree and matrx feasblty hold, then a graph G exsts that realzes < V, d, D >. Furthermore, such a graph can be constructed n tme polynomal n n.

Can one prove our conjectures usng the Jont Degree-Matrx Theorem?

Can we prove our conjectures usng the AGM-result? Is d = (10 6,4 8 ) potentally K 6 -graphc?

Can we prove our conjectures usng the AGM-result? Is d = (10 6,4 8 ) potentally K 6 -graphc? If a realzaton of d exsts that contans a copy of K 6, then, t s known, a realzaton exsts n whch the copy of K 6 les on the vertces of hghest degree.

Can we prove our conjectures usng the AGM-result? Is d = (10 6,4 8 ) potentally K 6 -graphc? If a realzaton of d exsts that contans a copy of K 6, then, t s known, a realzaton exsts n whch the copy of K 6 les on the vertces of hghest degree. So, by Theorem of AGM, we must smply construct a D for whch degree and matrx feasblty hold and for whch, n ths example, d 11 = ( 6 2) = 15.

Can we prove our conjectures usng the AGM-result? Is d = (10 6,4 8 ) potentally K 6 -graphc? If a realzaton of d exsts that contans a copy of K 6, then, t s known, a realzaton exsts n whch the copy of K 6 les on the vertces of hghest degree. So, by Theorem of AGM, we must smply construct a D for whch degree and matrx feasblty hold and for whch, n ths example, d 11 = ( 6 2) = 15. The value of d 11 forces, by degree feasblty, d 12 = d 21 = 30. In turn, by degree feasblty agan, we get d 22 = 1. It s now easy to check that matrx feasblty holds. Thus, d has a realzaton contanng a copy of K 6.

Summary Proved the EJL-conjecture Generalzed the EJL-conjecture and proved a specfc case Jont Degree-Matrx Theorem appears to be a useful tool. Can we use t to prove the generalzed EJL-conjecture?

Summary Proved the EJL-conjecture Generalzed the EJL-conjecture and proved a specfc case Jont Degree-Matrx Theorem appears to be a useful tool. Can we use t to prove the generalzed EJL-conjecture? Thank you!