Bernoulli Polynomials

Similar documents
Some Fun with Divergent Series

10.1 Sequences. Example: A sequence is a function f(n) whose domain is a subset of the integers. Notation: *Note: n = 0 vs. n = 1.

ζ (s) = s 1 s {u} [u] ζ (s) = s 0 u 1+sdu, {u} Note how the integral runs from 0 and not 1.

Bernoulli polynomials

Completion Date: Monday February 11, 2008

n=1 ( 2 3 )n (a n ) converges by direct comparison to

Convergence of sequences and series

8.1 Sequences. Example: A sequence is a function f(n) whose domain is a subset of the integers. Notation: *Note: n = 0 vs. n = 1.

3.4 Introduction to power series

Series Solutions. 8.1 Taylor Polynomials

Math 0230 Calculus 2 Lectures

Review: Power series define functions. Functions define power series. Taylor series of a function. Taylor polynomials of a function.

THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE DEPARTMENT OF MATHEMATICS BERNOULLI POLYNOMIALS AND RIEMANN ZETA FUNCTION

6 Lecture 6b: the Euler Maclaurin formula

MATH115. Infinite Series. Paolo Lorenzo Bautista. July 17, De La Salle University. PLBautista (DLSU) MATH115 July 17, / 43

Section Taylor and Maclaurin Series

MOMENTS OF HYPERGEOMETRIC HURWITZ ZETA FUNCTIONS

4 Uniform convergence

Bernoulli Numbers and their Applications

Taylor Series. Math114. March 1, Department of Mathematics, University of Kentucky. Math114 Lecture 18 1/ 13

Part 3.3 Differentiation Taylor Polynomials

f (x) = k=0 f (0) = k=0 k=0 a k k(0) k 1 = a 1 a 1 = f (0). a k k(k 1)x k 2, k=2 a k k(k 1)(0) k 2 = 2a 2 a 2 = f (0) 2 a k k(k 1)(k 2)x k 3, k=3

SECTION A. f(x) = ln(x). Sketch the graph of y = f(x), indicating the coordinates of any points where the graph crosses the axes.

Summation Techniques, Padé Approximants, and Continued Fractions

Math 259: Introduction to Analytic Number Theory More about the Gamma function

Some Applications of the Euler-Maclaurin Summation Formula

1 + lim. n n+1. f(x) = x + 1, x 1. and we check that f is increasing, instead. Using the quotient rule, we easily find that. 1 (x + 1) 1 x (x + 1) 2 =

A sequence { a n } converges if a n = finite number. Otherwise, { a n }

Math Assignment 11

Complex Series. Chapter 20

Mathematics 324 Riemann Zeta Function August 5, 2005

JUST THE MATHS UNIT NUMBER DIFFERENTIATION APPLICATIONS 5 (Maclaurin s and Taylor s series) A.J.Hobson

Mathematical Methods for Physics and Engineering

First, let me recall the formula I want to prove. Again, ψ is the function. ψ(x) = n<x

EQUIDISTANT SAMPLING FOR THE MAXIMUM OF A BROWNIAN MOTION WITH DRIFT ON A FINITE HORIZON

Evaluating Singular and Nearly Singular Integrals Numerically

Ma 530 Power Series II

NPTEL web course on Complex Analysis. A. Swaminathan I.I.T. Roorkee, India. and. V.K. Katiyar I.I.T. Roorkee, India

Considering our result for the sum and product of analytic functions, this means that for (a 0, a 1,..., a N ) C N+1, the polynomial.

Abstract. 2. We construct several transcendental numbers.

Summary: Primer on Integral Calculus:

Lecture 32: Taylor Series and McLaurin series We saw last day that some functions are equal to a power series on part of their domain.

TAYLOR AND MACLAURIN SERIES

AP Calculus Testbank (Chapter 9) (Mr. Surowski)

The University of British Columbia Final Examination - April 11, 2012 Mathematics 105, 2011W T2 All Sections. Special Instructions:

(b) Prove that the following function does not tend to a limit as x tends. is continuous at 1. [6] you use. (i) f(x) = x 4 4x+7, I = [1,2]

11.10a Taylor and Maclaurin Series

Complex Analysis Homework 9: Solutions

Introduction and Review of Power Series

APPENDIX. THE MELLIN TRANSFORM AND RELATED ANALYTIC TECHNIQUES. D. Zagier. 1. The generalized Mellin transformation

THE GAMMA FUNCTION AND THE ZETA FUNCTION

Power series solutions for 2nd order linear ODE s (not necessarily with constant coefficients) a n z n. n=0

MTH 133 Solutions to Exam 2 November 15, Without fully opening the exam, check that you have pages 1 through 13.

MTH4101 CALCULUS II REVISION NOTES. 1. COMPLEX NUMBERS (Thomas Appendix 7 + lecture notes) ax 2 + bx + c = 0. x = b ± b 2 4ac 2a. i = 1.

MATH 1242 FINAL EXAM Spring,

Fourier Sin and Cos Series and Least Squares Convergence

Taylor and Maclaurin Series

Infinite Series. 1 Introduction. 2 General discussion on convergence

e y [cos(x) + i sin(x)] e y [cos(x) i sin(x)] ] sin(x) + ey e y x = nπ for n = 0, ±1, ±2,... cos(nπ) = ey e y 0 = ey e y sin(z) = 0,

Second Midterm Exam Name: Practice Problems March 10, 2015

Notes for Expansions/Series and Differential Equations

8.5 Taylor Polynomials and Taylor Series

Is Analysis Necessary?

Section 11.1 Sequences

Power series and Taylor series

Last Update: March 1 2, 201 0

13 Definite integrals

Chapter 11 - Sequences and Series

The Value of the Zeta Function at an Odd Argument

cosh 2 x sinh 2 x = 1 sin 2 x = 1 2 cos 2 x = 1 2 dx = dt r 2 = x 2 + y 2 L =

Exercises for Part 1

Math 113 Winter 2005 Key

Solutions to Assignment-11

NORMAL NUMBERS AND UNIFORM DISTRIBUTION (WEEKS 1-3) OPEN PROBLEMS IN NUMBER THEORY SPRING 2018, TEL AVIV UNIVERSITY

f (r) (a) r! (x a) r, r=0

Math 113 (Calculus 2) Exam 4

Without fully opening the exam, check that you have pages 1 through 12.

Relevant sections from AMATH 351 Course Notes (Wainwright): Relevant sections from AMATH 351 Course Notes (Poulin and Ingalls):

AP Calculus (BC) Chapter 9 Test No Calculator Section Name: Date: Period:

MTH 133 Solutions to Exam 2 April 19, Without fully opening the exam, check that you have pages 1 through 12.

An Euler-Type Formula for ζ(2k + 1)

Complex Homework Summer 2014

Advanced Calculus Math 127B, Winter 2005 Solutions: Final. nx2 1 + n 2 x, g n(x) = n2 x

Conformal maps. Lent 2019 COMPLEX METHODS G. Taylor. A star means optional and not necessarily harder.

Friday 09/15/2017 Midterm I 50 minutes

Fourier Series. 1. Review of Linear Algebra

III. Consequences of Cauchy s Theorem

Research Article Operator Representation of Fermi-Dirac and Bose-Einstein Integral Functions with Applications

2 Write down the range of values of α (real) or β (complex) for which the following integrals converge. (i) e z2 dz where {γ : z = se iα, < s < }

Here are brief notes about topics covered in class on complex numbers, focusing on what is not covered in the textbook.

AP Calculus Chapter 9: Infinite Series

Complex Analysis Slide 9: Power Series

MATH 118, LECTURES 27 & 28: TAYLOR SERIES

MA2501 Numerical Methods Spring 2015

Analysis II: Basic knowledge of real analysis: Part V, Power Series, Differentiation, and Taylor Series

Asymptotic Expansions

MAS153/MAS159. MAS153/MAS159 1 Turn Over SCHOOL OF MATHEMATICS AND STATISTICS hours. Mathematics (Materials) Mathematics For Chemists

Without fully opening the exam, check that you have pages 1 through 12.

Topic 7 Notes Jeremy Orloff

MTH 133 Exam 2 November 16th, Without fully opening the exam, check that you have pages 1 through 12.

Transcription:

Chapter 4 Bernoulli Polynomials 4. Bernoulli Numbers The generating function for the Bernoulli numbers is x e x = n= B n n! xn. (4.) That is, we are to expand the left-hand side of this equation in powers of x, i.e., a Taylor series about x =. The coefficient of x n in this expansion is B n /n!. Note that we can write the left-hand side of this expression in an alternative form x e x = x e x/ ( e x/ e x/) = xex/ sinh x = x cosh x sinh x sinh x = x coth x x. (4.) Note that x coth x is an even function of x, while x is odd. Therefore we conclude that all but one of the Bernoulli numbers of odd order are zero: By writing x = iy and noting that B = (4.3a) B k+ =, k =,, 3,.... (4.3b) coth iy = i cot y, (4.4) 9 Version of September 6,

3 Version of September 6, CHAPTER 4. BERNOULLI POLYNOMIALS we conclude that or iy coth iy = y cot y = n= B n (iy) n, (4.5) (n)! y cot y = () n B n (n)! yn. (4.6) n= By straightforward expansion in powers of x we can read off the first few Bernoulli numbers: x coth x = x cosh x sinh x x ( + x ) (! + x ) 4 ( 4! + x ) 6 ( 6! + x ) 8 8! +... x + ( x ) 3 ( 3! + x ) 5 ( 5! + x ) 7 ( 7! + x ) 9 9! +... + ) ) 4 ) 6 ) 8 + + +! 4! 6! 8! { ) ) 4 ) 6 ) 8 + + + 3! 5! 7! 9! ) ) 4 ) 6 + + + 3! 5! 7! ) ) 4 3 ) 4 } + + + O(x ) 3! 5! 3! =... ( ) ( = + x + x4 ) ( ) ( + x6 + x8 ) +....! 6 4! 3 6! 4 8! 3 (4.7) So by comparison with Eq. (4.) we find B =, B = 6, B 4 = 3, B 6 = 4, B 8 = 3. (4.8) What is the radius of convergence of the series z e z = z + B n (n)! zn? (4.9) Recall that a power series converges everywhere within its circle of convergence, and diverges outside that circle. Since a uniformly convergent series must converge to a continuous function, the power series must converge to a well-behaved function within the circle of convergence. That is, the limit function must have a singularity somewhere on the circle of convergence, but must be singularityfree within the circle of convergence. The precise theorem, proved in Chapter n=

4.. BERNOULLI NUMBERS 3 Version of September 6, n B n Asymptotic value Relative error 3% 6 π 39% 3 3 π 7.6% 4 45 3 4 π.7% 6 4 3 35 π.4% 8 5 475 5 66 π.99% 6 69 73 467775 π.5% 7 7 6 8 367 5 6385875 9 43867 798 456755 4π 4.6% π 6.5% 9769469875 π 8.38% 746 33 987846385 π.95% Table 4.: The Bernoulli numbers B n for n from to, compared with the asymptotic values (4.). The last column shows the relative error of the asymptotic estimate. Note that the later rather rapidly approaches the true value. 5, is that the radius of convergence of a power series is the distance from the origin to the nearest singularity of the function the series represents. In this case, it is clear that the generating function is singular wherever e z =, except for z =. Thus the closest singularities to the real axis occur at ±πi, so that the radius of convergence is π. On the other hand (π) = ρ B n (n + )! = lim = lim (n + )(n + ) n (n)! B (n+) n B n B n+, (4.) from which we can infer the fact that the Bernoulli numbers grow rapidly with n, B n (n)!, n. (4.) (π) n We cannot deduce the sign or overall constant from this analysis: The true asymptotic behavior of B n is n+ (n)! B n (). (4.) (π) n The table shows the relative accuracy of the asymptotic approximation (4.).

3 Version of September 6, CHAPTER 4. BERNOULLI POLYNOMIALS 4. Bernoulli Polynomials The Bernoulli polynomials are defined by the generating function that is, according to Eq. (.94), F(x, s) = xexs e x = B n (s) = n= B n (s) xn n!, (4.3) ( ) n F(x, s) x. (4.4) x= From the properties of F(x, s) we can deduce all the properties of these polynomials:. Note that F(x, ) = x e x = x n B n n!. (4.5) Therefore, we conclude that the Bernoulli polynomials at zero are equal to the Bernoulli numbers, B n () = B n. (4.6). Next we notice that n= F(x, ) = xex e x = x x = ex e x = F(x, ), (4.7) so that by comparing corresponding terms in the generating function expansion, we find B n () = () n B n () = () n B n. (4.8) 3. If we differentiate the generating function with respect to its second argument, we obtain the relation But obviously s F(x, s) = x e xs e x = n= x e xs e x = xf(x, s) = n= so equating coefficients of x n /n! we conclude that B n (s)xn n!. (4.9) x n+ B n, (4.) n! B n(s) = nb n (s). (4.)

4.3. EULER-MACLAURIN SUMMATION FORMULA33 Version of September 6, (Note that B (s) = is consistent with this if B (s) is finite.) Again, by direct power series expansion of the generating function we can read off the first few Bernoulli polynomials: F(x, s) x + xs + (xs) + 6 (xs)3 x + x + 3! ( x3 + x s ) ( s + x s 6 + ) +..., (4.) 4 from which we read off B (s) =, B (s) = s, B (s) = s s + 6. (4.3a) (4.3b) (4.3c) By keeping two more terms in the expansion we find B 3 (s) = s 3 3 s + s, (4.3d) B 4 (s) = s 4 s 3 + s 3. (4.3e) Note that the properties (4.6) and (4.8) are satisfied. Note further we can use the property (4.) to derive higher Bernoulli polynomials from lower ones. Thus from Eq. (4.3c) we know that B 3(s) = 3s 3s +. (4.4) The expression for B 3 (s), (4.3d) is recovered, when it is recalled that B 3 =. 4.3 Euler-Maclaurin Summation Formula Using the above recursion relation (4.) we can deduce a very important formula which allows a precise relation between a discrete sum and a continuous integral. First note that since B = B (s) = we can write f(x)b (x)dx = f(x)dx, (4.5) valid for any function f. But now we can integrate by parts using B (x) = B (x) : (4.6) f(x)dx = f(x)b (x)dx

34 Version of September 6, CHAPTER 4. BERNOULLI POLYNOMIALS = f(x)b (x) Here, we have used the facts that x= f (x)b (x)dx = f() + f() f (x)b (x)dx. (4.7) B () = B =, B () = B =. (4.8a) (4.8b) Now we can continue integrating by parts by noting that B (x) = B (x), (4.9) so that f(x)dx = f() + f() f ()B () f ()B () + f (x)b (x)dx = f() + f() B f () f () + f (x)b (x)dx. (4.3) A general pattern is emerging. Let us assume the following formula holds for some integer k (we have just proved it for k = ): f(x)dx = f() + f() k m= + (k)! B m (m)! f (m) () f (m) () f (k) (x)b k (x)dx. (4.3) We shall then prove that the same formula holds for k k +, thereby establishing this formula, the Euler-Maclaurin summation formula, for all k. We proceed as follows. Note that B k (x) = B k+ (x) k + = B k+ (x) (k + )(k + ), (4.3) so that by integrating by parts, we rewrite the last term in Eq. (4.3) as (k)! f (k) (x) (k + )(k + ) B k+ (x)dx

4.3. EULER-MACLAURIN SUMMATION FORMULA35 Version of September 6, = = (k + )! (k + )! + where we have noted that for k > f (k) ()B k+() f (k) ()B k+() f (k+) (x)b k+(x)dx f (k+) ()B k+ () + f (k+) ()B k+ () f (k+) (x)b k+ (x)dx, (4.33) B k+ () = (k + )B k+() =, B k+() = (k + )B k+ () = (k + )B k+ () =. (4.34a) (4.34b) Hence f(x)dx = f() + f() + k+ m= B m (m)! (k + )! f (m) () f (m) () f (k+) (x)b k+ (x)dx. (4.35) This is exactly Eq. (4.3) with k replaced by k + ; so since the formula is true for k = it is true for all integers k. Notice that the last term in this formula, the remainder, can also be written in the form (k + 3)! Now consider the integral (N a positive integer) N f(s)ds = N k= k+ k f (k+3) (x)b k+3 (x)dx. (4.36) f(s)ds = N k= f(k + t)dt, (4.37) where we have introduced a local variable t. For the latter integral, we can use the Euler-Maclaurin sum formula, which here reads f(k + t)dt = f(k + ) + f(k) n B m f (m) (k + ) f (m) (k) (m)! m= + (n)! f (n) (k + t)b n (t)dt. (4.38)

36 Version of September 6, CHAPTER 4. BERNOULLI POLYNOMIALS Now when we sum the first term here on the right-hand side over k we obtain N k= N f(k + ) + f(k) = k= while the second term when summed on k involves N k= Thus we find f(k) f() + f(n), (4.39) f (m) (k + ) f (m) (k) = f (m) (N) f (m) (). (4.4) N f(s)ds = N k= f(k) f() + f(n) n m= + (n)! (m)! B m f (m) (N) f (m) () N k= f (n) (t + k)b n (t)dt. (4.4) Equivalently, we can write this as a relation between a finite sum and an integral, with a remainder R n : N f(k) = k= N where the remainder + f(s)ds + f() + f(n) n m= R n = (n)! (m)! B m f (m) (N) f (m) () + R n, (4.4) N k= f (n) (t + k)b n (t)dt. (4.43) is often assumed to vanish as n. Note that the remainder can also be written as R n = N f (n) (t)b n (t t ) dt, (4.44) (n)! where t signifies the greatest integer less than or equal to t. 4.3. Examples. Use the Euler-Maclaurin formula to evaluate the sum N n= cos(πn/n). N n= cos πn N N = dn cos πn N + ( + ) + =, (4.45)

4.3. EULER-MACLAURIN SUMMATION FORMULA37 Version of September 6, because and N f (m) () = f (m) (N) = (4.46) dn cos nπ N = N π π Of course, the sum may be carried out directly, N n= cos πn N = = N (e iπn/n + e iπn/n) dx cosx =. (4.47) e πi(n+)/n + eπi(n+)/n e πi/n e πi/n = ( + ) =. (4.48). The following sum occurs, for example, in computing the vacuum energy in a cosmological model: (l + )e l(l+)t. (4.49) l= How does this behave as t? We will answer this question by using the Euler-Maclaurin formula assuming that the remainder R n tends to zero as n. Thus we will write the limiting form of that sum formula as f(l) = l= dl f(l) + f( ) + f() + k= B k f (k) ( ) f (k) (). (4.5) (k)! Here f(l) = (l + )e l(l+)t, (4.5) so that f( ) = f (k) ( ) =, (4.5) while a very simple calculation shows f() =, (4.53a) f () = t, (4.53b) f () = t + t t 3, (4.53c) f (5) () = t 8t 3 + 3t 4 t 5, (4.53d) f (7) () = 68t 3 + 336t 4 84t 5 + 56t 6 t 7, (4.53e) f (k) () = O(t 4 ), k 5. (4.53f)

38 Version of September 6, CHAPTER 4. BERNOULLI POLYNOMIALS Thus Eq. (4.5) yields (l + )e l(l+)t = l= = t dl (l + )e l(l+)t + B f () B 4 4! f ()... du e u + + ( ) (t ) 6 + 4! ( 3 ) t + O(t ) + O(t ) = t + 3 + t 5 + 4 35 t + 35 t3 +.... (4.54) Here the integral was evaluated by making the substitution u = l(l + )t, du = (l+)t dl, and in the last line we have displayed the next two terms in this asymptotic expansion for small t. 3. The Riemann zeta function (.5) is defined by ζ(α) = nα, Re α >. (4.55) n= Suppose we approximate this by the first M terms in the sum occurring in the Euler-Maclaurin formula (4.4): ζ(α, M) = α + M m= B m (m)! f(m) (), (4.56) where f(n) = n α, and the first two terms here come from the integral and the f() terms in the EM formula. It is easy to see that f (m) () = Γ(α + m ). (4.57) Γ(α) Given the asymptotic behavior of the Bernoulli numbers in (4.), it is apparent that the limit M of ζ(α, M) does not exist. This limit is an example of an asymptotic series. However, in Table 4. we compare the sum of the first N terms of the series in (4.55) with the first N terms in the series defined by (4.56), that is ζ(α, N), for α = 3, where ζ(3) =.569. The original series converges monotonically to the correct limiting value, but not spectacularly fast. For N = 9 terms, the relative error is about.5%. The asymptotic series is divergent; however, the N = term is in error by only 4%, and the average of the N = and N = is larger than the true value by only +.5%. This illustrates a characteristic feature of asymptotic series: A few terms in the series approximates the function rather well, but as more and more terms are included the series deviates from the true value by an ever increasing amount.

4.3. EULER-MACLAURIN SUMMATION FORMULA39 Version of September 6, N N n= n3 ζ(3, N) ζ(3, N) + ζ(3, N + ).5.8.5.667.8 3.6.5.75 4.777..38 5.857.567.694 6.93.86 7.93 8.64 8.95 5.669 9.965 47.564 Table 4.: Two approximations compared for ζ(3) =.6...: N terms in the defining series (4.55) and N terms (without the remainder) in the Euler- Maclaurin sum (4.56). The former converges monotonically to the limit from below, while the later diverges, yet approximates the true value to better than % for low values of N.