ON A CONJECTURE OF RYSElt AND MINC

Similar documents
Derangements and Applications

Chapter 10. The singular integral Introducing S(n) and J(n)

LINEAR DELAY DIFFERENTIAL EQUATION WITH A POSITIVE AND A NEGATIVE TERM

The Matrix Exponential

Problem Set 6 Solutions

Probability and Stochastic Processes: A Friendly Introduction for Electrical and Computer Engineers Roy D. Yates and David J.

The Matrix Exponential

On the irreducibility of some polynomials in two variables

SECTION where P (cos θ, sin θ) and Q(cos θ, sin θ) are polynomials in cos θ and sin θ, provided Q is never equal to zero.

Cramér-Rao Inequality: Let f(x; θ) be a probability density function with continuous parameter

4. (5a + b) 7 & x 1 = (3x 1)log 10 4 = log (M1) [4] d = 3 [4] T 2 = 5 + = 16 or or 16.

Basic Polyhedral theory

NEW APPLICATIONS OF THE ABEL-LIOUVILLE FORMULA

BINOMIAL COEFFICIENTS INVOLVING INFINITE POWERS OF PRIMES. 1. Statement of results

Recall that by Theorems 10.3 and 10.4 together provide us the estimate o(n2 ), S(q) q 9, q=1

Hardy-Littlewood Conjecture and Exceptional real Zero. JinHua Fei. ChangLing Company of Electronic Technology Baoji Shannxi P.R.

10. Limits involving infinity

cycle that does not cross any edges (including its own), then it has at least

Homework #3. 1 x. dx. It therefore follows that a sum of the

Section 6.1. Question: 2. Let H be a subgroup of a group G. Then H operates on G by left multiplication. Describe the orbits for this operation.

DISTRIBUTION OF DIFFERENCE BETWEEN INVERSES OF CONSECUTIVE INTEGERS MODULO P

(Upside-Down o Direct Rotation) β - Numbers

SCHUR S THEOREM REU SUMMER 2005

An Application of Hardy-Littlewood Conjecture. JinHua Fei. ChangLing Company of Electronic Technology Baoji Shannxi P.R.China

BINOMIAL COEFFICIENTS INVOLVING INFINITE POWERS OF PRIMES

Equidistribution and Weyl s criterion

Square of Hamilton cycle in a random graph

Exercise 1. Sketch the graph of the following function. (x 2

INCOMPLETE KLOOSTERMAN SUMS AND MULTIPLICATIVE INVERSES IN SHORT INTERVALS. xy 1 (mod p), (x, y) I (j)

EEO 401 Digital Signal Processing Prof. Mark Fowler

u x v x dx u x v x v x u x dx d u x v x u x v x dx u x v x dx Integration by Parts Formula

COUNTING TAMELY RAMIFIED EXTENSIONS OF LOCAL FIELDS UP TO ISOMORPHISM

Lie Groups HW7. Wang Shuai. November 2015

Higher order derivatives

Introduction to Arithmetic Geometry Fall 2013 Lecture #20 11/14/2013

Construction of asymmetric orthogonal arrays of strength three via a replacement method

Solution: APPM 1360 Final (150 pts) Spring (60 pts total) The following parts are not related, justify your answers:

y = 2xe x + x 2 e x at (0, 3). solution: Since y is implicitly related to x we have to use implicit differentiation: 3 6y = 0 y = 1 2 x ln(b) ln(b)

1 Isoparametric Concept

Some remarks on Kurepa s left factorial

1 N N(θ;d 1...d l ;N) 1 q l = o(1)

Combinatorial Networks Week 1, March 11-12

ON A SECOND ORDER RATIONAL DIFFERENCE EQUATION

Note If the candidate believes that e x = 0 solves to x = 0 or gives an extra solution of x = 0, then withhold the final accuracy mark.

Search sequence databases 3 10/25/2016

ON THE DISTRIBUTION OF THE ELLIPTIC SUBSET SUM GENERATOR OF PSEUDORANDOM NUMBERS

Where k is either given or determined from the data and c is an arbitrary constant.

u 3 = u 3 (x 1, x 2, x 3 )

Quasi-Classical States of the Simple Harmonic Oscillator

Fourier Transforms and the Wave Equation. Key Mathematics: More Fourier transform theory, especially as applied to solving the wave equation.

22/ Breakdown of the Born-Oppenheimer approximation. Selection rules for rotational-vibrational transitions. P, R branches.

The Equitable Dominating Graph

Supplementary Materials

Differentiation of Exponential Functions

The graph of y = x (or y = ) consists of two branches, As x 0, y + ; as x 0, y +. x = 0 is the

Bifurcation Theory. , a stationary point, depends on the value of α. At certain values

That is, we start with a general matrix: And end with a simpler matrix:

COMPUTER GENERATED HOLOGRAMS Optical Sciences 627 W.J. Dallas (Monday, April 04, 2005, 8:35 AM) PART I: CHAPTER TWO COMB MATH.

Abstract Interpretation: concrete and abstract semantics

Einstein Equations for Tetrad Fields

Spectral Synthesis in the Heisenberg Group

On the number of pairs of positive integers x,y H such that x 2 +y 2 +1, x 2 +y 2 +2 are square-free

First derivative analysis

Ewald s Method Revisited: Rapidly Convergent Series Representations of Certain Green s Functions. Vassilis G. Papanicolaou 1

Limiting value of higher Mahler measure

MATH 319, WEEK 15: The Fundamental Matrix, Non-Homogeneous Systems of Differential Equations

Section 11.6: Directional Derivatives and the Gradient Vector

COHORT MBA. Exponential function. MATH review (part2) by Lucian Mitroiu. The LOG and EXP functions. Properties: e e. lim.

10. The Discrete-Time Fourier Transform (DTFT)

State-space behaviours 2 using eigenvalues

Elements of Statistical Thermodynamics

Ch. 24 Molecular Reaction Dynamics 1. Collision Theory

2.3 Matrix Formulation

Finding low cost TSP and 2-matching solutions using certain half integer subtour vertices

Integration by Parts

ECE602 Exam 1 April 5, You must show ALL of your work for full credit.

The pn junction: 2 Current vs Voltage (IV) characteristics

A Propagating Wave Packet Group Velocity Dispersion

Propositional Logic. Combinatorial Problem Solving (CPS) Albert Oliveras Enric Rodríguez-Carbonell. May 17, 2018

Another view for a posteriori error estimates for variational inequalities of the second kind

Deift/Zhou Steepest descent, Part I

An Extensive Study of Approximating the Periodic. Solutions of the Prey Predator System

BSc Engineering Sciences A. Y. 2017/18 Written exam of the course Mathematical Analysis 2 August 30, x n, ) n 2

PROOF OF FIRST STANDARD FORM OF NONELEMENTARY FUNCTIONS

The van der Waals interaction 1 D. E. Soper 2 University of Oregon 20 April 2012

Difference -Analytical Method of The One-Dimensional Convection-Diffusion Equation

Bernadette Faye, Florian Luca, and Amadou Tall

Mutually Independent Hamiltonian Cycles of Pancake Networks

5. B To determine all the holes and asymptotes of the equation: y = bdc dced f gbd

SOME PARAMETERS ON EQUITABLE COLORING OF PRISM AND CIRCULANT GRAPH.

Inference Methods for Stochastic Volatility Models

Superposition. Thinning

2008 AP Calculus BC Multiple Choice Exam

LINEAR SYSTEMS OF DIFFERENTIAL EQUATIONS DONE RIGHT KYLE ORMSBY

EECE 301 Signals & Systems Prof. Mark Fowler

Laboratory work # 8 (14) EXPERIMENTAL ESTIMATION OF CRITICAL STRESSES IN STRINGER UNDER COMPRESSION

Function Spaces. a x 3. (Letting x = 1 =)) a(0) + b + c (1) = 0. Row reducing the matrix. b 1. e 4 3. e 9. >: (x = 1 =)) a(0) + b + c (1) = 0

ON RIGHT(LEFT) DUO PO-SEMIGROUPS. S. K. Lee and K. Y. Park

ELECTRON-MUON SCATTERING

VSMN30 FINITA ELEMENTMETODEN - DUGGA

Transcription:

MA THEMATICS ON A CONJECTURE OF RYSElt AND MINC BY ALBERT NIJE~HUIS*) AND HERBERT S. WILF *) (Communicatd at th mting of January 31, 1970) 1. Introduction Lt A b an n x n matrix of zros and ons, and suppos that A has r, ons in its i t h row (i= 1,2,..., n). It has bn conjcturd [2] that for th prmannt of A w hav th inquality (1).. Pr A.;; II (r,!)l/ri. Th purpos of this papr is to prov Thorm 1: Thr is a univrsal constant r=.136708... such that.. (2) PrA.;; II {(rt!) l /ri+r}. (3) Prvious work on inqualitis of th form n Pr A.;; II q;(r,) has bn don by MIN [2] who showd that q;(x) = Hx+ 1) is admissibl in (3), MIN [3] who found q;(x) = (1 +V2)-1(x -;- V2) for (3), JURKAT and RYSER [4], WILF [5], and othrs, but no prvious q;(x) has givn vn th corrct first trm of th asymptotic bhaviour of (4) x! llx ~ ~ + log x + log ~ +o(1) 2 (x --+ x:» whras our thorm 1 abov corrctly givs th first two trms. 2. Th function q;(n) Lt q; b a fixd function of th positiv intgrs, and suppos (3) holds for all matrics A of zros and ons of ordr <:n-l. Now lt A b an n x n matrix of this typ, with row sums Tl,..., Tn. If any T, = 1, thn (3) holds for A providd only that q;(i) = 1. Othrwis, suppos th rows and columns of A hav bn prmutd, if ncssary, so that th *) Rsarch supportd in part by th National Scinc Foundation.

152 ons in th first column occur in th first c rows. Expanding by minors down th first column w find Pr A = L Pr (AH) c C n < L {II lfj(rk- 1) II lfj(rk)} i=1 k=i k~c+1 k*i 1 c n = L II lfj(rk - 1) II lfj(rk)' lfj(ri- 1) k-i k-c+l Comparing this xprssion with th right sid of (3) w s that in ordr to prov (3) for A it is nough to xhibit a function lfj such that (5) c 1 c lfj(rk - 1) 1 L )II < lfj(ri- 1 k-i lfj(rk) for all positiv intgrs c, rl,..., Tc- Considr th function lfj which is rcursivly dfind by (6) l(a) (b) lfj(i)=1 lfj(n+l)=lfj(n)1f'p(n). For this lfj, th lft sid of (5) is c L 1 c II -1/'P(rk-1) i= I lfj(ri - 1) k-i = { ± I} xp {_ ~ ± I} i= I lfj(ri- 1) k-i lfj(rk -1) =x-x / <max x- x / ",;;;'0 =1. It follows that th function lfj of (6) is admissibl for th inquality (3). Th rmaindr of this papr is dvotd to a clos study of th rcurranc (6) with a viw to stablishing th rlations n log n A (7) lfj(n) = + ~ + +0(1) (n ~ (0) and (8) A-Iog~ lfj(n) < n!i/n + (all n;» 1). 3. Asymptotic bhaviour ot lfj. W rmark first that putting bn=(lfj(n))-l in (6) (b) yilds bn +1 = b n-bn = F(bn)

153 whr P(x) = x:», Th asymptotic rlation (7) thn follows from wllknown thorms about th succssiv itrats of functions F which hav th form F(x)-=x-ax 2 +... (a> 0) nar th origin (s,.g., [6]). Nonthlss, w prov (7) indpndntly of thos rsults. First, from (6b), 1 cp(n+ 1);;;,cp(n)i and so (a) Nxt, from (6b), n cp(n»-. cp(n t-l) <; cp(n )+ t 2cp(n) which, with (10) (9), yilds n cp(n) <; - + O(log n) (n-)-oo). Now writ li(x)=x {1/x-l- ~ 1 } x 2x (11) j 2, = O(x- 2 ) (;l: -)- 0). Thn, with Yn=cp(n), w hav Summing, 1 Yn+1=Yn+ - -2 +H(Yn). Yn (lla) Yk ) n-] ] ] Ylc 1 n-l 1 n-l (k- ~n+-1+i I -k + t L -k- + I H(YI) =n+-1+!logn+'i..+! l (k- YIc) + ~H(YIc)+O(l) 2 1 kyl I =n+i log n+a +0(1) which provs (7), with r 00 (k - YI) 00 (llb) A=-1+2+tt kyl +fh(yi).

154 4. Proof of (8) W claim that th diffrncs incras monotonically. In viw of (7) and (4), this will stablish (8). W hav first, (12) n+l {I r'(x+1) 1,.' } =! I'( x + 1)1/% X F(x + 1) - x2log 1 (x 1 I) dx. From [1], p. 18, q. (27), (13) F'(x+ 1) 1 l'(x+i) <: log(x+i) - 2(x +I) <: log x+ 2x + 2x(x+ 1)' From [I], p. 22, q. (9), (14) ( log l'(x -) ;;;' (x+~) log(x -; I)-x- I+O ) (0 = log V2n) I :> x log X-X -i- -log x +C--. 2 4x 2 Th quantity in bracs in (12) is thrfor { } <:.!. _ log,x +.~ - 0 + 2. X 2x 2 x 2 x 3 (x ;;;. 2). Nxt, from [1], p. 21, q. (8), log P(x -t-i) <: (X 'l ~) log(x -l-i)-x-i+o +~ whr I I 11 K = max T t - 1 - t + '2 1>0 and aftr som calculation, on finds that (15) lf( '1)11 X log x 0 (log X)2 0 log X A 2 XT x <: - + -- + - + + + - 2 8x 2x x (x>xo)

155 whr Xo, A 2 ar any pair such that A A I. 0 (log X)2 Al log X Al (log x)2 "';;''''0 { 8x 2x 4x 2;> max -, + + 2 and A I = 2K + 1. Substituting (15), (14) in (12), (16) \ I n+1 {I 1 (log X)2 (n+l)!l/n+l n!l/n,,;; f +- " 2x - 8x2 (log x)3 [ 0 AI]} +-+- 8x x x2 (O-k)logx 2x2 A3}d - +- x x2 whr Xl, A 3 ar any pair such that A 3;> max {(A2 + ~) + C(~ - 0)1 (log X)2 + A 2 + 0 log X} "';;''''1 x 8x 2 x 2 2x 2 Thn (16) givs (n t-l)!l/n+l-n!l/n.;;; -1 -I- -1 - (log n)2 - (0-- -!) (17) 2n 8n j 2 2 logn A 4 --+n 2 n 2 whr n2, A 4 ar any pair such that n ;»n2 implis [log (n ;-1)]2 0 - t log (n +- 1), A 3 ---,- 8(n+ 1)2 2 (nl 1)2 n2 <; _ (0 - t) log n,a4 (log n )2 2 ----:n;2 T 1i2-8n2. Now from (6b), and sinc w find that qj(n-t-l);>qj(n)+ t 2 2qJ(n) n log n qj(n)<; - + -2- +A5 (18) log n A 5 qj(n+ 1)-qJ(n);> - + - +---- 2n 4n 2 2n 2

Subtracting (17) from (18), 156 [Ip(n+ 1)-(n f- l )!l /n+l ] _ [Ip(n ) - n!l /n] (log n)2 2C -1 log n As -I- 2A4 ;;> 8n2 -~ ~- n 2 for n ;;>n*. By simpl stimations, on can tak, succssivly, K = 1/12, (xo,a2)c=(1,1.5), (Xl, A s)=(i, 1.9), (n2,a 4)-=(1,2), As=l, and finally n* = l 2< 5000, which provs th monotonicity of th squnc for n ;;> 5000. By actual calculation l) on obsrvs th monotonicity for n «; 5000 also. complting th proof. 5. Conclusion Sinc {Ip(n)-n!lln} t w hav from (7) and (4) th dsird inquality (8). By computation w found, for xampl q;(09,000) = 36,422.65517926.... (!H},OOO)!l /99,OOO = 36,422.5186529... W rmark that th constant A can b xhibitd as th solution of a crtain functional quation, as is charactristic of problms of functional itration. Suppos w lt Ip(n,;) dnot th solution of th rcurrnc (6) (b) with th starting valu Ip(I, ;) =;, in plac of (6) (a). On has again th formula (7), n log n A(;) rp(n,;) ~ +2'--- +0(1) (n -)- CXJ). Now, sinc thr follows (19) A (;1M ) = 1+A(;). Thus, our constant A is A(I), whr A(;) is a solution of (19). W rmark finally that th mthod will not prov th conjctur in th sns that th function Ip(X) =X!l/x dos not satisfy th rlation (5), for xampl whn rl=... =rm=2m; r m+l= =rm+n=2n. Th constant A = 1.2905502 was computd from (llb) in which th sris shown ar rapidly convrgnt. Univrsity of Pnnsylvania 1) W gratfully acknowldg th assistanc of th Univrsity of Pnnsylvania Computr Cntr in providing tho tim for numrous calculations rlatd to this papr.

157 REFERENCES 1. BATEMAN, H. t al, Highr Transcndntal Functions, vol. 1, McGraw Hill, 1953. 2. MINc, H., Uppr Bounds for Prmannts of (O,I)-matrics, Bull. Amr. Math. Soc. 69, 789-791 (1963). 3., An inquality for prmannts of (0,l j-matrics, J. Combirratorial Thory, 2, 312-6 (1967). 4. JURKAT, W. B. and H. J. RYSER, Matrix Factorizations of dtrminants and prmannts, J. Algbra 3, 1-27 (1966). 5. WILF, H. S., A mchanical counting mthod and combinatorial applications, J. Combinatorial Thory 4, 246-258 (1968). 6. BRUIJN, N. G. DE, Asymptotic mthods in analysis, North-Holland Intrscinc, 1958.