Euler, Ioachimescu and the trapezium rule. G.J.O. Jameson (Math. Gazette 96 (2012), )

Similar documents
Review of Calculus, cont d

The Regulated and Riemann Integrals

An approximation to the arithmetic-geometric mean. G.J.O. Jameson, Math. Gazette 98 (2014), 85 95

Best Approximation. Chapter The General Case

7.2 The Definite Integral

Math 360: A primitive integral and elementary functions

Lecture 19: Continuous Least Squares Approximation

Advanced Calculus: MATH 410 Notes on Integrals and Integrability Professor David Levermore 17 October 2004

Math 554 Integration

n f(x i ) x. i=1 In section 4.2, we defined the definite integral of f from x = a to x = b as n f(x i ) x; f(x) dx = lim i=1

f(x)dx . Show that there 1, 0 < x 1 does not exist a differentiable function g : [ 1, 1] R such that g (x) = f(x) for all

UNIFORM CONVERGENCE MA 403: REAL ANALYSIS, INSTRUCTOR: B. V. LIMAYE

THE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS.

NUMERICAL INTEGRATION

The First Fundamental Theorem of Calculus. If f(x) is continuous on [a, b] and F (x) is any antiderivative. f(x) dx = F (b) F (a).

Overview of Calculus I

AN INTEGRAL INEQUALITY FOR CONVEX FUNCTIONS AND APPLICATIONS IN NUMERICAL INTEGRATION

Properties of Integrals, Indefinite Integrals. Goals: Definition of the Definite Integral Integral Calculations using Antiderivatives

Numerical Integration

Math& 152 Section Integration by Parts

Numerical Analysis: Trapezoidal and Simpson s Rule

Unit #9 : Definite Integral Properties; Fundamental Theorem of Calculus

Further integration. x n nx n 1 sinh x cosh x log x 1/x cosh x sinh x e x e x tan x sec 2 x sin x cos x tan 1 x 1/(1 + x 2 ) cos x sin x

Definite integral. Mathematics FRDIS MENDELU

A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H. Thomas Shores Department of Mathematics University of Nebraska Spring 2007

Exam 2, Mathematics 4701, Section ETY6 6:05 pm 7:40 pm, March 31, 2016, IH-1105 Instructor: Attila Máté 1

Definite integral. Mathematics FRDIS MENDELU. Simona Fišnarová (Mendel University) Definite integral MENDELU 1 / 30

CMDA 4604: Intermediate Topics in Mathematical Modeling Lecture 19: Interpolation and Quadrature

Integral points on the rational curve

The Riemann Integral

Lecture 1: Introduction to integration theory and bounded variation

1 The Lagrange interpolation formula

Convex Sets and Functions

Physics 116C Solution of inhomogeneous ordinary differential equations using Green s functions

Calculus I-II Review Sheet

We know that if f is a continuous nonnegative function on the interval [a, b], then b

New Expansion and Infinite Series

Sections 5.2: The Definite Integral

1 The Riemann Integral

Lecture 1. Functional series. Pointwise and uniform convergence.

38 Riemann sums and existence of the definite integral.

II. Integration and Cauchy s Theorem

NUMERICAL INTEGRATION. The inverse process to differentiation in calculus is integration. Mathematically, integration is represented by.

1. Gauss-Jacobi quadrature and Legendre polynomials. p(t)w(t)dt, p {p(x 0 ),...p(x n )} p(t)w(t)dt = w k p(x k ),

W. We shall do so one by one, starting with I 1, and we shall do it greedily, trying

Math 1B, lecture 4: Error bounds for numerical methods

Lecture 14: Quadrature

The Fundamental Theorem of Calculus. The Total Change Theorem and the Area Under a Curve.

Czechoslovak Mathematical Journal, 55 (130) (2005), , Abbotsford. 1. Introduction

Chapter 0. What is the Lebesgue integral about?

Review on Integration (Secs ) Review: Sec Origins of Calculus. Riemann Sums. New functions from old ones.

arxiv:math/ v2 [math.ho] 16 Dec 2003

Improper Integrals. Type I Improper Integrals How do we evaluate an integral such as

ODE: Existence and Uniqueness of a Solution

Chapter 1. Basic Concepts

Review of Riemann Integral

1 i n x i x i 1. Note that kqk kp k. In addition, if P and Q are partition of [a, b], P Q is finer than both P and Q.

Main topics for the First Midterm

Theoretical foundations of Gaussian quadrature

SOLUTIONS FOR ANALYSIS QUALIFYING EXAM, FALL (1 + µ(f n )) f(x) =. But we don t need the exact bound.) Set

1.9 C 2 inner variations

The Riemann-Lebesgue Lemma

Review of basic calculus

The Henstock-Kurzweil integral

Chapters 4 & 5 Integrals & Applications

Keywords : Generalized Ostrowski s inequality, generalized midpoint inequality, Taylor s formula.

Some estimates on the Hermite-Hadamard inequality through quasi-convex functions

UNIFORM CONVERGENCE. Contents 1. Uniform Convergence 1 2. Properties of uniform convergence 3

P 3 (x) = f(0) + f (0)x + f (0) 2. x 2 + f (0) . In the problem set, you are asked to show, in general, the n th order term is a n = f (n) (0)

f(x) dx, If one of these two conditions is not met, we call the integral improper. Our usual definition for the value for the definite integral

Math Calculus with Analytic Geometry II

Properties of the Riemann Integral

Euler-Maclaurin Summation Formula 1

Section 6.1 INTRO to LAPLACE TRANSFORMS

Riemann Sums and Riemann Integrals

11 An introduction to Riemann Integration

Math 118: Honours Calculus II Winter, 2005 List of Theorems. L(P, f) U(Q, f). f exists for each ǫ > 0 there exists a partition P of [a, b] such that

The Banach algebra of functions of bounded variation and the pointwise Helly selection theorem

Lecture 3. Limits of Functions and Continuity

III. Lecture on Numerical Integration. File faclib/dattab/lecture-notes/numerical-inter03.tex /by EC, 3/14/2008 at 15:11, version 9

Week 10: Riemann integral and its properties

3.4 Numerical integration

Orthogonal Polynomials and Least-Squares Approximations to Functions

INTRODUCTION TO INTEGRATION

7.2 Riemann Integrable Functions

x = b a N. (13-1) The set of points used to subdivide the range [a, b] (see Fig. 13.1) is

Riemann Sums and Riemann Integrals

MAA 4212 Improper Integrals

Homework 4. (1) If f R[a, b], show that f 3 R[a, b]. If f + (x) = max{f(x), 0}, is f + R[a, b]? Justify your answer.

63. Representation of functions as power series Consider a power series. ( 1) n x 2n for all 1 < x < 1

Advanced Computational Fluid Dynamics AA215A Lecture 3 Polynomial Interpolation: Numerical Differentiation and Integration.

Principles of Real Analysis I Fall VI. Riemann Integration

SYDE 112, LECTURES 3 & 4: The Fundamental Theorem of Calculus

Definition of Continuity: The function f(x) is continuous at x = a if f(a) exists and lim

MA 124 January 18, Derivatives are. Integrals are.

Integration Techniques

Suppose we want to find the area under the parabola and above the x axis, between the lines x = 2 and x = -2.

Lecture 20: Numerical Integration III

Taylor Polynomial Inequalities

New Integral Inequalities for n-time Differentiable Functions with Applications for pdfs

Transcription:

Euler, Iochimescu nd the trpezium rule G.J.O. Jmeson (Mth. Gzette 96 (0), 36 4) The following results were estblished in recent Gzette rticle [, Theorems, 3, 4]. Given > 0 nd 0 < s <, let x n = y n = n n ( + r) [ ( + n) s s], s s ( + r) [ ( + n ) s s] s s u n = (x n + y n ) = y n ( + n ) s Then (x n ) is incresing, (y n ) is decresing, nd both tend to limit, denoted by I s (), s n. Further, when n, n s (I s () x n ), ns (y n I s ()), ns+ (I s () u n ) s. The methods re quite lengthy, nd give the ppernce of being specific to this cse. Here we present different, rgubly simpler, pproch to results of this sort. Without ny extr effort, it delivers more generl version: x n is replced by n f(r) n f(x)dx, 0 where f(x) is function stisfying suitble conditions; in fct, the results gin in both clrity nd simplicity when presented in this wy. The first step, the convergence of (x n ) nd (y n ), is ctully no more thn the fmilir process leding to the existence of Euler s constnt γ. The hrder prt is the derivtion of the other limits. These re obtined in [] using vrious series expnsions. Our method is bsed insted on n estimte for the error in the trpezium rule which is well known to specilists in numericl nlysis, but possibly less so to Gzette reders. The uthor hopes tht it will be of interest to some of them. We describe two proofs, both quite elementry. We consider differentible function f(x) tht is positive nd decresing for x 0 nd stisfies f(x) 0 s x. Let I n = S n = f(0) + f() + + f(n ), n 0 f(x) dx, J r = r+ r f(x) dx, n = S n I n, Γ n = S n+ I n,

so tht Γ n = n + f(n). Clerly, I n = J 0 + J + + J n, so n n = [f(r) J r ]. () The bsic results bout these quntities derive from the following very simple inequlity. Since f(x) is decresing, we hve f(r + ) f(x) f(r) for r x r +, hence f(r + ) J r f(r). () By (), it follows tht n 0. Further: Proposition : ( n ) is incresing nd (Γ n ) is decresing. Both converge to the sme limit (sy L) s n, nd n L Γ n for ll n. Proof. We hve n+ n = f(n) J n 0, Γ n+ Γ n = f(n + ) J n 0. So Γ n is bounded below (by 0) nd decresing, hence it tends to limit, L, nd Γ n L for ll n. Since Γ n n = f(n) 0 s n, n lso tends to L. Exmple : The best known exmple of this process is given by f(x) = /(x + ). Then n = n r= log(n + ), nd L is Euler s constnt γ. r of [], Exmple : Let f(x) = /( + x) s, where > 0 nd 0 < s <. Then, in the nottion n = x n, Γ n = y n+, L = I s (). nd hence By (), we hve L = [f(r) J r ], (3) L n = [f(r) J r ], (4) Insted of (), we now consider the trpezium rule estimte for J r, tht is, T r = f(r) + f(r + ). This is the integrl of the liner function greeing with f(x) t r nd r +. In most cses, it is much more ccurte estimte of the integrl thn either f(r) or f(r + ). If f is convex

(i.e. curving upwrds), then it is obvious from digrm tht J r T r ; forml proof is contined in Theorem 3 below. A sufficient condition for convexity is f (x) 0 for ll x. It is stisfied by mny functions of the type we re considering, including the specific exmples given bove. Note tht Then hence n T r = f(0) + n f(r) + f(n) = S n f(0) + f(n). r= To tke dvntge of the better pproximtion given by T r, we introduce Λ n = n + Γ n = n + f(n). Λ n = S n + f(n) I n n = (T r J r ) + f(0), L Λ n = (T r J r ). (5) Compre this with (4). For convex f, these formule show t once tht (Λ n ) is incresing nd L Λ n. In the nottion of [], Λ n = u n+. We now estblish n estimte for the error in the trpezium rule, nd use it to derive pir of inequlities for L Λ n which in turn will imply the limits stted in []. The first step is to give n estimte for the error in liner pproximtion to function. This is ctully the cse n = of the more generl result on the polynomil interpolting function t n points (e.g. [, p. 4]). The proof is plesnt ppliction of Rolle s theorem. Proposition : Let f be twice differentible on [, b], nd let p(x) be the liner function greeing with f(x) t nd b. Let q(x) = (x )(b x). Then, given x in (, b), there exists ξ in (, b) such tht p(x) f(x) = q(x)f (ξ). Proof: We prove the sttement for chosen point x 0 in (, b). Let G(x) = p(x) f(x) kq(x), with k chosen so tht G(x 0 ) = 0, hence p(x 0 ) f(x 0 ) = kq(x 0 ). We hve to show tht k = f (ξ) for some ξ. Now G() = G(b) = G(x 0 ) = 0. By Rolle s theorem, pplied twice, there exists ξ in (, b) such tht G (ξ) = 0. Now p (x) = 0 nd q (x) =, so G (x) = f (x) + k for ll x. Hence k = f (ξ), s required. Then Theorem 3: Suppose tht m f (x) M on [, b], nd let T (f) = (b )[f()+f(b)]. m(b )3 T (f) 3 f(x) dx M(b )3.

Proof: T (f) = p(x) dx, where p(x) is s bove. Proposition, Write b = h nd substitute x = y: q(x) dx = mq(x) p(x) f(x) Mq(x). h So [p(x) f(x)] dx lies between mh3 nd Mh3. 0 y(h y) dy = [ hy 3 y3] h 0 = 6 h3. Now q(x) 0 on [, b], so by Note: We puse to sketch second, eqully ttrctive, proof of Theorem 3. c = ( + b) nd f(x) dx = I(f). Integrtion by prts gives (x c)f (x) dx = = [ ] b (x c)f(x) f(x) dx (b )[f(b) + f()] I(f) = T (f) I(f). Write Now integrte by prts the other wy round! With q(x) s bove, we hve q (x) = +b x = (c x), so we cn use q(x) s the ntiderivtive of x c. We obtin since q() = q(b) = 0. T (f) I(f) = sttement follows s before. [ = ] b q(x)f (x) + q(x)f (x) dx. q(x)f (x) dx Agin the integrnd lies between mq(x) nd Mq(x), nd the Clerly, if f (x) is decresing, then Theorem 3 pplies with M = f () nd m = f (b). Hence for T r nd J r defined s bove, if f (x) is decresing, then We cn now stte our bsic result on L Λ n. Theorem 4: Suppose tht f (r + ) T r J r f (r). () f(x) is positive nd decresing for x 0, (b) f(x) nd f (x) tend to 0 s x, (c) f (x) is positive nd decresing for x 0. Then f (n + ) L Λ n f (n ). 4

(Note tht f (x) is negtive.) Proof: By (5) nd Theorem 3, provided tht r= f (r) converges, we hve f (r) L Λ n f (r). + Since f (x) is decresing, () (pplied to successive intervls) gives f (r) f (x) dx = f (n ) nd + f (r) n n+ f (x) dx = f (n + ). The list of conditions on f(x) might seem rther long, but it is very esily seen tht they re ll stisfied by /( + x) s (where s > 0). Actully, with bit of effort, one cn show tht the condition f (x) 0 follows from the others. Exmple 3: To pply this to γ, tke f(x) = /(x + ), nd replce n by n in Theorem 4. Then Λ n = n r= r + n log n, nd we deduce tht γ = Λ n + R n, where (n + ) R n (n ). Even with n quite smll, this gives good pproximtion to γ. For exmple, when n = 4, the resulting lower nd upper bounds for γ re 0.57537 nd 0.5830. The sme work pplies, rther more directly, to the estimtion of the til of convergent series. Suppose tht f stisfies (), (b), (c) nd tht f(x) dx is convergent. Then, by (), n= f(n) is convergent (this is the integrl test for convergence ). Write Sn = f(n), In = f(x) dx. It is fmilir fct tht I n pproximtes S n in some sense. We cn now describe this pproximtion rther ccurtely. Clerly, T r = S n f(n) nd J r = I n. So the proof of Theorem 4 gives: Theorem 5: Under these conditions, Sn = In + f(n)+r n, where f (n+) R n f (n ). Exmple 4: Let f(x) = /x s, where s >. The sum of the series n= /ns is the Riemnn zet function ζ(s). In the nottion of Theorem 5, S n = (s )n s + n s + R n, 5 n

where s (n + ) s+ R n (n ) s+. Note: If f (x) is convex, then refinement of the second proof of Theorem 3 (which we will not describe here) leds to T r J r [f (r + ) f (r)]. This mens tht the f (n ) in Theorem 4 cn be replced by f (n), so tht the n cn be replced by n in Exmples 3 nd 4. Finlly, we return to the limits stted t the beginning. We ssume two further conditions which re clerly stisfied by /( + x) s. Agin, we will not tke ny trouble investigting the extent to which some of the conditions re implied by the others. Theorem 6: Let f(x) be s in Theorem 4, nd ssume further: (d) f (x + ) f (x) s x, (e) f (x) f(x) 0 s x. Then, when n, L Λ n f(n) 0, L Λ n f (n) L n f(n),, Γ n L f(n). Proof. The first sttement follows t once from Theorem 4 nd (d), since (d) lso implies tht f (x )/f (x) s x. Condition (e) now gives the second sttement. The lst two sttements follow t once, since n = Λ n f(n) nd Γ n = Λ n + f(n). This illustrtes nicely the fct tht Λ n is better pproximtion to L thn n or Γ n. These limits reproduce the ones from []. Consider, for exmple, the first one. With f(x) = /( + x) s nd the nottion of [], we hve so tht L Λ n f (n ) = (I s() u n ) ( + n )s+, s n s+ sn s+ L Λ n (I s () u n ) = ( + n ) s+ f (n ) s s n. A further limit stted in [] is lim n n s+ (z n I s ()) = s/6, where z n = (x n + y n ) (Theorem 3). Recll tht y n = Γ n. We leve it s n exercise for the reder to show tht under our conditions, lim n (Γ n Γ n )/f (n) = nd to derive the limit just stted. 6

References. Alin Sintmrin, Regrding generlistion of Iochimescu s constnt, Mth. Gzette 94 (00), 70 83.. Lee W. Johnson nd R. Den Riess, Numericl Anlysis, Addison-Wesley (98). Dept. of Mthemtics nd Sttistics, Lncster University, Lncster LA 4YF, UK e-mil: g.jmeson@lncster.c.uk 7