MAT 5930. Analysis for Teachers.

Wm C Bauldry

[email protected]

Fall ’13

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 1 / 147 Nordkapp, Norway. Today

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 2 / 147 Introduction and Considered

1 Course Introduction (Class page, Course Info, Syllabus)

2 Calculus Considered 1 A Standard Freshman Calculus Course B Topics List: MAT 1110; MAT 1120 B Refer to texts by Thomas (traditional), Stewart (very popular, but. . . ), and Hughes-Hallett, et al, or Ostebee & Zorn (“reform”). B Historical: L’Hopital’sˆ Analyse des Infiniment Petits pour l’Intelligence des Lignes Courbes, Cauchy’s Cours d’Analyse, and Granville, Smith, & Longley’s Elements of the Differential and Calculus

2 An AP Calculus Course(AB,BC) 1 Functions, Graphs, and Limits Representations, one-to-one, onto, inverses, analysis of graphs, limits of functions, asymptotic behavior, continuity, 2 Derivatives Concept, definitions, interpretations, at a point, as a , second derivative, applications, computation, numerical approx. 3 Concept, definitions, interpretations, properties, Fundamental Theorem, applications, techniques, applications, numerical approx.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 3 / 147 Calculus Considered

2 (Calculus Considered) 3 Calculus Problems §1 Precalculus material: function, induction, summation, slope, trigonometry §2 Limits and Continuity: Squeeze Theorem, discontinuity, removable discontinuity, different interpretations of limit expressions §3 Derivatives: trigonometric derivatives, power rule, indirect methods, Newton’s method, Mean Value Theorem,“Racetrack Principle” §4 Integration: Fundamental Theorem, Riemann sums, parts, multiple integrals §5 Infinite Series: geometric, integrals and series, partial fractions, convergence tests (ratio, root, comparison, integral), Taylor & Maclaurin

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 4 / 147 A Practical Rationale: Growth in AP Calculus

See: MAA-NCTM Joint Position Statement on Calculus of March, 2012.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 5 / 147 AP Calculus 2010/11 Results

North Carolina AP Score Distributions, 2010 – 2012

AB 2010 2011 2012 BC 2010 2011 2012 n 6,797 6,860 6,946 n 3,070 3,176 3,426 x¯ 2.63 2.63 2.78 x¯ 3.56 3.50 3.58 x ≥ 3 50.0% 50.3% 54.7% x ≥ 3 75.4% 74.3% 76.6%

ASU: MAT 1110 credit (4 cr) MAT 1110 & 1120 (8) UNC-CH: MATH 231 (3 cr) MATH 231 & 232 (6)

c 2013 The College Board.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 6 / 147 A Little Background

Our goal is A rigorous study of elementary calculus extending to a treatment of fundamental concepts of analysis involving functions of a real variable. (See “Course Objectives.”)

What this course is not A “refresher course” in calculus. A “how to” course for teaching AP calculus.

What this course does Consider calculus from an advanced view—that is, as . Relate analysis concepts to the calculus classroom for depth.

Develop the profound understanding of fundamental . . . need[ed] to become accomplished mathematics teachers. — Liping Ma in Knowing and Teaching Elementary Mathematics.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 7 / 147 Notes to Analysis Students: Learning Mathematics

According to Paul Halmos, to learn mathematics, you have to know:

The ‘What’: Know the definitions, statements of theorems, and formulas What is an open set? When does a function have to have a maximum? What is the limit formula for the derivative?

The ‘How’: Apply techniques and use formulas to solve problems Find the minimum of the derivative on an . Determine where a function is continuous. Calculate the sum of the reciprocals of the squares.

The ‘Why’: Understand why a technique works, why a theorem is true Why does a bounded infinite set have to have a limit point? Why does continuity imply integrability? Why can antiderivatives compute definite integrals?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 8 / 147 Notes to Analysis Students: Doing Mathematics

George Polya’s´ Four Steps in Problem Solving:

Step 1: Understand the Problem (“see clearly what is required”) What is the unknown? What are the data? Step 2: Make a Plan (“the idea of the solution”) Do you know a related problem? How are the items (the data and the unknown) connected? Step 3: Carry Out the Plan Check each step. Can you see clearly and prove that the step is correct? Step 4: Look Back at the Completed Solution (“review and discuss it”) Reconsider, reexamine the result & the path that led to it. Can you check the result? The argument?

For a complete version see “How to Solve It”

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 9 / 147 Notes to Analysis Students: Teaching Mathematics

George Polya’s´ Ten Commandments for Teachers:

1 Be interested in your subject. 2 Know your subject. 3 Know about the ways of learning: The best way to learn anything is to discover it by yourself. 4 Try to read the faces of your students, try to see their expectations and difficulties, put yourself in their place. 5 Give them not only information, but “know-how,” attitudes of mind, the habit of methodical work. 6 Let them learn guessing. 7 Let them learn proving. 8 Look out for such features of the problem at hand as may be useful in solving the problems to come—try to disclose the general pattern that lies behind the present concrete situation. 9 Do not give away your whole secret at once—let the students guess before you tell it—let them find out by themselves as much as is feasible. 10 Suggest it, do not force it down their throats.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 10 / 147 Fundamental Concept: Function

Who First used the term function? – Leibniz1 (1646 –1716)

Introduced f(x) notation and originally thought a function must be a single algebraic expression? – Euler (1707–1783) Shocked the math world by presenting a bizarre function? – Weierstrass (1815 –1897) Introduced the formal definition of function in terms of sets of ordered pairs? – Homework! When did the function concept become central in school math? – 1920’s

See: “The Function Concept in School Mathematics,” H Hamley, Math Gazette, 18(229), 169-179 Homework: Read “Subject-Matter Knowledge and Pedagogical Content Knowledge: Prospective Secondary Teachers and the Function Concept,” R Even, JRME 24, No. 2. (1993), pp. 94-116. 1See: Jeff Miller’s website Earliest Known Uses of Some of the Words of Mathematics Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 11 / 147 The Definition of Function

Definition (Function) When two variables are so related that the value of the first variable is determined when the value of the second variable is given, then the first variable is said to be a function of the second. From Elements of the Differential and Integral Calculus by Granville, Smith, & Longley (1934).

Definition (Function) Let X and Y be nonempty sets. Let f be a collection of ordered pairs (x, y) with x ∈ X and y ∈ Y. Then f is a function from X to Y if to every x ∈ X there is assigned a unique y ∈ Y. From Calculus and Analytic Geometry by Thomas (1968).

Definition (Function) A function is a rule for assigning to each member of one set, called the domain, one (and only one) member of another set, called the range. From Calculus from Graphical, Numeric, and Symbolic Points of View by Ostebee & Zorn (2002).

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 12 / 147 Elementary Properties of Functions

Definition Let f :X → Y be a function.

Well-defined: If x1 = x2, then f(x1) = f(x2).

One-to-one: If f(x1) = f(x2), then x1 = x2; or if x1 6= x2, then f(x1) 6= f(x2). Onto: For each y, there is an x such that f(x) = y.

Inverse image: If B ⊆ Y, then f −1(B) = {x ∈ X | f(x) ∈ B}.

Graph: The set of pairs {(x, f(x)) | x ∈ X} ⊂ X × Y is the graph of f.

Even: If f(−x) = +f(x) for each {x, −x} ⊆ X. Odd: If f(−x) = −f(x) for each {x, −x} ⊆ X.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 13 / 147 Interlude: Domains Matter

Example Is the statement

+1 = √+1 √+1 = √+1 +1 = ( 1) ( 1) = √ 1 √ 1 = 1 × × − × − − × − − correct? p If yes, then +1 = 1 −

If no, where’s the flaw?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 14 / 147 The Principle of Mathematical Induction

Theorem (Mathematical Induction) Let P (n) be a statement about the natural number n. If P (1) is true, and whenever P (k) is true, then P (k + 1) must also be true, then P (n) is true for all n N. ∈

Theorem (Strong Induction) Let P (n) be a statement about the natural number n. If P (1) is true, and whenever P (k),P (k 1),P (k 2), ..., P (2), and P (1) are true, then P (k + 1) must also be− true, − then P (n) is true for all n N. ∈

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 15 / 147 Summations and Induction, I

n n(n+1) Example ( k=1 k = 2 by Picture) P

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 16 / 147 Summations and Induction, II

n n(n+1) Example (Demonstrate k=1 k = 2 by Induction) Let P (n) be the proposition that Pn k is equal to n(n+1) . Pk=1 2 P1 1(1+1) Basis: Then P (1), which is k=1 k = 2 , is clearly true. Induction: Show that if P (n) is true, then P (n + 1) is true. Assume P (n) is true and add (n + 1) to both sides; i.e., n X n(n + 1) (n + 1) + k = (n + 1) + . 2 k=1 Pn+1 1 Combine terms to see k=1 k = 2 ((2n + 2) + n(n + 1)) . Simplify: n+1 X (n + 1)(n + 2) k = 2 k=1 which shows that P (n + 1) is true.

By the Principle of Mathematical Induction, the result holds.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 17 / 147 Summations and Induction, III

Exercise Illustrate the statement with a picture and prove it using induction. n 1 2k = n2 + n kX=1 n 2 2k 1 = ? (HINT:ASTACK OF TRIANGLES) − kX=1 n n(n + 1)(2n + 1) 3 k2 = 6 kX=1 4 2n < n! for n = 4, 5, 6,...

5 2 (n2 n) for n N | − ∈

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 18 / 147 Polya´ and Induction

George Polya´ constructed the example: Theorem All horses are of the same color.

Proof. Basis. If there’s only one horse, there’s only one color. Induction. Suppose within a set of n horses, there is only one color. Look at any set of n + 1 horses. Number them: 1, 2, 3, . . . , n, n + 1. Consider the sets 1, 2, . . . , n and 2, 3, . . . , n + 1 . Each is a set of only n horses, therefore within{ each} there{ is only one color.} But the two sets overlap, so there is only one color among all n + 1 horses.

Homework: Find the flaw.

n = 1 n = 2 n = ...

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 19 / 147 Important Inequalities ( 1.1) §

Theorem

Let a, b, ai, and bi all be real numbers and n ∈ N. Then 1 |a| ≤ b if and only if (iff) −b ≤ a ≤ b 2 |a + b| ≤ |a| + |b| Triangle inequality

" n #2 " n # " n # 3 X X 2 X 2 aibi ≤ ai · bi Cauchy-Bunyakovsky-Schwarz ineq. i=1 i=1 i=1 v v v u n u n u n uX 2 uX 2 uX 2 4 t (ai + bi) ≤ t (ai) + t (bi) Minkowski inequality i=1 i=1 i=1

4I. 4II. −|a| ≤ a ≤ |a| and − |b| ≤ b ≤ |b| 0 ≤ (a + b)2 = a2 + 2ab + b2 −|a| − |b| ≤ a + b ≤ |a| + |b| ≤ |a|2 + 2|a| |b| + |b|2 = (|a| + |b|)2 −(|a| + |b|) ≤ a + b ≤ (|a| + |b|) (a + b)2 ≤ (|a| + |b|)2

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 20 / 147 Compare: Definitions of Limit ( 1.2) § “Ghosts of departed Quantities” Definition (First Limit (d’Alembert)) One quantity is the limit of another if the second can approach the first nearer than by a given quantity, so that the difference between them is absolutely inassignable. from Encyclopedie´ (1754).

Definition (Intuitive Limit (Cauchy)) When the successive values attributed to a variable approach indefinitely a fixed value so as to end by differing from it by as little as one wishes, this last is called the limit of all the others. from Cours d’analyse (1823).

Definition (Formal Limit (Weierstrass and Bolzano))

A function f(x) has a limit L at x = x0 if for any positive number , there exists a δ such that |f(x) − L| <  for all x in the deleted interval 0 < |x − x0| < δ.

The notation “lim” was introduced by S.-A.-J. L’Huilier (1784) and “ lim ” is due to K. T. W. Weierstrass (1854). x→a

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 21 / 147 Limit Proofs, I

Example (-δ Proof) Prove that lim 2x + 3 = 7 x 2 → Proof. Let  > 0. Then we need to find a δ > 0 so that

f(x) L = (2x + 3) (7) <  | − | | − | 2x 4 <  | − | 2 x 2 <  | − | Choosing δ > 0 so that δ < /2 yields that if 0 < x 2 < δ, then it must follow that f(x) L = (2x + 3) (7) < . | − | | − | | − |

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 22 / 147 Limit Proofs, II

Example (-δ Proof) Prove: lim 4x2 1 = 35 x 3 → − Proof. Let  > 0. We need to find a δ > 0 so that for 0 < x 3 < δ, then | − | f(x) L = (4x2 1) (35) <  | − | − − 2 4x 36 = 2x + 6 2x 6 <  | − | | | · | − | (4 x + 3 ) x 3 <  | | · | − | Assume δ < 1. Then 1 < x 3 < 1 implies that 5 < x + 3 < 7, so that 20 < 4(x + 3) < 28. Choosing− −δ > 0 to be less than the minimum of /28 and 1 yields that if 0 < x 3 < δ, then it must follow that f(x) L = (4x|2 −1)| (35) < . | − | − −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 23 / 147 Limit Proofs, III

“For every  > 0 there is a δ > 0 such that for all x, we see P (x) is true” negated becomes “There is an  > 0 such that for every δ > 0 there is an xˆ with P (ˆx) false”

Example (-δ “Non-Proof”) x Demonstrate that lim | | does not exist. x 0 → x Proof. Suppose the limit is L. Let  = 1 and let δ be any positive number. Choose any x (0, δ). Then x /x = 1, so that f(x) L = 1 L . Choose any p ∈ | p| p | − | | − | xn ( δ, 0). Then xn /xn = 1, so that f(x) L = 1 L . We have that ∈ −  <| 1 | L < − 1 < 1| L <−1 | |0 −< L− < |2 and − − ⇒ − − ⇒  < 1 L <  1 < 1 L < 1 2 < L < 0 which is a contradiction.− − − Therefore⇒ − there− is− no limit L.⇒ −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 24 / 147 Limit Proofs, IV

Exercise Find the value and prove it correct for: 1 lim 4x 1 = x 3 → − 2 lim 3x + 5 = x 2 → − 3 lim 4 x2 = x 1 → − 4 lim 2x3 + 1 = x 0 → 3 5 lim 3x + x + 1 = (STUCK?LOOK AT APROOF .) x 1 →− 6 Why won’t this approach work for lim ln(2x 1) ? x 1 → − 7 Prove that lim 1 doesn’t exist. x 0 x →

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 25 / 147 Squeeze Theorem

Theorem (Squeeze Theorem or Sandwich Theorem) Suppose that m(x) f(x) M(x) on a deleted neighborhood 2of a and that ≤ ≤ lim m(x) = L = lim M(x). x a x a → → Then lim f(x) = L. x a →

1 Apply the theorem to f(x) = x2 sin(1/x) to determine a value for f(0) that makes f continuous everywhere. sin(x) 2 State an analogue of the theorem to use to determine lim . x →∞ x

2A deleted neighborhood of a is I0 = (a − δ, a + δ) − {a} for some δ > 0.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 26 / 147 Compare: Definitions of Continuity

Definition (Intuitive Continuity) The function f is continuous at x = a if and only if we can make f(x) arbitrarily close to f(a) whenever x is sufficiently close to a.

Definition (Formal Continuity—calculus level) Let f be defined on an open interval containing a. Then f is continuous at x = a if and only if for every  > 0 thereI is a corresponding δ > 0 such that whenever x is within δ of a, then f(x) must be within  of L. ∈ I

1 How do these definitions compare to the (calculus level) “limit definition” of continuity: Definition: f is continuous at a if and only if lim f(x) = f(a)? x a →

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 27 / 147 Borzellino’s Functions, I

Example (f(x) = x sin(1/x) ) Does f have a limit atpx = 0? Is f continuous at x = 0?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 28 / 147 Borzellino’s Functions, II

Example (g(x) = (x4 x2)3/2 ) − Does g have a limit at x = 0? Is g continuous at x = 0?

2

1

-2 -1 0 1 2

f(x)=(x4 x2)(3/2)

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 29 / 147 Types of Discontinuity

Definition (Four Principal Types of Discontinuity) Simple or First Kind Removable: The limit lim f(x) exists, but isn’t equal to f(a). x a → Jump: Both lim f(x) and lim f(x) exist, but have different values. x a+ x a → → − Essential or Second Kind Infinite: At least one of lim f(x) or lim f(x) is infinite. x a+ x a → → − Oscillating: At least one of lim f(x) or lim f(x) doesn’t exist, but is x a+ x a bounded. → → −

1 Find and graph an example of each type of discontinuity.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 30 / 147 Continuous and Discontinuous a Lot!

Example (Thomae’s Function3) A function that is continuous at each irrational point and discontinuous at each rational point in [0, 1].

( 1/q if x=p/q, (p, q)=1 D(x) = 0 otherwise

Can a function be discontinuous only on the irrationals?

Compare to Baire’s theorem: Baire 1 functions are continuous except on a first category set.

3 Also called Riemann’s function; Dirichlet’s function is χQ.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 31 / 147 Continuity Theorems

Theorem (Composition of Continuous Functions) The composition of continuous functions is a continuous function.

Theorem (Intermediate Value Theorem) Let f be continuous on the closed interval [a, b] and let yˆ be between f(a) and f(b). Then there is at least one xˆ [a, b] for which f(ˆx) =y. ˆ ∈ Bolzano’s original theorem: if f(a) f(b) < 0, then f must have a root. · Theorem (Extreme Value Theorem) Let f be continuous on the closed interval [a, b]. There are (at least) two values xm and xM in [a, b] such that f(x ) f(x) f(x ) m ≤ ≤ M for all x [a, b]; i.e., f must take on minimum and maximum values. ∈ EVT : Bolzano 1830 (pub 1930), rediscovered: Weierstrass, 1860.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 32 / 147 Homework

1 Problems: 1.1 & 1.2; pg. 36: 1a,§ 2, 4, 6, 11, 13, 17, 18, 21, 23, 24, 27, 29, 30.

2 Read 1.3 §

3 Enjoy the The MacTutor History of Mathematics archive web page for the Mathematicians of the day.

4 Look at the Animated Function Families web page.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 33 / 147 Derivatives ( 1.3) § Definition (The Derivative Function) The derivative of f(x) is given by d f(x + h) f(x) f(x) = f 0(x) = lim − dx h 0 h → if the limit exists.

Which came first differentiation or integration? Proposition (Basic Rules of Differentiation) Results: Reductions: 1 (ur)0 = r ur−1 · u0 1 (k · u)0 = k · (u0) 0 u 2 (u ± v)0 = (u0) ± (v0) 2 (eu)0 = eu · u0, ln(u)0 = u 3 (u · v)0 = (u0) · v + u · (v0) 3 sin(u)0 = + cos(u) · u0, etc.  u 0 (u0) · v − u · (v0) 0 4 = u 2 4 sin−1(u)0 = √ , etc. v v 2 1 − u 5 (u(v))0 = u0(v) · v0

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 34 / 147 Derivative People

Pierre de Fermat (1601-1665) Isaac Barrow (1630-1677)

Fermat: Solved extrema problems using tangents (derivatives) Barrow: “Differential triangle;” Newton attended Barrow’s lectures on geometry and optics

(For fun, look at the MAA’s MathDL“Portrait Gallery”.)

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 35 / 147 The Chain Rule and Inverse Functions

Theorem

If f :[a, b] R is differentiable and f 0 is not zero on [a, b], then, for y = f(x), we have → 1 1 f − 0(y) = f 0(x)

Proof. (Computation sketch.)

1 f f − (y) = y 1 f f− (y) 0 = 1 1 1 f 0 f − (y) f − 0(y) = 1 Chain Rule! · ← 1 f 0(x) f − 0(y) = 1 ·

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 36 / 147 Properties of the Derivative

Theorem If f is differentiable at x = a, then f is continuous at x = a.

Proof. f(x) f(a) lim − lim (x a) = lim [f(x) f(a)] x a x a × x a − x a − → − → →

Theorem (Darboux’s Intermediate Value Theorem)

If f is continuous on [a, b] and differentiable on (a, b), then f 0 has the intermediate value property.

But f 0 isn’t necessarily continuous!

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 37 / 147 Newton’s Method

Example (A Functional Version of Newton’s Method) f(x) Define the function Newton(x) = x . − f 0(x) Maple TI f(x) := ... y1 := ... df := D(f) y2 := (y10) Newton(x) := x - f(x)/df(x) y3 := x - y1/y2

Give an initial value then iterate: Maple TI 1.0 1.0 Newton(%) y3(ans) Newton(%) y3(ans) Et cetera.

1 Find all positive roots of f(x) = x7 1.4995x + 0.994. (Maple file.) −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 38 / 147 A Group Project

Project (The function M(x) = x f 0(x)/f 00(x)) − Determine what iterating f (x) M(x) = x 0 − f (x) does. 00 1 Test the function f(x) = x4 8x3 + 19x2 12x with different starting − − values for x0. 1 x0 = 1.0 2 x0 = 2.0 3 x0 = 3.0 2 Do the same with g(x) = x + 2 sin(x). 3 What is your conjecture? Can you prove it?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 39 / 147 The Mean Value Theorem and . . .

Theorem (The Mean Value Theorem; Cauchy – 1823) Let f be differentiable on (a, b) and continuous on [a, b]. Then there is a value f(b) f(a) c (a, b) so that f 0(c) = − . ∈ b a − Somewhere the instantaneous rate of change = the average rate of change.

Theorem (The “Racetrack Principle;” Davis, Porta, Uhl – 1994)

Suppose f(a) = g(a) and f 0(x) g0(x) for all x a. Then f(x) g(x) for all x a. ≤ ≥ ≤ ≥

Theorem (The “Speed Limit Law;” Ostebee and Zorn – 1997 )

Let f be differentiable on [a, b] and M R. If f 0(x) M for all x [a, b], then f(x) f(a) M (x a). ∈ ≤ ∈ − ≤ · −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 40 / 147 Derivative Tests for Extrema

Proposition (First Derivative Test) Suppose f is differentiable on an interval containing c, a critical point. If the derivative f 0 changes sign from + below c to above c, then f(c) is a local maximum, − from below c to + above c, then f(c) is a local minimum. − If f 0 doesn’t change sign, then f(c) is a terrace point, not an extrema.

Proposition (Second Derivative Test) Suppose f is twice differentiable on an interval containing c, a critical point.

If f 00(c) < 0, then f(c) is a local maximum,

If f 00(c) > 0, then f(c) is a local minimum.

If f 00(c) = 0, the test fails to classify the point.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 41 / 147 Homework

1 Problems: 1.3; pg. 36: 29,§ 34, 40b, 41, 44, 45, 46; Look at 39.

2 Read 1.4 §

3 Investigate the MAA Mathematical Sciences Digital Library

4 Explore the Mathworld website.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 42 / 147 Development of the Integral ( 1.4) § Hippocrates (430 BC): Quadrature of the Lune

Eudoxus (360 BC): Method of Exhaustion

Archimedes (250 BC): Vol(sphere) = 2 Vol(cylinder); parabolic area 3 ·

a an+1 Cavalieri (1635): xn dx = — without modern notation n + 1 Z0

a a(p/q)+1 Fermat (1679): xp/q dx = — without modern notation (p/q) + 1 Z0

— Cavalieri’s notation: omn. l. for omnes linea = “sum of the lines”

— Newton’s notation: x˙ & x Leibniz’s notation: dx & omn. (1686: dx) · R Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 43 / 147 The Fundamental Theorem, I

Isaac Newton (1642-1723) Gottfried Leibniz (1646-1716)

Developed the Fundamental Theorem independently Used either antiderivatives or geometry to calculate integrals Newton did not use transcendental functions, preferring infinite series; Leibniz avoided series

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 44 / 147 The Fundamental Theorem, II

Theorem (Fundamental Theorem, Version 1) Suppose that f is integrable 4on [a, b] and, for x [a, b], define F (x) by ∈ x F (x) = f(t) dt Za 5 Then F is continuous and if f is continuous at c, then F 0(c) = f(c).

Theorem (Fundamental Theorem, Version 2)

Let f be continuous on [a, b] with F 0(x) = f(x). Then

b f(x) dx = F (b) F (a) − Za

4Leibniz invented the R dx notation (1665); Fourier added the limits R b dx. (1822) · a · 5Watch out for singular functions! Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 45 / 147 Definite Integral Definitions

b n Cauchy (1823) f(x) dx = lim f(xn 1) (xn xn 1) n − · − − Za k=1 Theorem: Continuous ⇒Xintegrable

b n Riemann (1854) f(x) dx = lim f(cn) (xn xn 1) n · − − Za k=1 Theorem: Continuous exceptX on a “null set” ⇒ integrable

b n Stieltjes (1894) f(x) dg(x) = lim f(cn) (g(xn) g(xn 1)) n · − − Za k=1 Theorem: Continuous and gXmonotone ⇒ integrable

b 1 Lebesgue (1902) f(x) dµ(x) = lim f(cn) “length” f − ([yk 1, yk]) · − Za k Theorem: Bounded and “measurable”X ⇒ integrable 

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 46 / 147 Integral People

Augustin-L. Cauchy G. F. Bernhard Riemann Thomas J. Stieltjes Henri L. Lebesgue 1789-1857 1826-1866 1856-1894 1875-1941

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 47 / 147 Riemann Sums

Definition () Let f be a bounded function on [a, b]. Let the partition be = x = a < x < x < < x = b and = t beP a collection of points P { 0 1 2 ··· n } T { i} where ti [xi 1, xi] for each i = 1..n. The Riemann sum is ∈ − n (f, , ) = f(tk) (xk xk 1) R P T · − − kX=1

Definition (Riemann-Stieltjes Sum) Let f be bounded and g be increasing on [a, b]. Let the partition be = x = a < x < x < < x = b and = t be a collectionP of points P { 0 1 2 ··· n } T { i} where ti [xi 1, xi] for each i = 1..n. The Riemann-Stieltjes sum is ∈ − n (f, g, , ) = f(tk) [g(xk) g(xk 1)] RS P T · − − kX=1

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 48 / 147 Riemann’s Function (x) R From Riemann’s dissertation:

Define 0.75

0.5 ∞ X ((kx))

R(x) = 0.25 k2 k=1

0 0.25 0.5 0.75 1 1.25 1.5 where6  1 -0.25 x − bxc bxc ≤ x < bxc + 2  1 ((x)) = -0.5 0 x = bxc + 2  1 x − bxc − 1 bxc + 2 ≤ x < bxc + 1

-0.75

R(x) is discontinuous at all rationals x = p/q where q is even Z 1 R(x) is integrable What do you conjecture R(x) =? 0

6 Note: alternately ((x)) = (x − 1) − bx − 1/2c for x 6= bxc + 1/2. Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 49 / 147 A Function Defined by an Integral

Example +1 x < 1.5 x Define: f(x) = and F (x) = f(t) dt. 1 otherwise (− ) Z0

2

Query: Where is not F 1 differentiable?

-1 0 1 2 3

-1

-2

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 50 / 147 Functions Defined by Integrals

Many important functions that have no elementary expressions are defined by integrals. Definition (Several Special Functions) x 1 xn+1 1 ln(x) = dt Look at lim − n 1 n+1 1 t →− Z x 2 t2 erf(x) = e− dt,“error function” √π Z0 ∞ x 1 t Γ(x) = t − e− dt, Generalized factorial from Γ(n + 1) = n! Z0 x sin(t) Si(x) = dt,“sine integral”(Engineering) 0 t Z x (x) = sin π t2 dt,“Fresnel sine integral”(Transition curves) Fs 2 Z0  and hundreds more (A& S Handbook) . . . See Maple’s Special Functions Explorer

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 51 / 147 Which Witch is Which?

Example

3

2

1

-3 -2 -1 0 1 2 3

-1

-2

-3

The graph shows f(x), f 0(x), and F (x) = f(x) dx. Which is which? Z

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 52 / 147 A Special Integral

∞ x2 The integral e− dx is important in analysis and probability, but has no elementary antiderivativeZ−∞ — the Fundamental Theorem does not apply. Instead, consider

2 ∞ x2 ∞ y2 ∞ ∞ (x2+y2) = e− dx e− dy = e− dx dy. I × Z0 Z0 Z0 Z0 Change to polar coordinates with

[x, y] [r, θ] = x2 + y2 , arctan(y/x) and dx dy r dr dθ. 7→ 7→ hp i The transformed integral is π/2 2 ∞ r2 = e− r dr dθ, I Z0 Z0 which is not hard to evaluate. Homework!

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 53 / 147 A Mean Value Theorem for Integrals

Definition (Average Value) Let f be an integrable function on [a, b]. The average value f¯ is 1 Z b f¯ = f(t) dt b − a a

Theorem (Integral Mean Value Theorem) Let f be a continuous on [a, b]. There is a point c ∈ (a, b) where f(c) = f.¯

Proof.

1 f is continuous on [a, b], so f has a max M and a min m by the EVT b 2 R f is cont. so integrable. Then m(b − a) ≤ a f(x) dx ≤ M(b − a) b 3 1 R By the IVT, there is a pt. c ∈ (a, b) so that f(c) = b−a a f(x) dx

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 54 / 147 Homework

1 Problems: 1.4; pg. 36: 48,§ 49, 53, 55, 56.

2 Read 1.5, 1.6 §

3 Investigate the Earliest Known Uses of Some of the Words of Mathematics website

4 Explore the Earliest Uses of Symbols of Calculus website.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 55 / 147 Sequential Limits ( 1.5) §

Definition

Let an be a sequence. Then lim an = L if and only if for every  > 0 there n { } →∞ is an N N such that whenever n > N, then an L < . ∈ | − | Examples ( 1)n n + 1 Let a = − Let dn = n n + 1 2n 1 2 − Let bn = √4n n + 3 2n Let en = cos(nπ/2) − − n Let cn = sin(2nπ) Let fn = n

Theorem (Bolzano-Weierstrass Thm for Sequences7) A bounded sequence has a convergent subsequence.

7See Moore’s “Historians and Philosophers of Logic: Are They Compatible? The Bolzano-Weierstrass Theorem as a Case Study,” History And Philosophy Of Logic 20 (2000), 169-180. Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 56 / 147 Series Limits

Definition

∞ Let an be a series. Then an = L if and only if for every  > 0 there is an k=0 P X n N N such that whenever n > N, then an L < . ∈ − k=0 X

Examples

∞ 1 ∞ 1 rk = when r < 1 = e 1 r | | k! kX=0 − kX=0 ∞ 1 ∞ ( 1)n = − = ln( 1 ) n ∞ n 2 n=1 n=1 X X ∞ 1 π2 ∞ 1 = = 1 j2 6 j (j + 1) j=1 j=1 X X ·

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 57 / 147 Guido Grandi and Summing Series Divinely

Guido Grandi (1671-1742) discussed infinite series. In his book Quadratura circula et hyperbolæper infinitas hyperbolas geometrice exhibita (1703), he computed 0 = (1 + 1) + (1 + 1) + (1 + 1) + (1 + 1) + ··· 1 = 1 + ( 1 + 1) + ( 1 + 1) + ( 1 + 1) + and wrote: ··· By putting parentheses into the expression 1 1 + 1 1 + in different ways, I can, if I want, obtain 0 or 1. But− then− the idea··· of the creation ex nihilo is perfectly plausible.8 1 Although, he thought the true value of the series was 2 from setting x = 1 in 1 2 3 1+x = 1 x + x x in the 2nd edition of Quadratura. Euler (1707-1783) agreed. − − ···

8“Sed inquies: aggregatum ex infinitis differentiis infinitarum ipsi DV æqualium, sive continue,` sive alterne` sumptarum, est demum summa ex infinitis nullitatibus, seu 0, quomodo ergo quantitatem notabilem aggreget? At repono, eam Infiniti vim agnoscendam, ut etiam quod per se nullum est multiplicando, in aliquid commutet, sicuti finitam magnitudine´ dividendo, in nullam degenerare cogit: unde per infinitam Dei Creatoris potentiam omnia ex nihlo facta, omniaque in nihilum redigi posse: neque ade absurdum esse, quantitatem aliquam, ut ita dicam, creari per infinitam vel multiplicationem, vel additionem ipsius nihili, aut quodvis quantum infinita divisione, aut subductione in nihilum redigit.”

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 58 / 147 Convergence Tests for Series

Theorem

Ratio Test: [D’Alembert (1761)] Let an be a positive term series and set a r = lim n+1 . Then n an P 1 if 0 r < 1, the series converges. ≤ 2 if 1 < r , the series diverges. ≤ ∞ 3 if r = 1, the test fails.

Root Test: [Cauchy (1823)] Let an be a positive term series and set n ρ = lim √an. Then n P 1 if 0 ρ < 1, the series converges. ≤ 2 if 1 < ρ , the series diverges. ≤ ∞ 3 if ρ = 1, the test fails.

∞ ( 1)n n The root test is stronger than the ratio test. Check: 2 − − n=1 X

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 59 / 147 Very Useful Tests

Theorem

Limit Comparison: Let an and bn be positive series. Set L = lim an/bn. Then P P 1 if L = 0 and bn converges, then an converges. 2 if 0 < L < P, the two series eitherP converge or diverge together. ∞

3 if L = and b diverges, then a diverges. ∞ n n P P Integral Test: Let an = f(n) be non-negative and monotone-decreasing. Then ∞ an and f(x) dx converge or diverge together. Zk (AlsoP called the MaclaurinCauchy test.)

For more tests, visit Mathworld’s Convergence Tests page.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 60 / 147 Group Project

Project Test each series for convergence.

∞ 2 + n3 ∞ n 1 5 ln 1 + 2n3 3n + 1 n=0 n=1 X X   ∞ n ∞ 22n+1 2 6 ln(n) 5n n=2 n=0 X X ∞ ( 10)n ∞ √n + 1 √n 3 − 7 − n! n n=0 n=1 X X n n ∞ n 1 3 ∞ ( 1) x 4 ( 1) − n− 8 − − 22n n! n=1 n=0 X X

Look at the Calculus II “Convergence Handout.”

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 61 / 147 Taylor Polynomials and Series

Definition (Taylor Polynomial for f) Let f have n continuous derivatives. The Taylor polynomial for f of degree n is (n) f 00(a) 2 f (a) n T (x) = f(a) + f 0(a)(x a) + (x a) + + (x a) n − 2! − ··· n! − (n+1) Set Mn+1 = max f (t) on the interval [a, b]. Then the maximum error in approximating f by T for x [a, b] is bounded by n ∈ M Err n+1 (x a)n+1 n ≤ (n + 1)! −

Definition (Taylor Series for f) Let f have derivatives of all orders. The Taylor series for f centered at x = a is ∞ f (k)(a) T (x) = (x a)k k! − k=0 for x a < R where R 0 is theXradius of convergence. | − | ≥

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 62 / 147 Taylor Comparison

Example

sin(x) & T5, T15, and T30. tan(sin(x)) & T5, T15, and T30

The Taylor series for sine works well in contrast to the series for tan ◦ sin .

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 63 / 147 Complex Plots Show the Answer

Example

z = sin(x + iy) z = tan(sin(x + iy)) No| poles| | Lots of poles |

The radius of convergence of a Taylor series is the distance from its center to the nearest pole. Poles may lie in the plane, off the real axis.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 64 / 147 Homework

1 Problems: 1.5, 1.6; pg. 36: 36,§ 60, 61, 63, 66, 70, 72. Ponder the comment following #72

2 Read 2.1, 2.2 §

3 Investigate the Free Software Foundation’s“Directory of Math Software”

4 Explore uses of GeoGebra (free) software for Linux, Mac, and Windows.

5 Explore uses of Xcas (free) computer algebra system (CAS) for Linux, Mac, iPad, and Windows.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 65 / 147 Interior, Exterior, Boundary, & Limit Points ( 2.1) §

Definition Neighborhood: A (basic) neighborhood N(x) of x ∈ R is an open interval containing x; i.e., N(x) = (x − δL, x + δR). Deleted N’hood: A neighborhood N(x) minus x; i.e., N 0(x) = N(x) − {x}. Interior: The interior of a set is int(S) = {x | N(x) ⊆ S for some n’hood N(x)}. Exterior: The exterior of a set is ext(S) = {x | N(x) ⊆ Sc for some n’hood N(x)}. Boundary: The boundary of a set is  c ∂(S) = int(S) ∪ ext(S) . Limit Point: The point x is a limit point or an accumulation point of S iff every neighborhood N(x) contains a point of S different from x; i.e., S ∩ N 0(x) 6= ∅.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 66 / 147 Interior/Exterior in R2

Example

Exterior

Interior

Boundary

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 67 / 147 Interior/Exterior Example

Example Identify the points A and B as interior, exterior, or boundary points.

A B

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 68 / 147

A B

A B

(5) (4)

(9)

(8)

A B

(5) (4) Homework

1 Problems: 2.1. pg. 117, No§ 2.1, 2.2, 2.4; LA 2.8.

2 Read 2.2 §

3 Investigate the topic dense set.

4 Explore the taxi cab metric.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 69 / 147 Limit and Continuity Definitions ( 2.2) § Definition (Limit) Let f : D R and let a be an accumulation point of Dlet a be an accumulation→ point of D. Then lim f(x) = L if and only if for every  > 0 there x a is a δ > 0 such that if 0 < x a →< δ, then f(x) L < . | − | | − |

Definition (Continuity) Let f : D R and let a Dlet a D. Then f is continuous at x = a if and only if for→ every  > 0 there∈ is a δ∈ > 0 such that if x a < δ, then f(x) f(a) < . | − | | − |(∀ > 0) (∃δ > 0) (∀x ∈ D)[d(x, a) < δ =⇒ d(f(x), f(a)) < ].

Definition (Continuity — General Topological Version) Let f : D R and let a D. Then f is continuous at x = a iff for every open → ∈ 1 set R containing f(a), the inverse image f − ( ) is an open set in D. O ⊆ O

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 70 / 147 Limit and Continuity Example

Example

3 Consider

√x2 1 1 x 2 f(x) = − ≤ | | (0 x = 0

1

1 What is the lim f(x) ? x 0 → The limit does not exist as 0 is not -2 -1 0 1 2 an accumulation point of dom(f).

2 Is f continuous at x = 0? -1 Yes!

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 71 / 147 Uniform Continuity

Definition A function f is uniformly continuous on a set S iff for any  > 0, there is a δ > 0 such that for any x1, x2 S with x1 x2 < δ, we have f(x ) f(x ) < . ∈ | − | | 1 − 2 | A function f is not uniformly continuous on a set S iff there is an  > 0 for which any δ > 0 has points x1, x2 S with x1 x2 < δ, but f(x ) f(x ) . ∈ | − | | 1 − 2 | ≥ Theorem If f is continuous on a compact set, then f is uniformly continuous on it. If f is uniformly continuous on (a, b), then f can be extended to be (uniformly) continuous on [a, b].

If f 0 is bounded on (a, b), then f is uniformly continuous on (a, b).

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 72 / 147 Homework

1 Problems: 2.2. pg. 117, No§ 2.9, 2.11, 2.13, 2.18, 2.22.

2 Read 2.4 §

3 Investigate the topic “limit superior and limit inferior.”

4 Explore the Conway base 13 function and the IVP.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 73 / 147 Derivatives ( 2.3) § Definition (Derivative) Suppose f :D R and a D is an accumulation point of D. Then → ∈ f(x) f(a) f 0(a) = lim − x a x a → − (1) if the limit exists. Set f (x) = f 0(x), and for n > 1, define

(n) (n 1) 0 f (x) = f − (x) .   Example (How Many Derivatives Does It Have?)

For each N N, define fn by ∈ xn x 0 fn(x) = ≥ (0 otherwise

(n 1) (n) (n+1) What is fn − (0)? fn (0)? fn (0)?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 74 / 147 Cauchy’s MVT

Theorem (Cauchy’s Mean Value Theorem)

Suppose that f, g :[a, b] R are both continuous on [a, b] and→ differentiable on (a, b). Then there is a c (a, b) where ∈ f 0(c) [g(b) g(a)] = g0(c) [f(b) f(a)]. · − · − If denominators aren’t zero, then f 0(c) f(b) f(a) = − g (c) g(b) g(a) 0 −

Theorem (Darboux’s IVT for Derivatives) Derivatives have the Intermediate Value Property.

Corollary Derivatives can only have discontinuities of the second kind.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 75 / 147 Proving Darboux

Proof. (Darboux’s IVT for Derivatives.

Assume ‘wolog’ that f 0(a) < k < f 0(b).

1 Define g(x) = f(x) kx. Then g0(x) = f 0(x) k. So g0(a) < 0 < g0(b). − − 2 g is continuous, so it has a minimum m. Let g(c) = m. 3 If c = a, then g(x) g(a) g(x) m − = − 0 = g0(a) > 0 x a x a ≥ ⇒ →← − − 4 If c = b, then g(b) g(x) m g(x) − = − 0 = g0(a) < 0 b x b x ≤ ⇒ →← − − 5 Whence c (a, b). Therefore, since c is a minimum, g0(c) = 0. I.e., ∈ f 0(c) = k.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 76 / 147 Reconciliation Bill

Reconcile This with Darboux! x x 0 Define f(x) = ≥ . (0 otherwise

1 x > 0 0 Whence f 0(x) = . (NB: f (0) DNE) (0 x < 0

1 1

-2 -1 0 1 2 -2 -1 0 1 2

-1 -1

Darboux’s hypotheses: f is continuous on [a, b] and differentiable on [a+, b ]. − Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 77 / 147 Tailored Ends

Theorem (Remainders for Taylor’s Theorem) For f “suitably nice,”

n f(x) = f (k)(c)(x c)k + R (x) − n kX=0 where: f (n+1)(c) Lagrange Form: R (x) = (x a)n+1 n (n + 1)! · −

f (n+1)(c) Cauchy Form: R (x) = (x c)n(x a) n n! · − − 1 x Integral Form: R (x) = (x t)n f (n+1)(t) dt n n! − · Za for some c between a and x.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 78 / 147 Homework

1 Problems: 2.3 pg. 117. # 23,§ 24, 26, 29, 34, 35, 36

2 Read 2.4 §

3 Investigate singular functions

4 Explore the CORDIC algorithm

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 79 / 147 Riemann Integrals ( 2.4) §

Definitions

Partition: A set of n + 1 points where a = x0, b = xn, and P = x0 < x1 < x2 < < xn . Set k = [xk 1, xk]. { ··· } I − Tagged Partition: A partition P along with a set of tags T = tk tk [xk 1, xk] for k = 1..n denoted by PT − { | ∈ n }

Riemann Sum: RS(PT , f) = f(tk)∆xk k=1 X n

Darboux Upper Sum: U(P, f) = Mk ∆xk where Mk = sup f(x) x k k=1 ∈I Xn Darboux Lower Sum: L(P, f) = mk ∆xk where mk = inf f(x) x k ∈I kX=1

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 80 / 147 Riemann-Darboux Integrals

Definitions b Upper Integral: f(x) dx = inf U(P, f) (Recall: infimum = g.l.b) a P R b Lower Integral: a f(x) dx = sup L(P, f) (Recall: supremum = l.u.b.) P Riemann-DarbouxR Integral: A function bounded on [a, b] is integrable iff b b b a f(x) dx = a f(x) dx. The common value is a f(x) dx R R R Proposition For any partition P , we see L(P, f) b f(x) dx b f(x) dx U(P, f) ≤ a ≤ a ≤ R R Theorem Suppose f is bounded on [a, b]. Then f is integrable iff for any  > 0 there is a partition P ∗ such that U(P ∗, f) L(P ∗, f) < . −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 81 / 147 Integral Examples

Example Let D(x) be Dirichlet’s function. Then 1 1 D(x) dx = 0 = 1 = D(x) dx, 6 Z0 Z0 1 therefore D(x) dx DNE. Z0

Example Let T (x) be Thomae’s function. Then 1 1 T (x) dx = 0 = T (x) dx, Z0 Z0 1 therefore T (x) dx = 0. Z0

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 82 / 147 Integral Theorems

Theorem (Continuity Implies Integrability) If f is continuous on [a, b], then f is integrable on [a, b].

Proof. U(P ) L(P ) = (M m )∆x = [f(x ) f(x )]∆x • − k − k k k,M − k,m k f(xk,M ) f(xk,m) /(b a) and ∆xk = b a • − P ≤ − P − P Theorem (Monotonicity Implies Integrability) If f is monotone on [a, b], then f is integrable on [a, b].

Proof.

U(P ) L(P ) = (Mk mk)∆xk = [f(xk) f(xk 1)]∆xk • − − − − with ∆xk = (b a)/n and [f(xk) f(xk 1)] = f(b) f(a) • −P −P − − P

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 83 / 147 A Series of Steps

Example

∞ 1 1 S(x)= U x k2 · − k k￿=1 ￿ ￿

1 Is S integrable on [0, 1]? Yes! S(x) dx 0.44 (See a Maple worksheet) 0 ≈ Wm C Bauldry ([email protected]) MATR 5930. Analysis for Teachers. Fall ’13 84 / 147 Integral Mean Value Theorem

Theorem (Bonnet’s Mean Value Theorem) Suppose f is continuous and g is nonnegative and integrable on [a, b]. Then there is a value c (a, b) such that ∈ b b f(x)g(x) dx = f(c) g(x) dx Za Za Proof. f is continuous on a closed interval = f has max & min values ⇒ m g f g M g, then divide by g (nonzero else done) · ≤ ≤ · IVT = there is a c s.t. f(c) = ( f g)/( g) R ⇒ R R R R R Very useful inequality (from the proof): b b b m g(x) dx f(x)g(x) dx M g(x) dx · ≤ ≤ · Za Za Za

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 85 / 147 Fundamental Theorem of Calculus

Theorem (Fundamental Theorem of Calculus)

If F is differentiable on [a, b] and F 0 = f is integrable on [a, b], then b f(x) dx = F (b) F (a). − Za Proof. Let ε > 0. n n n X X 0 X 1 F (b) − F (a) = F (xk) − F (xk−1) = F (tk)∆xk = f(tk)∆xk for all P k=1 k=1 k=1 n X Z b 2 f integrable =⇒ there is a P s.t. f(tk)∆xk − f(x) dx < ε k=1 a Z b Z b

3 Thus [F (b) − F (a)] − f(x) dx < ε. So f(x) dx = F (b) − F (a) a a

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 86 / 147 Fundamental Example

Example Problem. Suppose that f(x) 0 for all x [a, b], ≥ ∈ f is continuous on [a, b], and b a f(x) dx = 0. ThenRf(x) = 0. Solution. Assume that at some c [a, b] we have f(c) > 0. ∈ 1 Since f is contin., there is a neighborhood Nδ(c) where f(x) 2 f(c) > 0 for all x N (c). ≥ ∈ δ Every partition P has to include c in some interval k = [xk 1, xk]. Refine P until N (c). I − Ik ⊂ δ Then L(P, f) 1 f(c) δ > 0. ≥ 2 Hence b f(x) dx 1 f(c) δ > 0 which is a contradiction. a ≥ 2 R

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 87 / 147 A Brief Integral Timeline

Development of the Integral c. 400 BC Eudoxus Method of Exhaustion c. 250 BC Archimedes Quadrature of the Lune 1635 Cavalieri Geometria indivisibilibus continuorum nova quadam ratione promota 1669 Barrow Lectiones Opticæ et Geometricæ 1687 Newton Philosophi Naturalis Principia Mathematica 1684 Leibniz Nova methodus pro maximis et minimis, itemque tangen- tibus 1696 L’Hopitalˆ Analyse des inifiniment petits 1823 Cauchy Cours d’Analyse 1854 Riemann Uber¨ die Darstellbarkeit einer Function durch eine trigonometrische Reihe 1870 Darboux Sur les equations´ aux deriv´ ees´ du second ordre 1894 Stieltjes Recherches sur les fractions continues 1902 Lebesgue Integrale,´ longueur, aire

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 88 / 147 Homework

1 Problems: 2.4 pg. 117. # 39,§ 41, LA 42

2 Read 2.5 §

3 Investigate Frank Burk’s A Garden of Integrals. Especially, read Chapter 1, “An Historical Overview,” pp. 1-26.

4 Explore Gauss-Kronrod quadrature on TI calculators.

5 Study Biographies of Women in Mathematics.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 89 / 147 Interlude: Fundamental Theorems

Theorem (The Fundamental Theorem of Arithmetic) Let n be an integer other than 0 or 1. Then there is a unique expansion ± n = sign(n) pe1 pe2 pek · 1 2 ··· k where p1 < p2 < < pk are primes and ei N. ··· ∈ Theorem (The Fundamental Theorem of Algebra) Let p(z) be a non-constant polynomial of degree n. Then there are n numbers zk (possibly complex, possibly repeating) such that

p(z) = a(z z0) (z z1) (z zn 1) − · − ··· − − Theorem (The Fundamental Theorem of Calculus) x Let f be a continuous real function on [a, b]. Then F (x) = f(t) dt is a dZ x differentiable at every point in [a, b], and f(x) = F 0(x) = f(t) dt. dx a Let f be a continuous real function on [a, b] and let F be a functionZ b differentiable on [a, b] such that F 0 = f. Then f(t) dt = F (b) F (a). − Za

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 90 / 147 Sequence & Series ( 2.5) § Theorem The limit of a sequence is unique.

Proof. Suppose there are two limits M and L. Let ε > 0. a L = a L < ε/2 eventually9 n → ⇒ | n − | a M = a M < ε/2 eventually n → ⇒ | n − | L M L a + a M < ε, so L = M | − | ≤ | − n| | n − | Theorem A convergent sequence is bounded.

Proof. a a L + L < ε + L < 1 + L eventually | n| ≤ | n − | | | | | | | M max a1 , a2 ,..., an 1 , 1 + L ≥ {| | | | | − | | |} 9“P (n) is true eventually” means ∃N ∈ N s.t. P (n) is true for all n > N. Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 91 / 147 A Sequence of Results

Theorem A bounded, increasing sequence of reals converges.

Theorem (Bolzano-Weierstrass Theorem) Every bounded, infinite set of reals has an accumulation point.

Definition (Cauchy Sequence)

A real sequence an is Cauchy iff for every  > 0 there is an N N such that if n, m > N, then{a } a < . ∈ | n − m|

Theorem A sequence of real numbers is Cauchy iff it converges.

These theorems all fail if we replace R with Q. Why?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 92 / 147 Upper and Lower Limits

Definition For a sequence a , set { n} lim sup an = inf sup ak n n k n →∞ →∞ ≥

lim inf an = sup inf ak n n k n →∞ →∞ ≥ Let S be the range of a . Then { n} lim supn an is the largest accumulation point of S →∞ lim infn an is the smallest accumulation point of S →∞ Theorem A sequence converges iff lim sup and lim inf are finite and equal.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 93 / 147 Series Convergence Tests, I

Let an and bn be series. P P Theorem (Comparison & Limit Comparison Tests) Suppose 0 < a b . Then n ≤ n if bn converges, so does an if P an diverges, so does Pbn. Suppose a 0 and b > 0. Then P n ≥ n P if lim an/bn = L with 0 < L < , then an and bn converge or diverge together. ∞ P P

Theorem (Integral Test)

If a:[1, ) R is continuous, positive, and decreasing, then an and ∞ → ∞ a(x) dx converge or diverge together. 1 P R

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 94 / 147 Series Convergence Tests, II

P Let an be a series with positive terms. (See Convergence Tests) Theorem (Ratio & Root Tests) √ n Set r = limn→∞ an+1/an. Then Set ρ = limn→∞ an. Then P P if r < 1, an converges, if ρ < 1, an converges, if r = 1, the test fails, if ρ = 1, the test fails, P P if r > 1, an diverges. if ρ > 1, an diverges.

Theorem (Raabe’s Test & Cauchy’s Condensation Test)   an Set R = lim n · − 1 . Then If an is decreasing, then n→∞ an+1 P P if L > 1, then an converges, an converges iff if L = 1, the test fails, n P 2 a n converges. P 2 if L < 1, then an diverges.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 95 / 147 Basic Convergence Test Flow Chart

Choosing a Convergence Test for Infinite Series Courtesy David J. Manuel

Do Series Diverges by the individual No terms approach 0? the Divergence Test.

Yes

Does Do Use Ratio Test the series Yes individual terms Yes alternate have factorials or (Ratio of Consecutive signs? exponentials? Terms) No No

Is Use Alternating individual term Yes Use Integral Test Series Test easy to integrate? (do absolute value of terms go to 0?) No

Do individual terms Use Comparison or involve fractions with Yes Limit Comp. Test powers of n? (Look at Dominating Terms) No

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 96 / 147 Homework

1 Problems: 2.5 pg. 117. # 49,§ 52, 54, 57, 58, 62a,b.

2 Read 2.6 §

3 Investigate the “Interactive Real Analysis” web pages for convergence tests

TM 4 Explore The On-Line Encyclopedia of Integer Sequences

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 97 / 147 Pointwise & Uniform Convergence ( 2.6) § Let fn :D → R for all n ∈ N. Definition P Pointwise convergence: {fn(x)} or n fn(x) converges pointwise on D to f(x) iff for every ε > 0, for every x ∈ D there is an N = N(ε, x) ∈ N such that if n > N, then

X |fn(x) − f(x)| < ε or fn(x) − f(x) < ε n P Uniform convergence: {fn(x)} or n fn(x) converges uniformly on D to f(x) iff for every ε > 0 there is an N = N(ε) ∈ N such that for any x ∈ D if n > N, then

X |fn(x) − f(x)| < ε or fn(x) − f(x) < ε n

PWC: (∀ε > 0) (∀x∈D)(∃N(, x))[ ... ] &. UC: (∀ε > 0) (∃N(ε))(∀x∈D)[... ]

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 98 / 147 A Sequence of Functions

Example

1/n fn(x)=n x 1 − f2 ￿ ￿ f3 f5 f15

f(x)=ln(x)

Is the convergence uniform on [1, 10]? On (0, 10)?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 99 / 147 Uniform Convergence and Continuity

Theorem

If fn is continuous on [a, b] for all n and fn f uniformly, then f is continuous and → lim lim fn(x) = lim lim fn(x). x c n n x c → →∞ →∞ →

Theorem

If fn is continuous on [a, b] for all n and fn converges uniformly, then the sum is continuous and P ∞ ∞ lim fn(x) = lim fn(x). x c x c → n=0 n=0 → X X

1/n x Q Set Υn(x) = ∈ . Is Υn(x) continuous? Is limn Υn(x)? 0 x / Q ( ∈

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 100 / 147 Uniform Convergence and Integration

Theorem

If fn is integrable on [a, b] for all n and fn f uniformly, then f is integrable and b → b lim fn(x)dx = lim fn(x)dx. n n Za →∞ →∞ Za

Theorem

If fn is integrable on [a, b] for all n and fn converges uniformly, then the sum is integrable and b b ∞ P ∞ fn(x)dx = fn(x)dx. a n=0 n=0 a Z X X Z

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 101 / 147 A Good Sequence of Functions

Example

0 x<0 1 1 fn(x)= x x 0 x  · n − ≤ ≤ n 0 1

f1

f2

f3

f4

Find fn, limn fn, limn fn, and limn fn. R R R Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 102 / 147 A Bad Sequence of Functions

Example

0 x<0 f4 2 1 2n x 0 x< 2n fn(x)= ≤ 2n(1 nx) 1 x< 1  2n n  − 1 ≤ 0 n x f3 ≤  

f2

f1

Find fn, limn fn, limn fn, and limn fn. R R R Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 103 / 147 Uniform Convergence and Differentiation

Differentiation does not behave as well as continuity and integration. Example 1 2  Let fn(x) = n sin n x . Show that fn converges uniformly on R 0 fn doesn’t even converge pointwise anywhere.

Theorem

Suppose fn :[a, b] → R. If

1 each fn is continuously differentiable on [a, b],

2 fn → f pointwise on [a, b], and

3 0 {fn} converges uniformly on [a, b], 0 0 d d then fn → f uniformly, fn → f uniformly, and lim fn(x) = lim fn(x). dx n→∞ n→∞ dx

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 104 / 147 The Best Uniform Convergence Test

The Weierstrass M–test provides a very useful method for testing uniform convergence. Theorem (The Weierstrass M–Test)

Let fn be a sequence of functions. If there is a sequence of constants Mn such that

1 fn(x) Mn for all n N and | | ≤ ∈ ∞ 2 Mn converges, n=1 X ∞ then fn converges uniformly (and absolutely). n=1 X 1 1 Does f (x) converge where f (x) = sin n2x ? n n n n2 2 Why doesn’tP the test work for fn0 ? 

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 105 / 147 A Series of Steps

Example

∞ 1 1 S(x)= U x k2 · − k k￿=1 ￿ ￿

Is the convergence uniform?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 106 / 147 Bernstein’s Polynomials

Theorem (Weierstrass)

Let f :[0, 1] → R be continuous. For any  > 0, there is a polynomial Bn(x) such that |f(x) − Bn(x)| <  for all x ∈ [0, 1].

Proof. Set n   ! X k n k n−k B (x) = f · x (1 − x) n n k k=0

B1 = f(0) (1 − x) + f(1) x 2 1 B2 = f(0) (1 − x) + 2f( ) x(1 − x) B5(x) 2 + f(1) x2 B20(x) 3 1 2 B3 = f(0) (1 − x) + 3f( 3 ) x(1 − x) 7 f(x)= x sin π√x − 2 2 2 3 ￿ ￿ + 3f( 3 ) x (1 − x) + f(1) x

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 107 / 147 Powerful Series

Theorem ∞ X k Let R be the radius of convergence of the power series ak(x − c) . Then the power series k=0 1 converges absolutely for all x s.t. |x − c| < R; 2 converges uniformly for all x s.t. |x − c| ≤ r < R; 3 diverges for all x s.t. |x − c| > R.

Example ∞ X nxn Problem: Find the radius of convergence of . 2n+1 n=0

(n+1)|x|n+1 n|x|n |x| Solution: Consider rn := un+1/un = 2n+1 / 2n → 2 as n → ∞. Applying the ratio test gives convergence for |x|/2 ≤ 1 or |x| < 2. So the radius of convergence is R = 2.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 108 / 147 Differentiating and Integrating Power Series

Theorem

∞ n Suppose that f(x) = anx has radius of convergence R > 0. Then n=0 X 1 f is continuous for x < R | |

∞ n 1 2 f is differentiable for x < R and f 0(x) = n a x − | | n n=0 X x ∞ an 3 f is integrable for x < R and f(t) dt = xn+1 | | n + 1 0 n=0 Z X

Theorem Series representation is unique on the interval of convergence.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 109 / 147 Taylor Series

Theorem (Taylor Series) ∞ X n Suppose that f(x) = anx has radius of convergence R > 0. Then f is infinitely n=0 (n) differentiable on |x − a| < R and the coefficients an are given by an = f (a)/n!.

Examples (Important Maclaurin Series)

∞ ∞ 2k+1 1 X k −1 X k x 1 = x for |x| < 1 4 tan (x) = (−1) ; x ∈ R 1 − x 2k + 1 n=0 n=0 ∞ 2k+1 ∞ k X k x x X x 2 sin(x) = (−1) ; x ∈ R 5 e = ; x ∈ R (2k + 1)! k! n=0 n=0 ∞ 2k ∞ k X k x X k−1 x 3 cos(x) = (−1) ; x ∈ R 6 ln(1 + x) = (−1) ; |x| < 1 (2k)! k n=0 n=1 See Wikipedia’s list.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 110 / 147 Maclaurin Series

A Maclaurin series is a Taylor series centered at a = 0. Alternate techniques can be useful for finding series expansions without finding a formula for the nth derivative. Example The series for sec(x) can be found from 1 1 = cos(x) x2 x4 x6 1 − 2! + 4! − 6! ± · · · 1 5 61 = 1 + x2 + x4 + x6 + ··· 2 24 720

To find the inverse of the cosine series, use (see Landau symbols for “O(xn)” )  2 4  x x 6 2 3 4 5  1 = 1 − + + O(x ) × a0 + a1x + a2x + a3x + a4x + O(x ) 2! 4!  a0  2  a1  3  a2 a0  4 5 1 = a0 + a1x + a2 − x + a3 − x + a4 − + x + O(x ) 2 2 2 24 1 5 ⇒ a0 = 1 ⇒ a2 = ⇒ a4 = ⇒ ... ; a1 = 0 ⇒ a3 = 0 ⇒ a5 = 0, &c 2 24

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 111 / 147 Maclaurin Series, II

Example (Differentiation/Integration) The Maclaurin series for ln(1 + x) can be found as follows (subject to conditions): 1 ln(1 + x) = dx 1 + x Z 1 = 1 x + x2 x3 + x4 x5 1 + x − − − ± · · · x2 x3 x4 ln(1 + x) = x + ... − 2 3 − 4 ± We need to investigate what conditions are needed to be able to integrate term by term; i.e., when is ∞ ∞ an(x) dx = an(x) dx ? Z kX=0 kX=0 Z

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 112 / 147 Off On a Tangent

What is the Maclaurin series for f(x) = tan(x)? Using the definition, we need the derivatives of tan(x): tan(0)(x) = tan(x) tan(1)(x) = sec2(x) tan(2)(x) = 2 sec2(x) tan(x) tan(3)(x) = 2 sec4(x) + 4 sec2(x) tan2(x) tan(4)(x) = 16 sec4(x) tan(x) + 8 sec2(x) tan3(x) tan(5)(x) = 16 sec6(x) + 88 sec4(x) tan2(x) + 16 sec2(x) tan4(x) tan(6)(x) = 272 sec6(x) tan(x) + 416 sec4(x) tan3(x) + 32 sec2(x) tan5(x) tan(7)(x) = 272 sec8(x) + 2880 sec6(x) tan2(x) + 1824 sec4(x) tan4(x) + 64 sec2(x) tan6(x) tan(8)(x) = 7936 sec8(x) tan(x) + 24576 sec6(x) tan3(x) + 7680 sec4(x) tan5(x) + 128 sec2(x) tan7(x)

etc. (Volunteers?) 1 3 2 5 17 7 62 9 1382 11 21844 13 tan(x) = x + 3 x + 15 x + 315 x + 2835 x + 155925 x + 6081075 x + ···

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 113 / 147 Patterns

Circular Function Hyperbolic Function

∞ ∞ X x2n+1 X x2n+1 sin(x) = (−1)n sinh(x) = (2n + 1)! (2n + 1)! n=0 n=0 ∞ ∞ X x2n X x2n cos(x) = (−1)n cosh(x) = (2n)! (2n)! n=0 n=0

∞ 2n−1 ∞ 2n−1 X n n n x X n n x tan(x) = B2n(−1) 4 (1−4 ) tanh(x) = B2n4 (1 − 4 ) (2n)! (2n)! n=0 n=0

∞ ∞ X (2n)! x2n+1 X (2n)! x2n+1 arcsin(x) = arcsinh(x) = (−1)n 4n(n!)2 (2n + 1) 4n(n!)2 (2n + 1) n=0 n=0 ∞ ∞ X x2n+1 X x2n+1 arctan(x) = (−1)n arctanh(x) = (2n + 1) (2n + 1) n=0 n=0

Bn is the nth Bernoulli number; B0 = 1,B2 = 1/6,B4 = 1/30,B6 = 1/42,B8 = 1/30,... . − −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 114 / 147 Circular versus Hyperbolic

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 115 / 147 Homework

1 Problems: 2.6 pg. 117; # 38,§ 65, 68, 71, 74.

2 Read 3.1 §

3 Investigate the HMC Math ‘Convergence Tests’ web site.

4 Explore the text Real Infinite Series by Bonar and Khoury.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 116 / 147 In The Woods

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 117 / 147 Algebras ( 3.1) §

Definition (Algebra of Sets) A collection A of subsets of a set X is an algebra iff 1 X ∈ A 2 if A ∈ A, then Ac ∈ A 3 if A, B ∈ A, then A ∪ B ∈ A

Examples

c {∅,X} {A ⊆ R | either A or A is finite} c {∅, {1, 2}, {3, 4}, {1, 2, 3, 4}} {A ⊆ R | A or A is countable} {∅, (−∞, 0], (0, ∞), R} P(R), the set of all subsets of R (the “power set”)

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 118 / 147 Duality, σ-Algebras

Theorem (De Morgan’s Laws) (A ∪ B)c = Ac ∩ Bc and (A ∩ B)c = Ac ∪ Bc

Theorem If A is an algebra, then 1 ∅ ∈ A 2 if A, B ∈ A, then A ∩ B ∈ A

Definition (σ-Algebra of Sets)

An algebra of sets A is a σ-algebra iff whenever Ak ∈ A for k = 1, 2,... , then the ∞ ∞ [ \ union Ak ∈ A. (And the intersection Ak ∈ A, too.) k=1 k=1

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 119 / 147 Open Covers, Length, and Outer Measure

Definition

A collection of open sets is an open cover of a set A iff A O O. O ⊆ ∈O Define the length of an open interval I = (a, b) to be `(I) = b Sa. The “outer measure” of a set is the minimal length of open covers of the− set. Definition (Outer Measure)

The outer measure of a set E R is µ∗(E) = inf `(I) where is ⊆ O an open cover of E by open intervals. O I X∈O

Examples

1 The outer measure of a finite set is 0.

2 Outer measure of an open interval: I = (a, b) is µ∗(I) = b a. − 3 Outer measure of a closed interval: J = [a, b] is µ∗(J) = b a. −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 120 / 147 Properties

Proposition Outer measure is:

nonnegative: for any set E R, µ∗(E) 0 ⊆ ≥ monotone: if A B, then µ∗(A) µ∗(B) ⊆ ≤ subadditive: for any two sets A and B, µ∗(A B) µ∗(A) + µ∗(B) ∪ ≤ countably subadditive: if E is a sequence of sets, then { n}

µ∗ E µ∗(E ) n ≤ n n ! n [ X Proof. (Countably Subadditivity). ∗ Wolog µ (En) < ∞ for all n. Each En has a countable cover {In,j } of open intervals P ∗ n S S s.t. j `(In,j ) ≤ µ (En) + /2 . Then n{In,j } covers E = En. So ∗ P P ∗ n P ∗ µ (E) ≤ n,j `(In,j ) ≤ n (µ (En) + /2 ) = n (µ (En)) + .

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 121 / 147 Lebesgue’s Measure (1902)

Goals for a set measure function m: 1 m(an interval) = it’s length 2 m(E + x) = m(E) (translation invariant ) countably 3 if E is pairwise disjoint, then m( E ) = m(E ) { n} n n n n additive 4 every set can be measured by m. S P

The continuum hypothesis implies that 1, 3 and 4 are incompatible. Additionally, 2 and 4 are incompatible. We insist on 1, 2, and (nearly) 3. So we give up 4: there will be sets that can’t be measured.

Definition (Caratheodory’s´ Condition) A set E is Lebesgue measurable iff for every ‘test set’ A R, c ⊆ µ∗(A) = µ∗(A E) + µ∗(A E ) ∩ ∩ Define M to be the family of all measurable sets of real numbers.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 122 / 147 Measuring Properties

Proposition If E measurable, then Ec is measurable; i.e., E M Ec M. ∈ ⇐⇒ ∈

Proposition

If µ∗(E) = 0, then E is measurable.

Proof. c c 1 A = (A E) (A E ), therefore µ∗(A) µ∗(A E) + µ∗(A E ) † ∩ ∪ ∩ ≤ ∩ ∩ 2 A E E, thus µ∗(A E) µ∗(E) = 0 ∩ ⊆ ∩ ≤ c c 3 A E A, thus µ∗(A E ) µ∗(A) ∩ ⊆ ∩ ≤ c 4 So µ∗(A) µ∗(A E) + µ∗(A E ) 0 + µ∗(A). E M ≤ ∩ ∩ ≤ ∴ ∈

†This useful inequality is always true by monotonicity. Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 123 / 147 Measure Creates a σ-Algebra

Proposition If E and E M, then E E M. If E M, then E M. 1 2 ∈ 1 ∪ 2 ∈ n ∈ n n ∈ Proof. S

c 1 Let A be any set and use A ∩ E1 as a test-set for E2. So ∗ c ∗ c ∗ c c µ (A ∩ E1) = µ ((A ∩ E1) ∩ E2) + µ ((A ∩ E1) ∩ E2) c 2 A ∩ (E1 ∪ E2) = (A ∩ E1) ∪ (A ∩ E2 ∩ E1) ∗ ∗ c 3 µ (A ∩ (E1 ∪ E2)) + µ (A ∩ (E1 ∪ E2) ) | {z } ∗ ∗ c c = µ (A ∩ (E1 ∪ E2)) +µ (A ∩ (E1 ∩ E2)) | {z } h ∗ ∗ c i ∗ c c ≤ µ (A ∩ E1) + µ (A ∩ E2 ∩ E1) + µ (A ∩ E1 ∩ E2)

∗ h ∗ c ∗ c c i ≤ µ (A ∩ E1) + µ ((A ∩ E1) ∩ E2) + µ ((A ∩ E1) ∩ E2)

∗ | ∗ c ∗ {z } ≤ µ (A ∩ E1) + µ (A ∩ E1) = µ (A) ∗ ∗ ∗ c ∗ 4 Then µ (A) ≤ µ (A ∩ (E1 ∪ E2)) + µ (A ∩ (E1 ∪ E2) ) ≤ µ (A)

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 124 / 147 A Big σ-Algebra

Theorem M is a σ-algebra.

Definition Borel σ-algebra: the smallest σ-algebra that contains all open intervals

Proposition The Borel σ-algebra (R) is a subalgebra of M. Moreover, B (R) M (R). B $ $ P There are Lebesgue measurable sets that are not Borel sets and there are subsets of R that are not measurable. (Both are very complex.)

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 125 / 147 Hm. . .

I wonder. . .

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 126 / 147 a + b = c if and only if

( M) (0, a) M ( x, y) (x, y) M (x0, y0) M (b, c) M ∀ ∈ ∧ ∀ ∈ → ∈ → ∈ n h i o The Measure

Definition (Lebesgue Measure)

For E M, define µ(E) = µ∗(E). That is, µ = µ∗ . ∈ M

Definition (Almost Everywhere) A property P holds almost everywhere, or a.e., iff the set on which P does not hold has measure 0.

Example

χ 1 x Z The characteristic function of the integers is Z(x) = ∈ . Then (0 x / Z χ ∈ Z(x) = 0 a.e.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 127 / 147 Countable Additivity of Lebesgue Measure

Theorem

Let {En} be a countable sequence of pairwise-disjoint measurable sets. Then S∞  P∞ µ n=1 En = n=1 µ(En) .

Proof. I.FINITESEQUENCE. Basis: For n = 1, it’s trivially true. Induction: Assume the results holds for some value n − 1. A finite union of sets in M is in M, so

1 Sn  Sn   Sn  c  µ k=1 Ek = µ k=1 Ek ∩ En + µ k=1 Ek ∩ En Disjoint!

2 Sn−1  = µ(En) + µ k=1 Ek

3 Pn−1 Pn = µ(En) + k=1 µ(Ek) = k=1 µ(Ek) II.INFINITE SEQUENCE.

1 Sn S∞ Sn  Pn S∞  k=1 Ek ⊆ k=1 Ek, so µ k=1 Ek = k=1 µ(Ek) ≤ µ k=1 Ek

2 P∞ S∞  The left side converges (bounded & incr), ∴ k=1 µ(Ek) ≤ µ k=1 Ek 3 The opposite inequality follows from subadditivity.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 128 / 147 More Properties of µ

Proposition If A E and µ(E) = 0, then A M and µ(A) = 0. ⊆ ∈ Theorem

Let En n∞=1 be a sequence of nested, measurable sets with µ(E1) < . Then{ } ∞ ∞ µ En = lim µ(En) n n=1 ! →∞ \ Theorem (Regularity of µ) Let A M. Then for any  > 0 there is an open set O and a closed set F such that i) ∈F A O and ii) µ(O F ) < . ⊆ ⊆ −

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 129 / 147 Homework

1 Problems: 3.1 pg. 166; # 1,§ 2, 5, 7, LA 8, 11.

2 Read 3.2 §

3 Investigate David Bressoud’s Engaging Students through the History of Mathematics talk (Powerpoint slides).

4 Explore the text A Garden of Integrals by Frank Burk.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 130 / 147 The Lebesgue Integral ( 3.2) §

Definition (Measurable Function) Let f be an extended real-valued function on a measurable domain D. TFAE: For each r ∈ R, f −1((r, ∞)) ∈ M ⇔ f −1([r, ∞)) ∈ M ⇔ f −1((−∞, r)) ∈ M ⇔ f −1((−∞, r]) ∈ M A function f that satisfies this criteria is called a measurable function.

Definition (Step Function) A function φ:[a, b] → R is a step function iff there is a partition of [a, b] such that φ is n X χ constant on each interval Ik = (xk−1, xk). Then φ(x) = ak Ik (x). k=1

Definition (Simple Function) n A function ψ :[a, b] → R is a simple function iff the range of ψ is a finite set {ak}k=1 and n −1 X χ each Ak = ψ (ak) is measurable. Then ψ(x) = ak Ak (x). k=1

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 131 / 147 Measurable Functions

Proposition A continuous function is measurable. Theorem (Algebra of Measurable Functions) Let f and g be measurable functions on a common domain D, and let c R. Then f + c, cf, f g, f 2, and fg are all measurable. If a function h is equal∈ to f a.e., then h is measurable.±

Definition (Convergence Almost Everywhere) A sequence f converges to f a.e., written f f a.e., iff the set of points { n} n → where fn fails to converge has measure zero. Theorem

A function f is measurable iff there is a sequence of simple functions ψn converging to f a.e. { }

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 132 / 147 Lebesgue’s Diagram

From Lebesgue’s 1926 address to the “Societ´ e´ Mathematique”´ in Copenhagen.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 133 / 147 Integral of a Simple Function

Definition Let ψ be a simple function that is zero outside of a set of finite measure with n χ 1 ψ(x) = ak Ak (x) where Ak = ψ− (ak). Then kX=1 n ψ(x) dµ = ak µ(Ak) Z kX=1 If E is a measurable set, then n ψ(x) dµ = ak µ(Ak E) E ∩ Z kX=1 Examples 1 Find the integral of Dirichlet’s function on the interval [0, 1]. X 2 Find the integral of Thomae’s function on the interval [0, 1]. How?

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 134 / 147 Integral of a Bounded Function

Definition Let f be a bounded measurable function on a set E having finite measure. Then f dµ = inf ψ dµ ψ f ZE ≥ ZE where ψ is a simple function.

Theorem Let f be bounded and Riemann integrable on [a, b]. Then f is measurable and b f(x) dx = f dµ Za Z[a,b]

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 135 / 147 Integral Properties

Theorem Let f and g be bounded and measurable functions on a set E of finite measure, and let c ∈ R. Then Z Z 1 cf dµ = c f dµ ZE EZ Z 2 f + g dµ = f dµ + g dµ E E Z E Z 3 if f ≤ g a.e., then f dµ ≤ g dµ E E Z 4 if m ≤ f(x) ≤ M a.e. on E, then m · µ(E) ≤ f dµ ≤ M · µ(E) Z E Z Z 5 if E1 ∪ E2 = E with E1 ∩ E2 = ∅, then f dµ + f dµ = f dµ E1 E2 E Z Z

6 f dµ ≤ |f| dµ E E

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 136 / 147 Homework

1 Problems: 3.2 pg. 166; # 21,§ 24, 25, 26, 45.

2 Read 4.2 §

3 Study Lebesgue’s 1926 address to the “Societ´ e´ Mathematique”´ in Copenhagen. See the appendix of Chae’s for an English translation.

4 Explore Lebesgue’s criterion for Riemann integrability: Theorem (Riemann-Lebesgue Theorem)

Let f be bounded on [a, b] and define Df to be the set of discontinuities of f. Then f R([a, b]) µ (D ) = 0 ∈ ⇐⇒ f

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 137 / 147 TI and ( 4.2+) § Definition (TI Floating Point Number (decimal value)) A decimal value is coded internally as a 10 byte1 object:

sign biased mantissa exponent (1 bit) (15 bits) (64 bits: 14 4 bit BCD #’s, 0, 0) × s (exp −16384)B x = (−1) ×mantissa BCD ×10 (TI math functions restrict exponent range to [−999, +999])

Examples +π = 0 16384 3 1415926 535897 00 ↑implied decimal point 10 = 1 16385 1 0000000 000000 00 − 0.1 = 1 16383 1 0000000 000000 00 − 1TI-83 versions used 9 bytes objects that had a smaller exponent. Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 138 / 147 Gaussian Quadrature

The ‘calculus-course techniques’ for numerical integration—left sum, trapezoid rule, Simpson’s rule—all use equally spaced partitions. Can we do better by adjusting the points? n n Z 1 X X The basic formula is f(x) dx ≈ f(xk) ∆xk = f(xk) wk −1 k=1 k=1 Gauss treated both xk and wk as unknowns and looked for a set that would integrate 2n − 1 degree polynomials exactly over [−1, 1]. Let’s take n = 3, so 2n − 1 = 5 and f(x) = 1, x, x2, . . . , x5, successively. Then   w1 + w2 + w3 = 2   x1w1 + x2w2 + x3w3 = 0  n Z 1  2 2 2  X n n x1w1 + x2w2 + x3w3 = 2/3 xk wk = x dx =⇒ 3 3 3 −1 x1w1 + x2w2 + x3w3 = 0 k=1  4 4 4  x1w1 + x2w2 + x3w3 = 2/5  5 5 5  x1w1 + x2w2 + x3w3 = 0  p p giving x1 = − 3/5, x2 = 0, x3 = 3/5 and w1 = 5/9, w2 = 8/9, w3 = 5/9.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 139 / 147 Gaussian Quadrature Examples

Examples

1 [n = 3, 5th degree.] Set p(x) = x5 − 2x4 − x3 + 7x2 + 8x + 1. Then Z 1 88 p(x) dx = −1 15 and     88 GQ = p −p3/5 · (5/9) + p(0) · (8/9) + p p3/5 · (5/9) = 15 RelativeError = 0

2 [n = 3, 6th degree.] Set q(x) = 3x6 − 2x5 + 7x4 − 4x2 + 7x − 8. Then Z 1 1576 q(x) dx = − −1 105 and     1136 GQ = q −p3/5 · (5/9) + q(0) · (8/9) + q p3/5 · (5/9) = − 75

RelativeError = (−1576/105 − −1136/75)/(−1576/105) ≈ 2%

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 140 / 147 Gaussian Quadrature Error

Advantages: Simple weighted sum formula Minimal number of function evaluations Published tables list nodes and weights for many n Disadvantages: Nodes don’t take into account a function’s properties Transformations for domains other than [ 1, 1] New nodes and weights discard old computations−

Theorem (Error Estimate for Gaussian Quadrature) 2n (2n) If f C [ 1, 1] with M2n = maxx [a,b] f (x) , then the error ∈ − ∈ | | 1 n 22n+1[n!]4 f(x) dx wkf(xk) 3 M2n 1 − ≤ (2n + 1) [(2n)!] Z− k=1 X

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 141 / 147 C.F. Gauss, A.S. Kronrod, and TI

Aleksandr Kronrod (1964) discovered a way to extend Gaussian quadrature by adding nodes to an existing set rather than using all new nodes. The error analysis is relatively difficult, but provides a good heuristic. Definition (Gauss-Kronrod 7-15 Quadrature)

Gauss-Kronrod 7-15 quadrature (GK15) is a combination of a 7 node Gaussian (GQ7) with 8 new nodes and weights that gives exact integrals for polynomials up to degree 29. A “conservative, realistic” error estimate is 15 Z 1 √ 1.5 X f(x) dx − f(xk)wk ≤ 2000 2 GQ7 − GK15 −1 k=1

GQ7 nodes weights GK15 new nodes weights 0.94910791234276 0.12948496616887 0.99145537112081 0.02293532201053 ±0.74153118559939 0.27970539148928 ±0.86486442335977 0.10479001032225 ±0.40584515137740 0.38183005050512 ±0.58608723546769 0.16900472663927 ± 0.0 0.41795918367347 ±0.20778495500790 0.20443294007530 ±

Explore Gauss-Kronrod 7-15 quadrature with a Java applet. Consider f(x) = x sin(x2) + x . | |

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 142 / 147 The Points

GK15 points on f

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 143 / 147

GQ7 points GK15 points Example

1 x2 Compute e− dx. (We know this function has no elementary antiderivative.) 1 Z−

TI-84: 1.493648266 TI-89: 1.4936482656246 (but to 6-digits) TI-Nspire CAS: 1.4936482656246 (but to 6-digits) Maple 17: 1.493648266; using GK15 gives 1.493648264 Sage: 1.49364826562485 PocketCAS: 1.49364826562

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 144 / 147 TI’s Quadrature

From the TI Guidebook: TI-84: fnInt( (function integral) returns the numerical integral (Gauss-Kronrod method) of expression with respect to variable, given lower limit, upper limit, and a tolerance (if not specified, the default is 1E 5). fnInt( is valid only for real numbers. [TI-84: − p. 61; TI-83: p. 71]

TI-8910: nInt( returns an approximation of (expression1, var, lower, upper). This approximation is a weighted average of some sample values of the integrand inR the interval lower < var < upper.[TI-89: p. 472; TI-Nspire CAS: p. 76] (Uses adaptive G-K with tolerance = “ 6 significant digits.”)

10Also for TI-92, Voyage 200, Nspire, and Nspire CAS

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 145 / 147 Homework

1 Problems: 4.2 pg. 214; # 13,§ 14, 15, 24, 59.

2 Read everything

3 Investigate numerical quadrature.

4 Explore David Bressoud’s texts A Radical Approach to Real Analysis and A Radical Approach to Lebesgue’s Theory of Integration. Two great books.

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 146 / 147 Here Enˇth t˙ Le&on...

Wm C Bauldry ([email protected]) MAT 5930. Analysis for Teachers. Fall ’13 147 / 147