BOUNDS FOR DETERMINANTS WITH POSITIVE DIAGONALS

Similar documents
Chapter 5: Inner Product Spaces

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 8

In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

Irreducible polynomials with consecutive zero coefficients

Infinite Sequences and Series

SAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx

THE ABRACADABRA PROBLEM

Asymptotic Growth of Functions

A probabilistic proof of a binomial identity

ON AN INTEGRAL OPERATOR WHICH PRESERVE THE UNIVALENCE

3. Greatest Common Divisor - Least Common Multiple

On the L p -conjecture for locally compact groups

Lecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem

FIBONACCI NUMBERS: AN APPLICATION OF LINEAR ALGEBRA. 1. Powers of a matrix

Factors of sums of powers of binomial coefficients

Lecture 5: Span, linear independence, bases, and dimension

Convexity, Inequalities, and Norms

Department of Computer Science, University of Otago

Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009)


. P. 4.3 Basic feasible solutions and vertices of polyhedra. x 1. x 2

Section 11.3: The Integral Test

Theorems About Power Series

THE LEAST COMMON MULTIPLE OF A QUADRATIC SEQUENCE

Overview on S-Box Design Principles

5 Boolean Decision Trees (February 11)

Elementary Theory of Russian Roulette

Our aim is to show that under reasonable assumptions a given 2π-periodic function f can be represented as convergent series


Chapter 7 Methods of Finding Estimators

1. C. The formula for the confidence interval for a population mean is: x t, which was

4.3. The Integral and Comparison Tests

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES

A Note on Sums of Greatest (Least) Prime Factors

Overview of some probability distributions.

Degree of Approximation of Continuous Functions by (E, q) (C, δ) Means

I. Chi-squared Distributions

MATH 304 Linear Algebra Lecture 9: Subspaces of vector spaces (continued). Span. Spanning set.

Trigonometric Form of a Complex Number. The Complex Plane. axis. ( 2, 1) or 2 i FIGURE The absolute value of the complex number z a bi is

THIN SEQUENCES AND THE GRAM MATRIX PAMELA GORKIN, JOHN E. MCCARTHY, SANDRA POTT, AND BRETT D. WICK

Incremental calculation of weighted mean and variance

Soving Recurrence Relations

Bond Valuation I. What is a bond? Cash Flows of A Typical Bond. Bond Valuation. Coupon Rate and Current Yield. Cash Flows of A Typical Bond

The following example will help us understand The Sampling Distribution of the Mean. C1 C2 C3 C4 C5 50 miles 84 miles 38 miles 120 miles 48 miles

ON THE DENSE TRAJECTORY OF LASOTA EQUATION

THE HEIGHT OF q-binary SEARCH TREES

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n

Properties of MLE: consistency, asymptotic normality. Fisher information.

Week 3 Conditional probabilities, Bayes formula, WEEK 3 page 1 Expected value of a random variable

Definition. A variable X that takes on values X 1, X 2, X 3,...X k with respective frequencies f 1, f 2, f 3,...f k has mean

How To Solve An Old Japanese Geometry Problem

1. MATHEMATICAL INDUCTION

Lecture 4: Cheeger s Inequality

2. Degree Sequences. 2.1 Degree Sequences

5.3. Generalized Permutations and Combinations

Chapter 14 Nonparametric Statistics

A Recursive Formula for Moments of a Binomial Distribution

Sampling Distribution And Central Limit Theorem

MATH 304 Linear Algebra Lecture 18: Rank and nullity of a matrix.

Notes on Determinant

1 Computing the Standard Deviation of Sample Means

Output Analysis (2, Chapters 10 &11 Law)

3 Basic Definitions of Probability Theory

CS103A Handout 23 Winter 2002 February 22, 2002 Solving Recurrence Relations

Hypothesis testing. Null and alternative hypotheses

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights

7 Gaussian Elimination and LU Factorization

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Discrete Mathematics and Probability Theory Spring 2014 Anant Sahai Note 13

How To Solve The Homewor Problem Beautifully

Case Study. Normal and t Distributions. Density Plot. Normal Distributions

Ekkehart Schlicht: Economic Surplus and Derived Demand

Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling

Factoring x n 1: cyclotomic and Aurifeuillian polynomials Paul Garrett <garrett@math.umn.edu>

Multiple Representations for Pattern Exploration with the Graphing Calculator and Manipulatives

AN ALGORITHM FOR DETERMINING WHETHER A GIVEN BINARY MATROID IS GRAPHIC

Maximum Likelihood Estimators.

Example 2 Find the square root of 0. The only square root of 0 is 0 (since 0 is not positive or negative, so those choices don t exist here).

Approximating Area under a curve with rectangles. To find the area under a curve we approximate the area using rectangles and then use limits to find

Class Meeting # 16: The Fourier Transform on R n

Confidence Intervals for One Mean

A Test of Normality. 1 n S 2 3. n 1. Now introduce two new statistics. The sample skewness is defined as:

Confidence Intervals. CI for a population mean (σ is known and n > 30 or the variable is normally distributed in the.

NATIONAL SENIOR CERTIFICATE GRADE 11

Systems Design Project: Indoor Location of Wireless Devices

A Combined Continuous/Binary Genetic Algorithm for Microstrip Antenna Design

Analysis Notes (only a draft, and the first one!)

The Gompertz Makeham coupling as a Dynamic Life Table. Abraham Zaks. Technion I.I.T. Haifa ISRAEL. Abstract

T ( a i x i ) = a i T (x i ).

Chapter 7 - Sampling Distributions. 1 Introduction. What is statistics? It consist of three major areas:

PROCEEDINGS OF THE YEREVAN STATE UNIVERSITY AN ALTERNATIVE MODEL FOR BONUS-MALUS SYSTEM

Swaps: Constant maturity swaps (CMS) and constant maturity. Treasury (CMT) swaps

Lecture 3. denote the orthogonal complement of S k. Then. 1 x S k. n. 2 x T Ax = ( ) λ x. with x = 1, we have. i = λ k x 2 = λ k.

AP Calculus AB 2006 Scoring Guidelines Form B

CS103X: Discrete Structures Homework 4 Solutions

Decomposition of Gini and the generalized entropy inequality measures. Abstract

DATA ANALYSIS II. Matrix Algorithms

Journal of Combinatorial Theory, Series A

Notes on Symmetric Matrices

Transcription:

BOUNDS FOR DETERMINANTS WITH POSITIVE DIAGONALS BY EMILIE V. HAYNSWORTH(i) 1. Itroductio. I this paper upper ad lower bouds are foud for the determiat of a real, X matrix A = (a<,-)> with positive diagoal elemets satisfyig (1) a ^ X) I a.71. i = 1,2,,, ad a lower boud is foud for determiats whose elemets satisfy (2) a ^ At ^ ««, i = 1, 2,,, j **«1 / \ At = I max a + max a# 1. 2 \ j*i j*i / G. B. Price [6], A. Ostrowski [4] ad [5], J. L. Breer [l] ad [2], ad H. Scheider [7] have give lower ad upper bouds for the absolute value of determiats satisfyig more geeral coditios tha (1). However, the followig theorem, proved i 2, is ot implied by ay of the above results : Theorem 1. If A = (o ) has elemets satisfyig (l),itis possible to defie Lt ad Rt such that a = Li + Ri,,,. Li ^ X) I o.71. (i = 1,, w). The, for ay choice of Lt ad R{ satisfyig (3), (4) E(l[i. R) det A II (Li + 2Ri)Lk f[ Rh Zc=0 \»=1 >=i+l / Ji=.0 j=l i=k+l a empty product is defied to be 1. Preseted to the Society, Jauary 28, 1960; received by the editors November 2, 1959. (') The author is presetly at Aubur Uiversity, Aubur, Alabama. 395

396 E. V. HAYNSWORTH [September Ostrowski prove Theorem [5] has show that if A satisfies (2) the det A 2;0. I 3 we 2. If A satisfies (2), the det A ^ II ( E a - At). «=i V j=i ' I provig this we prove a result which may be used to improve ay boud depedig oly o the odiagoal elemets i the row, whe the diagoal elemets are positive. 2. Proof of Theorem 1. To prove Theorem 1 we eed the followig boud give by Price [6]: If (1) holds, the (5) II (««- r.) ^ det A g II (<*«+ r<) r,= E>>< o,j. We proceed by iductio o. Let > represet det A, whe ^4 is of order, ad suppose the elemets of A satisfy (3). 1. For m = 2, >2 = Li + Ri ai2 #21 l-»2 I -*^2 Expadig Z?2 by the diagoal elemets, Therefore Li «i2 Li 0 i?x 0 jri ai2 A> = + + + 0 L,2 0 i?2 021 Ri #21 '-'2 Z-iL2 + Z.1^2 + R1R2 ^ Z>2 ^ R1R2 + LA + (Li + 2Ri)U, sice 0 g g (Ui + a12)jl2 < 2RiLt, by (3) ad (5). 2. Assume that for ay matrix of order w 1 with elemets satisfyig (3), (6) e ( l, r) ^ z? _i ^ e ( a*+m^u r). k=0 \ i=l «'-*+l / i-0 \ i-1 t=*+l / If Z> = det A, A = (0,7), i, j=l, (3), partitio D as follows:,, ad the elemets ay satisfy

19601 BOUNDS FOR DETERMINANTS WITH POSITIVE DIAGONALS 397 D = Ai a2 03 L + R Ai=(aij),i,j=l,, \; a2 is the colum vector with compoets ai, i=l,,«1;o3 is the row vector with compoets aj, j = 1,, 1; ad, as i (3), J-> "T* -K- == O, -1 L ^ X) I Oi I, J?^ 0. The we ca write > as the sum of two determiats, i.e., (7) D = A + Rdet Ax A = Ai a-i But the elemets of A satisfy (1), hece, by (3) ad (5), (8) A ^ fl (a -ri)^f[ (a - Rt) = jj L{,»=i t 1 t-i -l -1 ad A ^ L II (o» + u) ^ i II {U + 22?,). i l t 1 ff3 Also, by the iductive assumptio, sice A\ is of order 1, ad, by (3), L we have, usig (6), (7) ad (8), ad Ri^L t, \ai,-\ ^ a*, 1 / fc 1 \ / ft V D * l* + R z( ii &) = z( < *.), D g L (l,+2*0+* z ( ( <+2Ri)Lk *.) = il(l[(li+2ri)lk rx k 0 \ t=l»'-*:+1 /

398 E. V. HAYNSWORTH [September 3. Proof of Theorem 2. Suppose we have a lower boud for the determiat of A which holds whe the diagoal elemets are idividually bouded from below. That is, suppose (9) a ; a(a) 0, i = 1, 2, «, implies det A ^ m(a) ^ 0, ct(a) is a fuctio of the odiagoal elemets i the ith row, with Ci{A) =0 if atj = 0 for all j^i, ad m(a) is a fuctio of all the odiagoal elemets of A. Let Z) = diag(di, dit, d) with tf\^0. We shall show that (10) detu + D) ^ w(^) + II <* ' t=i It is this result which we will use to improve Ostrowski's lower boud for determiats satisfyig (2). Sice du^ci(d) =0 for all i l, 2,,, (10) is a direct cosequece of the followig theorem which we proved i [3]: // two matrices A ad B of the same order satisfy the same row hypothesis (i.e., (9)) which is, i tur, sufficiet to prove det.4^0 ad det B^0, the det (A +B) ^det A +det B. A determiat whose elemets satisfy (2) does ot ecessarily have a domiat diagoal uless A* = 0 (» = 1,,»). Ostrowski's result shows, however, that such determiats are ever egative, ad, as a corollary he proved that, if X is ay root of A, = mi f E a ~ Af\. Sice the determiat is the product of the roots, this would imply det A ^ ". This boud is, i geeral, less tha that give by Theorem 2. If A satisfies (2), let bij = O, bu = Af E a- j ** h The B?=At, ad B satisfies (2). Thus det 5^0. Let

19601 BOUNDS FOR DETERMINANTS WITH POSITIVE DIAGONALS 399 di = a ba = zl ai3- Af, i = 1, 2,,, 3-1 ad set D = dia.g(di, d2,, d). Sice (2) implies au^bu each i=l, 2,,. The (10) implies we have di?z0 for which proves Theorem 2. det A = det (B + D) ^ II o\ t=i Refereces 1. J. L. Breer, A boud for a determiat with domiat mai diagoal, Proc. Amer. Math. Soc. vol. 5 (1954) pp. 631-634. 2. -, Bouds for determiats. II, Proc. Amer. Math. Soc. vol. 8 (1957) pp. 532-534. 3. E. Haysworth, Note o bouds for certai determiats, Duke Math. J. vol. 24 (1957) pp. 313-320. 4. A. Ostrowski, Note o bouds for determiats with domiat pricipal diagoal, Proc. Amer. Math. Soc. vol. 3 (1952) pp. 26-30. 5. -, Note o bouds for some determiats, Duke Math. J. vol. 22 (1955) pp. 95-102. 6. G. B. Price, Determiats with domiat pricipal diagoal, Proc. Amer. Math. Soc. vol. 2 (1951) pp. 497-502. 7. H. Scheider, A iequality for latet roots applied to determiats with domiat pricipal diagoal, J. Lodo Math. Soc. vol. 28 (1953) pp. 8-20. Natioal Bureau of Stadards, Washigto, D. C.