Counterexamples to the Connectivity Conjecture of the Mixed Cells

Similar documents
Linear, affine, and convex sets and hulls In the sequel, unless otherwise specified, X will denote a real vector space.

arxiv: v1 [math.co] 1 Mar 2014

Convexity preserving interpolation by splines of arbitrary degree

Min Cut, Fast Cut, Polynomial Identities

APPENDIX A Some Linear Algebra

The L(2, 1)-Labeling on -Product of Graphs

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness.

SUPER PRINCIPAL FIBER BUNDLE WITH SUPER ACTION

Calculation of time complexity (3%)

DIFFERENTIAL FORMS BRIAN OSSERMAN

A MODIFIED METHOD FOR SOLVING SYSTEM OF NONLINEAR EQUATIONS

Genericity of Critical Types

More metrics on cartesian products

Lecture 20: Lift and Project, SDP Duality. Today we will study the Lift and Project method. Then we will prove the SDP duality theorem.

Affine transformations and convexity

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

An efficient algorithm for multivariate Maclaurin Newton transformation

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

A new Approach for Solving Linear Ordinary Differential Equations

The Pseudoblocks of Endomorphism Algebras

Bézier curves. Michael S. Floater. September 10, These notes provide an introduction to Bézier curves. i=0

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

Errata to Invariant Theory with Applications January 28, 2017

General viscosity iterative method for a sequence of quasi-nonexpansive mappings

COMPLEX NUMBERS AND QUADRATIC EQUATIONS

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0

The Order Relation and Trace Inequalities for. Hermitian Operators

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

where a is any ideal of R. Lemma 5.4. Let R be a ring. Then X = Spec R is a topological space Moreover the open sets

FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP

REAL ANALYSIS I HOMEWORK 1

MAT 578 Functional Analysis

Approximations of Set-Valued Functions Based on the Metric Average

On quasiperfect numbers

The equation of motion of a dynamical system is given by a set of differential equations. That is (1)

Changing Topology and Communication Delays

Foundations of Arithmetic

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg

Every planar graph is 4-colourable a proof without computer

2.3 Nilpotent endomorphisms

THE CHVÁTAL-ERDŐS CONDITION AND 2-FACTORS WITH A SPECIFIED NUMBER OF COMPONENTS

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family

Modulo Magic Labeling in Digraphs

Beyond Zudilin s Conjectured q-analog of Schmidt s problem

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

Zeros and Zero Dynamics for Linear, Time-delay System

a b a In case b 0, a being divisible by b is the same as to say that

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 )

Design and Optimization of Fuzzy Controller for Inverse Pendulum System Using Genetic Algorithm

The Myerson value in terms of the link agent form: a technical note

The lower and upper bounds on Perron root of nonnegative irreducible matrices

Problem Set 9 Solutions

Some basic inequalities. Definition. Let V be a vector space over the complex numbers. An inner product is given by a function, V V C

Lecture 10: May 6, 2013

Valuated Binary Tree: A New Approach in Study of Integers

MATH 5707 HOMEWORK 4 SOLUTIONS 2. 2 i 2p i E(X i ) + E(Xi 2 ) ä i=1. i=1

Discrete Mathematics

DEGREE REDUCTION OF BÉZIER CURVES USING CONSTRAINED CHEBYSHEV POLYNOMIALS OF THE SECOND KIND

SOME RESULTS ON TRANSFORMATIONS GROUPS OF N-LINEAR CONNECTIONS IN THE 2-TANGENT BUNDLE

Perron Vectors of an Irreducible Nonnegative Interval Matrix

Discrete Mathematics. Laplacian spectral characterization of some graphs obtained by product operation

On the Multicriteria Integer Network Flow Problem

Research Article Relative Smooth Topological Spaces

Lecture 12: Discrete Laplacian

Affine and Riemannian Connections

Stability of Switched Linear Systems on Cones: A Generating Function Approach

where a is any ideal of R. Lemma Let R be a ring. Then X = Spec R is a topological space. Moreover the open sets

On C 0 multi-contractions having a regular dilation

The Analytical Solution of a System of Nonlinear Differential Equations

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso

THE WEIGHTED WEAK TYPE INEQUALITY FOR THE STRONG MAXIMAL FUNCTION

Canonical transformations

REDUCTION MODULO p. We will prove the reduction modulo p theorem in the general form as given by exercise 4.12, p. 143, of [1].

Lecture Notes on Linear Regression

42. Mon, Dec. 8 Last time, we were discussing CW complexes, and we considered two di erent CW structures on S n. We continue with more examples.

Module 9. Lecture 6. Duality in Assignment Problems

CHAPTER 17 Amortized Analysis

Perfect Competition and the Nash Bargaining Solution

Exercise Solutions to Real Analysis

Iterative General Dynamic Model for Serial-Link Manipulators

Chapter Twelve. Integration. We now turn our attention to the idea of an integral in dimensions higher than one. Consider a real-valued function f : D

arxiv: v1 [quant-ph] 6 Sep 2007

Time-Varying Systems and Computations Lecture 6

MMA and GCMMA two methods for nonlinear optimization

TANGENT DIRAC STRUCTURES OF HIGHER ORDER. P. M. Kouotchop Wamba, A. Ntyam, and J. Wouafo Kamga

Introductory Cardinality Theory Alan Kaylor Cline

Deriving the X-Z Identity from Auxiliary Space Method

On intransitive graph-restrictive permutation groups

Modelli Clamfim Equazione del Calore Lezione ottobre 2014

Lecture 7: Gluing prevarieties; products

BOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS

Section 3.6 Complex Zeros

STEINHAUS PROPERTY IN BANACH LATTICES

THE FUNDAMENTAL THEOREM OF CALCULUS FOR MULTIDIMENSIONAL BANACH SPACE-VALUED HENSTOCK VECTOR INTEGRALS

Learning Theory: Lecture Notes

REGULAR POSITIVE TERNARY QUADRATIC FORMS. 1. Introduction

Vertex Graceful Labeling-Some Path Related Graphs

Transcription:

Dscrete Comput Geom 2:55 52 998 Dscrete & Computatonal Geometry 998 Sprnger-Verlag New York Inc. Counterexamples to the Connectvty Conjecture of the Mxed Cells T. Y. L and X. Wang 2 Department of Mathematcs Mchgan State Unversty East Lansng MI 48824 USA 2 Department of Mathematcs Unversty of Central Arkansas Conway AR 7235 USA Abstract. In [4] a conjecture concernng the connectvty of mxed cells of subdvsons for Mnkowsk sums of polytopes was formulated. Ths conjecture was n fact orgnally proposed by Pedersen [3]. It turns out that a postve confrmaton of ths conjecture can substantally speed up the algorthm for the dynamcal lftng developed n [4]. In the mean tme when the polyhedral method s used for solvng polynomal systems by homotopy contnuaton methods [2] the postveness of ths conjecture also plays an mportant role n the effcency of the algorthm. Very unfortunately we found that ths conjecture s naccurate n general. In Secton a counterexample s presented for a general subdvson. In Secton 2 another counterexample shows that even restrcted to regular subdvsons nduced by lftngs ths conjecture stll fals to be true.. Counterexample for General Subdvsons For m n let A = A...A m be a sequence of fnte subsets of R n whose unon affnely spans R n.forq = conva the convex hull of A =...m ther Mnkowsk sum s defned by Q + +Q m ={x + +x m x Q for =...m}. By a cell of A we mean a tuple C = C...C m of subsets C A for =...m. Defne typec := dmconvc...dmconvc m N m convc := convc + +C m R n. The research of the frst author was supported n part under NSF Grant DMS-954953 and a Guggenhem fellowshp. The second author s research was supported n part by a UCA summer stpend.

56 T. Y. L and X. Wang Here N s the set of natural numbers. Cells of the same type are called the mxed cells of that type. A face of a cell C s a subcell F = F...F m where F C and some lnear functonal α R n v attans ts mnmum over C at F for =...m. Such an α s called an nner normal of F. IfF s a face of C then convf s a face of the polytope convc for =...m. Defnton. such that A fne mxed subdvson of A s a collecton S ={C...C k } of cells a dmconvc j = n for all j =...k b convc j convc l s a proper common face of convc j and convc l when t s nonempty for j l c k j= convc j = conva d for j =...k wrte C j = C j j...c m ; then each convc j s a smplex of dmenson #C j and for each j dmconvc j + + dmconvc m j = n. Remark. In fact the notaton A = A...A m s just a shorthand abstracton of conva = conva + +A m. So s a cell C = C...C m n the subdvson of A where C A for =...m; namely t means convc = convc + +C m. 2 Later when we draw fgures for A = A...A m or subdvsons of A they actually represent and 2. For a gven fne mxed subdvson S of A = A...A m let S ={B...B l } S be the set of all mxed cells n S havng the same type k...k m wth k > for all =...m. Wrte B j = B j j...b m for j =...l. Two cells D and E n S are sad to be connected f there exsts a sequence of cells {B j...b jd } n S such that a B j = D and B jd = E and b dmconvb j p convb j p+ k for all =...m and p =...d. It s conjectured n [3] and [4] that all mxed cells of the same type n S are connected. The followng example shows that ths conjecture may not hold n general. Example. Let A = { a = b = 2 c = d = 2

Counterexamples to the Connectvty Conjecture of the Mxed Cells 57 Fg. and A 2 = { e = f = g =. Q = conva Q 2 = conva 2 and Q + Q 2 are shown n Fg.. Consder the collecton of cells C = C C 2 C 3 C 4 C 5 where C ={a d e f C 2 ={b d c e C 3 ={b c e g C 4 ={b e f g C 5 ={a b d f. Clearly C s a fne mxed subdvson of A = A A 2 and cells C and C 3 are the only cells of type. They are not connected because a d b c =. 2. Counterexample for Regular Subdvsons For A = A...A m as n Secton each A s a fnte subset of R n choose realvalued functons ω : A R for =...m. The m-tuple ω = ω...ω m s called a lftng functon on A. We say that ω lfts A to ts graph  ={qω q : q A } R n+. Ths noton s extended n the obvous way:  = Â... m Q = convâ Q = Q + + Q m etc.

58 T. Y. L and X. Wang Let S ω be the set of cells C of A whch satsfy a dmconvĉ = n and b Ĉ s a facet face of dmenson n ofaˆ whose nner normals α R n+ v have postve last coordnate. In other words convĉ s a facet of the lower hull of ˆQ. It s known [] that S ω s a subdvson of A and for a generc lftng functon ω S ω s always a fne mxed subdvson. A subdvson nduced by a lftng functon s called a regular subdvson. When the polyhedral method s used for solvng polynomal systems by homotopy contnuaton methods [2] the lftng functon ω provdes a nonlnear homotopy for the algorthm and ts nduced subdvson S ω for the supports of the polynomals plays an essental role for the whole process of the algorthm. For the effcency of the algorthm t s very desrable that the mxed cells of a fxed type of ths subdvson are connected. In fact Pedersen showed [3] that when n = 2 mxed cells of a fxed type are ndeed connected for regular subdvsons. The dea of the proof was sketched n [4]. Unfortunately the followng example shows ths conjecture s ncorrect when n = 3. Example 2. Let A = a = A 2 = A 3 = b = c = a 2 = b 2 = c 2 = a 3 = b 3 = c 3 = c 4 = and ts lftng  =  2 =  3 = { a { b { c a2 b2 c2 a3 b3 c3 4 c4. The regular subdvson of A = A A 2 A 3 nduced by the above lftng s shown n Fg. 2. In ths fne mxed subdvson there are eght cells they are: C ={a b b 2 b 3 c 2 c 4 C 2 ={a b 2 b 3 c 2 c 3 c 4 C 3 ={a b b 3 c c 2 c 4 C 4 ={a a 3 b 2 b 3 c 3 c 4 C 5 ={a a 2 b b 3 c c 2 C 6 ={a a 3 b 3 c 2 c 3 c 4 C 7 ={a a 2 b 3 c c 2 c 4 C 8 ={a a 2 a 3 b 3 c 2 c 4.

Counterexamples to the Connectvty Conjecture of the Mxed Cells 59 Fg. 2 see Fg. 3. Clearly there are two cells of type they are C 4 and C 5.Onthe other hand α = 2 and α 2 = are the nner normals of Ĉ 4 and Ĉ 5 respectvely. Obvously C 4 and C 5 are not connected snce c c 2 c 3 c 4 =. Remark. The dynamc lftng algorthm n [4] constructs placeable subdvsons whch only requre mxed cells of a certan type to be fne. We trust that counterexamples to the connectvty of mxed cells n placeable regular subdvsons can also be constructed. 2

52 T. Y. L and X. Wang Fg. 3

Counterexamples to the Connectvty Conjecture of the Mxed Cells 52 References. I. M. Gel fand M. Kapranov and A. Zelevnsky Multdmensonal Determnants Dscrmnants and Resultants Brkhäuser Boston 994. 2. B. Huber and B. Sturmfels A polyhedral method for solvng sparse polynomal systems. Math. Comp. 6422:54 555 995. 3. P. Pedersen Personal communcaton at the occason of the AMS IMS SIAM Summer Research Conference on Contnuous Algorthm and Complexty Mount Holyoke College South Hadley MA June 994. 4. J. Verschelds K. Gatemann and R. Cools Mxed volume computaton by dynamcal lftng appled to polynomal system solvng Dscrete Comput. Geom. 6:69 2 996. Receved September 8 996 and n revsed form February 7 997.