VOLUMES OF CONVEX POLYTOPES

Similar documents
arxiv: v1 [math.co] 1 Apr 2011

On the ratio of maximum and minimum degree in maximal intersecting families

arxiv: v1 [math.co] 6 Mar 2008

Math 301: The Erdős-Stone-Simonovitz Theorem and Extremal Numbers for Bipartite Graphs

(n 1)n(n + 1)(n + 2) + 1 = (n 1)(n + 2)n(n + 1) + 1 = ( (n 2 + n 1) 1 )( (n 2 + n 1) + 1 ) + 1 = (n 2 + n 1) 2.

Quasi-Randomness and the Distribution of Copies of a Fixed Graph

Solution to HW 3, Ma 1a Fall 2016

A Bijective Approach to the Permutational Power of a Priority Queue

ON THE INVERSE SIGNED TOTAL DOMINATION NUMBER IN GRAPHS. D.A. Mojdeh and B. Samadi

arxiv: v1 [math.co] 4 May 2017

Journal of Inequalities in Pure and Applied Mathematics

On the ratio of maximum and minimum degree in maximal intersecting families

Auchmuty High School Mathematics Department Advanced Higher Notes Teacher Version

On the Number of Rim Hook Tableaux. Sergey Fomin* and. Nathan Lulov. Department of Mathematics. Harvard University

Enumerating permutation polynomials

On decompositions of complete multipartite graphs into the union of two even cycles

THE JEU DE TAQUIN ON THE SHIFTED RIM HOOK TABLEAUX. Jaejin Lee

Semicanonical basis generators of the cluster algebra of type A (1)

On the Quasi-inverse of a Non-square Matrix: An Infinite Solution

Math Notes on Kepler s first law 1. r(t) kp(t)

working pages for Paul Richards class notes; do not copy or circulate without permission from PGR 2004/11/3 10:50

The Congestion of n-cube Layout on a Rectangular Grid S.L. Bezrukov J.D. Chavez y L.H. Harper z M. Rottger U.-P. Schroeder Abstract We consider the pr

A NOTE ON ROTATIONS AND INTERVAL EXCHANGE TRANSFORMATIONS ON 3-INTERVALS KARMA DAJANI

Berkeley Math Circle AIME Preparation March 5, 2013

3.1 Random variables

Euclidean Figures and Solids without Incircles or Inspheres

COLLAPSING WALLS THEOREM

A generalization of the Bernstein polynomials

Lecture 18: Graph Isomorphisms

OLYMON. Produced by the Canadian Mathematical Society and the Department of Mathematics of the University of Toronto. Issue 9:2.

Pascal s Triangle (mod 8)

Chromatic number and spectral radius

Relating Branching Program Size and. Formula Size over the Full Binary Basis. FB Informatik, LS II, Univ. Dortmund, Dortmund, Germany

When two numbers are written as the product of their prime factors, they are in factored form.

Fractional Zero Forcing via Three-color Forcing Games

Miskolc Mathematical Notes HU e-issn Tribonacci numbers with indices in arithmetic progression and their sums. Nurettin Irmak and Murat Alp

18.06 Problem Set 4 Solution

New problems in universal algebraic geometry illustrated by boolean equations

SUFFICIENT CONDITIONS FOR MAXIMALLY EDGE-CONNECTED AND SUPER-EDGE-CONNECTED GRAPHS DEPENDING ON THE CLIQUE NUMBER

arxiv: v1 [math.nt] 12 May 2017

24. Balkanska matematiqka olimpijada

Chapter 5 Linear Equations: Basic Theory and Practice

On a generalization of Eulerian numbers

A Short Combinatorial Proof of Derangement Identity arxiv: v1 [math.co] 13 Nov Introduction

A STUDY OF HAMMING CODES AS ERROR CORRECTING CODES

THE LAPLACE EQUATION. The Laplace (or potential) equation is the equation. u = 0. = 2 x 2. x y 2 in R 2

J. N. R E DDY ENERGY PRINCIPLES AND VARIATIONAL METHODS APPLIED MECHANICS

PROBLEM SET #1 SOLUTIONS by Robert A. DiStasio Jr.

No. 39. R.E. Woodrow. This issue we give another example of a team competition with the problems

The Erdős-Hajnal conjecture for rainbow triangles

Matrix Colorings of P 4 -sparse Graphs

An intersection theorem for four sets

Problem Set #10 Math 471 Real Analysis Assignment: Chapter 8 #2, 3, 6, 8

Probablistically Checkable Proofs

ON INDEPENDENT SETS IN PURELY ATOMIC PROBABILITY SPACES WITH GEOMETRIC DISTRIBUTION. 1. Introduction. 1 r r. r k for every set E A, E \ {0},

15.081J/6.251J Introduction to Mathematical Programming. Lecture 6: The Simplex Method II

Lecture 7: Angular Momentum, Hydrogen Atom

6 PROBABILITY GENERATING FUNCTIONS

MODULE 5a and 5b (Stewart, Sections 12.2, 12.3) INTRO: In MATH 1114 vectors were written either as rows (a1, a2,..., an) or as columns a 1 a. ...

Heronian Triangles of Class K: Congruent Incircles Cevian Perspective

Measure Estimates of Nodal Sets of Polyharmonic Functions

The height of minimal Hilbert bases

9.1 The multiplicative group of a finite field. Theorem 9.1. The multiplicative group F of a finite field is cyclic.

Available online through ISSN

Modified Linear Programming and Class 0 Bounds for Graph Pebbling

gr0 GRAPHS Hanan Samet

SOME GENERAL NUMERICAL RADIUS INEQUALITIES FOR THE OFF-DIAGONAL PARTS OF 2 2 OPERATOR MATRICES

Brief summary of functional analysis APPM 5440 Fall 2014 Applied Analysis

MATH 220: SECOND ORDER CONSTANT COEFFICIENT PDE. We consider second order constant coefficient scalar linear PDEs on R n. These have the form

Suborbital graphs for the group Γ 2

The Chromatic Villainy of Complete Multipartite Graphs

Using Laplace Transform to Evaluate Improper Integrals Chii-Huei Yu

EQUI-PARTITIONING OF HIGHER-DIMENSIONAL HYPER-RECTANGULAR GRID GRAPHS

PDF Created with deskpdf PDF Writer - Trial ::

556: MATHEMATICAL STATISTICS I

The Catalan numbe sequence Abstact ; ; ; 5; 4; 4;:::; n + n ;::: n is one of the most impotant sequences in all of enumeative combinatoics. Richad Sta

CALCULUS II Vectors. Paul Dawkins

A Crash Course in (2 2) Matrices

On a quantity that is analogous to potential and a theorem that relates to it

Classical Worm algorithms (WA)

Fall 2014 Randomized Algorithms Oct 8, Lecture 3

Math 124B February 02, 2012

Math 259 Winter Handout 6: In-class Review for the Cumulative Final Exam

Bounds on the performance of back-to-front airplane boarding policies

On the Structure of Linear Programs with Overlapping Cardinality Constraints

Lecture 1.1: An introduction to groups

Chapter 3: Theory of Modular Arithmetic 38

An upper bound on the number of high-dimensional permutations

RELAXED COMPLETE PARTITIONS: AN ERROR-CORRECTING BACHET S PROBLEM

THE CONE THEOREM JOEL A. TROPP. Abstract. We prove a fixed point theorem for functions which are positive with respect to a cone in a Banach space.

As is natural, our Aerospace Structures will be described in a Euclidean three-dimensional space R 3.

arxiv: v2 [math.ag] 4 Jul 2012

Markscheme May 2017 Calculus Higher level Paper 3

An Application of Fuzzy Linear System of Equations in Economic Sciences

Vanishing lines in generalized Adams spectral sequences are generic

Failure Probability of 2-within-Consecutive-(2, 2)-out-of-(n, m): F System for Special Values of m

The Tutte polynomial of a graph, depth-rst. search, and simplicial complex partitions

A proof of the binomial theorem

MULTILAYER PERCEPTRONS

CERFACS 42 av. Gaspard Coriolis, Toulouse, Cedex 1, France. Available at Date: April 2, 2008.

Transcription:

VOLUMES OF CONVEX POLYTOPES Richad P. Stanley Depatment of Mathematics M.I.T. 2-375 Cambidge, MA 02139 stan@math.mit.edu http://www-math.mit.edu/~stan Tanspaencies available at: http://www-math.mit.edu/~stan/tans.html 1

P = convex polytope in R n intege polytope: vetices 2 Z n V(P) = volume of P If P is an intege polytope, let ev(p) = n! V (P) 2 Z; the nomalized volume of P. 2

Why compute V (P)? Let P = intege polytope = the nomal fan of P X = toic vaiety coesponding to ) deg(x ) = e V (P) (Gelfand, Kapanov, Zelevinsky) The numbe of linealy independent solutions to a geneic hypegeometic system with Newton polytope P is ev (P). 3

Two Renements of Volume Let P be an intege polytope and let 1. Dene P = fv : v 2 Pg i(p; ) = # (P \ Z n ) ; the Ehhat polynomial of P. i(p; ) is a polynomial in i(p; 0) = 1 If > 0, then i(p; ) = ( 1) dim P #(int(p)\z n ) i(p; ) = V (P) n + O( n 1 ). 4

Let dim P = n and X i(p; )x = h 0 + h 1 x + + h n x n (1 x) n+1 : 0 Then h j 2 Z, h j 0, and X h j = V e (P): j 5

Example. P = unit squae: 0 x 1; 0 y 1 0 4 ev (P) = 2! 1 = 2 i(p; ) = ( + 1) 2 X 0 i(p; ) = ( 1) 2 i(p; )x = 1 + x (1 x) 3 6

Let P 1 ; : : : ; P k be convex polytopes (o any convex bodies) in R n. Let t 1 ; : : : ; t k 2 R 0 : Dene the Minkowski sum t 1 P 1 + +t k P k = ft 1 v 1 + +t k v k : v i 2 P i g: Theoem (Minkowski) Thee exist V (P a 1 ) 0 (mixed volumes) k 1 ; : : : ; Pa k such that V (t 1 P 1 + + t k P k ) = X a 1 ++a k =n n a 1 ; : : : ; a k V (P a 1 1 ; : : : ; Pa k k )ta 1 1 ta k k : 7

Wite V (sp+tq) = so nx j=0 n j V j (P; Q)s n j t j ; V 0 (P; Q) = V (P); V n (P; Q) = V (Q): Minkowski poved V n 1 V n 1 V 0 n and conjectued V i 2 V i 1 V i+1 (Alexandov- Fenchel inequalities). Let B n be the unit ball in R n. Then V 1 (P; B n ) = 1 (suface aea of P): n 8

V 1 (P; B ) n V 0 (P; B ) n 1 V n (P; B ) ) aea(p) n V (P) n 1 V (B n ) n! isopeimetic inequality: 9

Theoem (Benstein). Let f 1 ; : : : ; f n be complex polynomials in the vaiables x 1 ; : : : ; x n. Let New(f j ) be the Newton polytope (convex hull of exponent vectos) of f j. If f 1 ; : : : ; f n ae othewise geneic, then the numbe Z(f 1 ; : : : ; f n ) of solutions to with no x i = 0 is f 1 = = f n = 0 Z(f 1 ; : : : ; f n ) = n! V (New(f 1 ); : : : ; New(f n )): Example. f 1 (x; y) = 1 + x + y 2, f 2 (x; y) = x + y 4. 10

2s+4t 4 4t 2 2s 1 t s+t 2 V (s New(f 1 )+t New(f 2 )) = s 2 + 2st 1 ) V (New(f 1 ); New(f 2 )) = 2! 2 = 4 11

A common genealization of Ehhat polynomials and mixed volumes (McMullen). Fo any convex body P, let N(P) = # (P \ Z n ) : Theoem. Let P 1 ; : : : ; P k be intege polytopes, and let Then t 1 ; : : : ; t k 2 N = f0; 1; : : :g: N(t 1 P 1 + + t k P k ) 2 Q [t 1 ; : : : ; t k ] (mixed lattice point enumeato of P 1 ; : : : ; P k ). 12

The degee n pat of N(t 1 P 1 + + t k P k ) is given by N(t 1 P 1 + + t k P k )j n = V (t 1 P 1 + +t k P k ): 13

Catalan Numbes C n = 1 2n n + 1 n tiangulations of a convex (n+2)-gon into n tiangles by n 1 diagonals that do not intesect in thei inteios binay tees with n vetices @ @@ @ @@ @ @ 14

lattice paths fom (0; 0) to (n; n) with steps (0; 1) o (1; 0), neve ising above the line y = x sequences of n 1's and n 1's such that evey patial sum is nonnegative (with 1 denoted simply as below) 111 11 1 11 1 1 11 1 1 1 15

Fo 62 additional combinatoial intepetations of C n, see Execise 6.19 of R. Stanley, Enumeative Combinatoics, volume 2, Cambidge Univesity Pess (just published). 16

Flow Polytopes and Kostant's Patition Function (with A. Postnikov) Let B m denote the Bikho polytope of all m m doubly stochastic matices a ij, i.e., X Xi j a ij 0 a ij = 1 a ij = 1: Open: V (B m ) =??? (as a polytope of dimension (m 1) 2 ). 17

Chan and Robbins (1998) dened the Chan-Robbins polytope CR m by CR m = f a ij 2 Bm : a ij = 0 if j > i+1g (a face of B m ). 0 0 0 dim CR m = m 2 Chan and Robbins conjectued that ev (CR m ) = C 1 C 2 C m 2 : 18

Flow polytopes. Let G a diected gaph with vetices 1; : : : ; m + 1 and edge set E such that if i! j is an edge, then i < j. Call G a ow gaph. Dene the ow polytope F G to be the set of all f 2 R E X 0 satisfying (1;j)2E X (j;m+1)2e f(1; j) = 1 f(j; m + 1) = 1 X f(i; j) = X f(j; k); i : (i;j)2e k : (j;k)2e fo 2 j m. 19

.1.6.2.3.1.7.7 1 2 3 4 5 total ow out of 1 and into m + 1 is 1 ow into an intenal vetex = ow out 20

Fact: If E = f(i; j) : 1 i < j m + 1g; then F G is \unimodulaly equivalent" to CR m (so same volume and Ehhat polynomial). Call G the complete ow gaph on m + 1 vetices. 21

Kostant's patition function fo A n 1. Let e i = ith unit coodinate vecto in R n : Wite e ij = e i e j. Let A + n 1 = fe ij : 1 i < j ng: 22

v 2 N A + n 1 = 8 < : X 1i<jn a ij e ij : a ij 2 N 9 = ; : # K(v) = n aij 1i<jn : v = X a ij e ij o : If v = (v 1 ; : : : ; v n ) and x v = x v 1 1 xv n n, then K(v) = coef. of x v in Q 1i<jn 1 : 1 xi =x j 23

Example. K(2; 1; 0; 1) = 4, since (2; 1; 0; 1) = 2e 12 + e 23 + e 34 = 2e 12 + e 24 = e 12 + e 13 + e 34 = e 12 + e 14 : Gelfand: Evey subject has one \tanscendental aspect." Fo the epesentation theoy of semisimple Lie algebas, it is Kostant's patition function. 24

Moe geneally, if S A +, then den 1 ne the esticted Kostant's patition function by K S (v) = # n aij e ij 2S : v = X e ij 2S a ij e ij 9 > = >; : 25

Let G be a ow gaph on the vetex set 1; : : : ; m + 1 with edge set E. Let G i = estiction of G to vetices i; : : : ; m + 1: Let t i 0 and F G (t 1 ; : : : ; t m 1 ) = t 1 F G1 + +t m 1 F Gm 1 : Note. F G (t 1 ; : : : ; t m 1 ) is the polytope of all f 2 R E 0 satisfying X k : (j;k)2e f(j; k) X i : (i;j)2e f(i; j) = t j ; fo 1 j m 1. In othe wods, thee is an excess ow of t j out of vetex j. 26

Let S = S G = fe ij : (i; j) is an edge of Gg. Theoem. If each t j 2 N then N(F G (t 1 ; : : : ; t m 1 )) = K SG (t 1 + +t m 1 ; 0; t m 1 ; : : : ; t 1 ): Note: K SG (t 1 + +t m ; t m ; : : : ; t 1 ) = K S (t 1 + +t m 1 ; 0; t m 1 ; : : : ; t 1 ) if e 12 2 S G. (Othewise K SG (t 1 + + t m ; t m ; : : : ; t 1 ) = 0 unless t m = 0.) Coollay. Fo t i 2 N we have K S (t 1 + + t m 1 ; 0; t m 1 ; : : : ; t 1 ) 2 Q [t 1 ; : : : ; t m 1 ]: 27

Example: G = complete ow gaph on f1; : : : ; m + 1g, so F G = CRm and Then F Gi = CRm i+1 : N(F G (t 1 ; : : : ; t m 1 )) = K(t 1 + +t m 1 ; 0; t m 1 ; : : : ; t 1 ): E.g., K(a+b; 0; b; a) = 1 6 (a+1)(a+2)(a+3b+3) K(a + b + c; 0; c; b; a) = 1 (a + 1)(a + 2)(a + 3)(a + b + 3c + 3) 360 (a 2 + 5ab + 10b 2 + 9a + 30b + 20): 28

Theoem. K(t 1 + + t m 1 ; 0; t m 1 ; : : : ; t 1 ) is divisible by (t 1 + 1) (t 1 + m 1). Conjectue. K(t 1 + + t m 1 ; 0; t m 1 ; : : : ; t 1 ) is divisible by t 1 + t 2 + + t m 2 + 3t m 1 + 3: Moe stongly, 3K(t 1 + +t m 1 ; 0; t m 1 ; : : : ; t 1 ) = (t 1 + + t m 2 + 3t m 1 + 3) K no e23 (t 1 + +t m 2 ; 0; 0; t m 2 ; : : : ; t 1 ): 29

Poblem (not caefully looked at). Ae the coecients of the polynomial K S (t 1 + +t m 1 ; 0; t m 1 ; : : : ; t 1 ) nonnegative? 30

Recall: Let q = #E(G), so Then F G (t 1 ; : : : ; t m 1 ) R q : K SG (t 1 + + t m 1 ; 0; t m 1 ; : : : ; t 1 ) q = V (t 1 F G1 + + t m 1 F Gm 1 ): Theoem. Let a 1 + +a m 1 = q, a i 0. Let Then i = outdeg(i) 1: ev (F a 1 ; : : : ; F a m 1 ) G 1 G m 1 = K SG (a 1 1 ; : : : ; a m 1 m 1 ): 31

Example. 1 2 3 4 5 ( 1 ; 2 ; 3 ) = (3; 1; 1) S = fe 12 ; e 13 ; e 14 ; e 23 ; e 34 g 1 s3 t 2 V (sf G1 + tf G2 + uf G3 ) = 3 tu 3! 2! +1 s 3! 1! 1! +2 s 4! 1! +1s 4! 1! +2s 5! : 4 t 4 u 5 32

K(0; 1; 1) = 1 K(0; 0; 0) = 1 K(1; 0; 1) = 2 K(1; 1; 0) = 1 K(2; 1; 1) = 2 33

Coollay. ev (F G ) = K SG (q m+1 1 ; 2 ; : : : ; m 2 ) Fo the complete ow gaph G with m + 1 vetices, m 1 K( 2 E.g., ev (F G ) = e V (CRm ) = ; (m 2); (m 3); : : : ; 1): K(6; 3; 2; 1) = C 1 C 2 C 3 = 1 2 5 = 10: 34

Coollay. Let CT denote the constant tem of a Lauent seies. Then CT ny i=1 ev (CR n+2 ) = Y (1 x i ) 2 (x j x i ) 1 : Theoem (Mois). CT = 1 n! ny i=1(1 x i ) a n Y i=1 n 1 Y j=0 1i<jn x b i Y 1i<jn (x j x i ) 2c (a + b + (n 1 + j)c) (c) (a + jc) (c + jc) (b + jc + 1) : Coollay (Zeilbege) e V (CRn ) = C 1 C n 2. 35

Note: Let 0 a b. poduct fomula fo K b + 1 2 a 2 9 simple ; b; b + 1; : : : ; a : 36

One Futhe Flow Gaph 5 6 1 2 3 4 E(G) = f(i; i + 1) : 1 i mg [ f(1; i) : 2 i mg [ f(i; m + 1) : 2 i mg: Theoem. V e (FG ) = C m 2. 37

The Catalanotope C n = conv(a + n [ f0g) R n+1 : e -e 2 3 e -e 1 3 0 e -e 1 2 38

Let T be a tee with vetex set f1; : : : ; n + 1g and edge set E. Dene the simplex T = conv fe ij : ij 2 E; i < jg [ f0g : 3 6 2 7 4 1 5 T = convfe 16 ; e 25 ; e 36 ; e 46 ; e 56 ; e 57 ; 0g 39

T is altenating if eithe evey neighbo of vetex i is less than i o is geate than i. T is noncossing if thee ae not edges ik and jl whee i < j < k < l. 1 2 3 4 5 6 7 8 40

Theoem (A. Postnikov). The simplices T, whee T anges ove all noncossing altenating tees with vetex set f1; : : : ; n+1g, ae the maximal faces of a tiangulation of C n. e -e 2 3 e -e 1 3 0 e -e 2 1 Easy: e V (T ) = 1. 41

Lemma. The numbe of noncossing altenating tees with vetex set f1; : : : ; n + 1g is the Catalan numbe C n. 1 2 3 4 5 6 7 8 Coollay. V e (Cn ) = C n. 42

Theoem. X i(c n ; )x = 0 P n 1 j=0 n 1 n j n j+1 x j (1 x) n+1 Hee 1 n n j n j+1 is a Naayana numbe. 43

Example (an ode polytope). Let O mn be the set of all points satisfying aij 1im 1jn R mn 0 a ij 1 a ij a i 1;j (1) a ij a i;j 1 : (2) Vetices ae the (0; 1)-matices in O mn. 44

Note: i(o mn ; ) is the numbe of matices a ij 1im 1jn and a ij satisfying (1), (2), 2 f0; 1; : : : ; g (plane patition with m ows, n columns, and lagest pat ). 0 0 1 2 1 3 3 3 3 3 4 5 5 7 7 9 45

O mn has a tiangulation whose facets (maximal faces) T ae indexed by standad Young tableaux T of shape (n; : : : ; n) (m n's). Example. T = 1 2 4 3 5 6 T : 0 a 11 a 12 a 21 a 13 a 22 a 23 Each T is pimitive, i.e., ev ( T ) = 1: 46

Coollay. We have ev (O mn ) = numbe of SYT of shape (n; : : : ; n) n! = Q Q m n i=1 j=1 (i + j 1) Theoem (MacMahon). We have i(o mn ; ) = my ny i=1 j=1 + i + j 1 i + j 1 : 47

Special case: m = 2. Then ev (O 2n ) = C n = 1 n + 1 the nth Catalan numbe. 2n n ; X 0 i(o 2n ; )x = P n 1 j=0 n 1 n j n j+1 x j (1 x) 2n+1 ; whee 1 n n j n j+1 is a Naayana numbe as above. 48