Integration Theory: Lecture notes 2013

Size: px
Start display at page:

Download "Integration Theory: Lecture notes 2013"

Transcription

1 Itegratio Theory: Lecture otes 203 Joha Joasso September 203 Preface These lecture otes are writte whe the course i itegratio theory is for the first time i more tha twety years, give joitly by the the two divisios Mathematics ad Mathematical Statistics. The major source is G. B. Follad: Real Aalysis, Moder Techiques ad Their Applicatios. However, the parts o probability theory are mostly take from D. Williams: Probability with Martigales. Aother source is Christer Borell s lecture otes from previous versios of this course, see 2 Itroductio This course itroduces the cocepts of measures, measurable fuctios ad Lebesgue itegrals. The itegral used i earlier math courses is the so called Riema itegral. The Lebesgue itegral will tur out to be more powerful i the sese that it allows us to defie itegrals of ot oly Riema itegrable fuctios, but also some fuctios for which the Riema itegral is ot defied. Most importatly however, is that it will allow us to rigorously prove may results for which proofs of the correspodig results i the Riema settig are usually ever see by studets at the basic ad itermediate level. Such results iclude precise coditios for whe we ca chage order of itegrals ad limits, chage order of itegratio Chalmers Uiversity of Techology Göteborg Uiversity joasso@chalmers.se

2 i multiple itegrals ad whe we ca use itegratio by parts. Of course, we will also prove may ew results. The cocept of measurability is a advaced oe, i the sese that a lot of people at first fid it difficult to master; it teds to feel fudametally more abstract tha thigs oe has ecoutered before. Therefore, a atural first questio is why the cocept is eeded. To aswer this, cosider the followig example. Let X = R/Z, the circle of circumferece, with additio ad multiplicatio defied modulo. Suppose we wat to itroduce the cocept of the legth of subsets of X. A atural first assumptio is that oe should be able to do this so that the legth is defied for all subsets of X. It is also extremely atural to claim that the legth l, should satisfy l( ) = 0, l(x) =, l( )A = l(a ) for all disjoit A, A 2,..., l(a + x) = l(a) for all A X ad x X. However, if we isist o defiig l for all subsets, this turs out to be impossible. Let us see why. Partitio X ito equivalece classes by sayig that x ad y are equivalet if x y is a ratioal umber. By the axiom of choice, there exists a set A cotaiig exactly oe elemet from each equivalece class. For each q Q X, let A q = A + q. The q A q = X, for sice for each x X, A cotais a elemet y equivalet to x, i.e. x A x y ad x y Q. O the other had, the A q s are disjoit, for if x A q A q2, the x = y+q = z + q 2 for two elemets y, z A. However, the y z = q 2 q Q, so y ad z are equivalet, cotradictig the costructio of A. If we could assig legths to the A q s, the these legths must be equal by the fourth coditio o l. O the other had, the legths of the A q s must sum to by the third coditio. However, these two coditios are mutually exclusive. The moral of the example is that the set A must be declared o-measurable; o legth of A ca be defied. The costructio of the example is based o the axiom of choice ad it ca be show that all costructios of o-measurable sets must rely o the axiom of choice. There are eve more absurd examples tha this oe. The famous Baach- Tarski paradox proves, usig the axiom of choice, that for ay two bouded compact sets i R 3, the oe ca be divided ito a fiite umber of parts which ca be 2

3 traslated ad rotated ad mirrored ad the put back together to form the other. For example: ay grai of sad ca be divided ito a umber of pieces that ca be put back together to form a ball the size of the earth! Clearly theses pieces caot have a well defied volume. Examples like these call for a theory of measures ad measurable sets. 3 Measures We are goig to cosider measures i a very geeral framework: we will cosider measures o a a abstract space X o we which we make o iitial assumptios whatsoever. As the above example revealed, it is ot always possible with meaigful measures defied o all subsets of X. Hece a cocept of what classes of subsets to defie a desired measure o, is eeded. The two last coditios o a legth measure i the above example were atural i that particular situatio, but it is easy to thik of other situatios where either of them is atural or eve meaigful. The two first coditios however, are such that they should hold for aythig that deserves to be called a measure, o matter what structure X has. Thus we keep those two coditios i mid, ad ask for classes of subsets large eough to esure that all iterestig set operatios o measurable sets results i a measurable set, but restrictive eough to make sure that o coflict with the basic assumptios arises. The aswer is σ-algebras. 3. Algebras ad σ-algebras Defiitio 3. Let A be a class of subsets of X such that (i) X A, (ii) E c A wheever E A, (iii) E F A wheever E, F A. The A is called a algebra (o X). Note that by (i) ad (ii), = X c A. Also, if E, F A, the E F = (E c F c ) c A by (ii) ad (iii). Defiitio 3.2 Let M be a class of subsets of X such that 3

4 (i) X M, (ii) E c M wheever E M, (iii) = E M wheever E, E 2,... M. The M is called a σ-algebra. Clearly ay σ-algebra is a algebra. As above M, ad aalogously, if E, E 2,... M, the E = ( Ec ) c M. A measure will always be defied o a σ-algebra. The smallest possible σ- algebra o ay space X is {, X}. The largest σ-algebra is P(X), the class of all subsets of X (but we have see that meaigful measures caot always be defied o this σ-algebra). If M is a σ-algebra o X, the the pair (X, M) is called a measurable space ad a set E M is called M-measurable. 3.2 Geerated σ-algebras Let C be a arbitrary class of subsets of X. We defie the σ-algebra geerated by C as the smallest σ-algebra cotaiig C, i.e. σ(c) = {F : F σ-algebra, F C}. (It is a easy exercise to show that ay itersectio of σ-algebras is a σ-algebra.) The most importat example is the Borel σ-algebra; if X is a topological space ad T is the class of ope sets, the the Borel σ-algebra, B(X), is give by B(X) = σ(t ). Sice ay ope set i R is a coutable uio of ope itervals, it follows that B(R) = σ((a, b) : a, b R). It is ow easy to see (check this!) that we also have B(R) = σ([a, b) : a, b R) = σ((a, b] : a, b R) = σ([a, b] : a, b R) = σ((, b) : b R) = σ((a, ) : a R). I itegratio theory, oe ofte works with the exteded real lie, R = [, ] ad, eve more, with the exteded positive half-lie R + = [0, ]. Here the arithmetics ivolvig the poits ad work as oe would ituitively guess, ad a subset is regarded as ope if it is either a subset of R ad ope as such, of the form [, a) or (a, ], or the whole space. It is ow straightforward to prove aalogous expressios for B(R) ad B(R + ). 4

5 3.3 Measures If C is a class of subsets of X ad µ 0 : C R +, the µ 0 is called a set fuctio. Let A be a algebra. If µ 0 is a set fuctio o A such that µ 0 ( ) = 0 ad E, F A, E F = implies µ 0 (E F ) = µ 0 (E) + µ 0 (F ), the µ 0 is said to be additive. If µ 0 ( ) = 0 ad µ 0 satisfies the stroger coditio that µ 0 ( E ) = µ 0(E ) wheever E, E 2,... A ad E A, the µ 0 is said to be coutably additive or a premeasure. (Stroger sice additivity follows from coutable additivity by takig E = E, E 2 = F ad E 3 = E 4 =... =.) Defiitio 3.3 Let M be a σ-algebra ad µ a set fuctio defied o M. If µ is coutably additive, the µ is said to be a measure. Let µ be a measure o the σ-algebra M. Here are a few classificatios. µ is said to be fiite if µ(x) <. µ ca be said to be a probability measure if µ(x) =. µ is said to be σ-fiite if there exist sets E, E 2,... M such that E = X ad µ(e ) < for all. µ is said to be semi-fiite if for every E M such that µ(e) =, there exists a set F E such that 0 < µ(f ) <. The trivial measure is the measure µ with µ(e) = 0 for all E M. Clearly ay probability measure is fiite, ay fiite measure is σ-fiite ad every σ-fiite measure is semi-fiite. Example. Let µ( ) = 0 ad µ(e) = for ay oempty measurable E. The µ is a measure which is ot eve semi-fiite. Example. Legth measure o [0, ] (which, to be true, we have ot defied yet) is a probability measure. Legth measure o R is σ-fiite; take e.g. E = (, ). Whe M is a σ-algebra o X ad µ is a measure o M, the triple (X, M, µ) is called a measure space. If µ(x) =, the we may also speak of (X, M, µ) as a probability space ad if we do that, we usually refer to M-measurable sets as evets. 5

6 Remark. Suppose that µ(x) =. The we ca choose to call µ a probability measure ad (X, M, µ) a probability space. Whether or ot we actually do that depeds o the poit of view we wat to adopt. I may situatios it is either our mai purpose to model a radom experimet or it is istructive or useful for some other reaso to thik of the poits x X as the possible outcomes of a radom experimet. If this is ot the case, we may istead prefer to just refer to µ as a fiite measure of total mass. Some geeral properties of measures follow. I all of these, it is assumed that (X, M, µ) is a measure space. Propositio 3.4 (a) E, F M, E F µ(e) µ(f ). (b) E, E 2,... M µ( E ) µ(e ), (c) If µ(x) <, the µ(e F ) = µ(e) + µ(f ) µ(e F ), (d) If µ(x) <, E, F M ad E F, the µ(f \ E) = µ(f ) µ(e). Proof. By additivity of µ, µ(f ) = µ(e) + µ(f \ E) wheever E F. This proves (d) ad sice µ(f \ E) 0, (a) follows too. For (b), let F = E ad recursively F = E \ F j, = 2, 3,.... The the F s are disjoit ad F = E, so by (a) µ( E ) = µ(f ) µ(e ). Fially (c) follows from µ(e F ) = µ(e) + µ(f \ E F ) = µ(e) + µ(f ) µ(e F ) by additivity ad (d). Propositio 3.5 (Cotiuity of measures) (a) If E E 2... ad E = E, the µ(e) = lim µ(e ). (b) If F F 2..., F = F ad µ(f ) <, the µ(f ) = lim µ(f ). 6

7 Proof. For (a), let A = E ad recursively A = E \ E. The E = A ad the A s are disjoit, so µ(e) = µ(a j ) = lim µ(a j ) = lim µ(e ) sice E = A j. Now (b) follows from applyig (a) to E = F \ F ad E = F \ F ad usig Propositio 3.4(d). Corollary 3.6 If µ(n ) = 0 for all, the µ( N ) = 0. Proof. Apply e.g. Propositio 3.4(b). 3.4 Almost everywhere ad completeess Let S be a propositio about poits of X ad suppose that F = {x : S(x) is false} is measurable. If µ(f ) = 0, the S is said to hold almost everywhere (with respect to µ if other measures are also uder discussio), abbreviated a.e. I case µ is a probability measure, oe ofte istead says that S holds almost surely, abbreviated a.s. If S holds a.e. ad T is aother propositio such that T (x) is true wheever S is true, the oe would clearly wat to thik of T as also holdig a.e. However this is ot so i geeral, sice eve if µ(f ) = 0, it may be the case that some subset E of F is ot measurable. If (X, M, µ) is such that E M wheever E F, F M ad µ(f ) = 0, the the measure space is said to be complete ad µ is said to be a complete measure. If µ is ot complete, the oe ca always exted the measure space, by defiig the larger σ-algebra M = {E F : E M, N M : F N, µ(n) = 0} (exercise: prove that M is a σ-algebra) ad the measure µ o M by µ(e F ) = µ(e). The (X, M, µ) is complete ad µ is called the completio of µ. 3.5 Dyki s Lemma ad the Uiqueess Theorem Dyki s Lemma will be a fudametal tool for theorem provig. It is based o the cocepts of π-systems ad d-systems. A π-system is a class I of subsets of X 7

8 that is closed uder fiite itersectios, i.e. E F I wheever E, F I. The defiitio of a d-system follows. Defiitio 3.7 Let D be a class of subsets of X. The D is said to be d-system if (a) X D, (b) E, F D, E F F \ E D, (c) E D, E E E D. Geerated d-systems are defied aalogously with geerated σ-algebras: d(c) = {D C : D d-system}s. (Check that ay itersectio of d-systems is a d-system.) Theorem 3.8 Let M be a class of subsets of X. The M is a σ-algebra if ad oly if it is π-system ad a d-system. Proof. The oly if-directio is obvious. The if directio follows from that X M by (a) i the defiitio of a d-system, E c = X \ E M wheever E M by (b) ad if E M, =, 2,..., the F := E j = ( Ec j) c M sice M is a π-system, so E := E j M by (c) sice F E. Sice ay σ-algebra is also a d-system, it follows that σ(c) d(c) for ay C. Dyki s Lemma provides a aswer to whe we have equality. Theorem 3.9 (Dyki s Lemma) If I is a π-system, the d(i) = σ(i). Proof. It suffices to prove that d(i) σ(i). By Theorem 3.8 it thus suffices to prove that d(i) is a π-system. I other words, it suffices to prove that D 2 := {B d(i) : B C d(i) for all C d(i)} equals d(i). The proof is doe i two similar steps. For step, defie D := {B d(i) : B C d(i) for all C I}. Sice I is a π-system, D cotais I, so if we ca show that D is a d-system, the D = d(i). Part (a) i the defiitio of a d-system obviously holds. If B, B 2 8

9 D ad B B 2, the for ay C I, (B 2 \B ) C = (B 2 C)\(B C) d(i) sice d(i) is a d-system. Hece part (b) holds for D. Fially if B D ad B B, the B C B C, so B D sice d(i) is a d-system. That D = d(i) meas that D 2 I, so it suffices ow to prove that D 2 is a d-system, which is ow doe i complete aalogy with step. (Check that you ca fill this i.) Our first applicatio is the followig uiqueess theorem for measures. Theorem 3.0 (Uiqueess of fiite measures) Suppose that I is a π-system ad M = σ(i). If µ ad µ 2 are two measures o M such that µ (X) = µ 2 (X) < ad µ (I) = µ 2 (I) for all I I, the µ = µ 2. Proof. By Dyki s Lemma, it suffices to prove that D := {E M : µ (E) = µ 2 (E)} is a d-system. That X D follows from the first part of the assumptio. If E, F D ad E F, the µ (F \ E) = µ (F ) µ (E) = µ 2 (F ) µ 2 (E) = µ 2 (F \ E), so F \ E D. Fially if E D ad E E, the µ (E ) = µ 2 (E ), so µ (E) = µ 2 (E) by the cotiuity of measures. Corollary 3. If two probability measures agree o I, the they are equal. 3.6 Borel-Catelli s First Lemma Defiitio 3.2 Let E, E 2,... be subsets of X. The Note that lim sup E := lim if E := m= =m m= =m E E. lim sup E = {x X : x E for ifiitely may } ad lim if E = {x X : x E for all but fiitely may }. 9

10 Oe sometimes writes E i.o. for lim sup E, where i.o. stads for ifiitely ofte. (There is o correspodig abbreviatio for lim if E.) Let (X, M, µ) be a measure space ad suppose that E, E 2,... M. Sice a σ-algebra is closed uder coutable itersectios ad uios, it is clear that lim sup E ad lim if E are the also measurable. Lemma 3.3 (Borel-Catelli s Lemma I) If = µ(e ) <, the µ(lim sup E ) = 0. Proof. Write F m = =m E ad F = lim sup E. The F F. Sice M = E F it follows from the cotiuity of measures (from below) ad the hypothesis that M µ(f ) = lim µ( E ) lim M M M µ(e ) = µ(e ) <. Hece the cotiuity of measures (from above) ad the hypothesis imply that µ(f ) = lim µ(f m ) µ(e ) = 0. m =m The Borel-Catelli Lemma is a importat tool, i particular i probability theory. Example. (The doublig strategy.) Assume that (X, M, P) is a probability space ad suppose that E, E 2,... are evets such that P(E ) = 2, =, 2,.... The by the Borel-Catelli Lemma, P(lim sup E ) = P(E i.o.) = 0. Oe way to describe this i words is the followig. Suppose we play a sequece of games such that at the th game we wi oe c.u. with probability 2 ad lose 2 c.u. with probability 2. Each game is fair i terms of expectatio, but by the Borel-Catelli Lemma, we will almost surely lose moey oly fiitely may times. Hece, over the whole ifiite sequece of games, we will almost surely wi a ifiite amout of moey. (I practice this strategy fails, of course, sice there are always some bouds that will set thigs up, e.g. oe ca oly play a certai umber of games i a lifetime.) 0

11 3.7 Carathéodory s Extesio Theorem A set fuctio µ : P(X) [0, ] is said to be a outer measure if µ ( ) = 0, µ (E) µ (F ) wheever E F, µ ( = E ) = µ (E ) for all sets E, E 2,.... If µ is a outer measure, the we say that a set A P(X) is µ -measurable if, for all E P(X), µ (E) = µ (E A) + µ (E A c ). By the defiitio of outer measure, it is immediate that the left had side is bouded by the right had side, so to prove that a give set A is µ -measurable, it suffices to show that µ (E) µ (E A) + µ (E A c ) for arbitrary E with µ (E) <. Theorem 3.4 (Carathéodory s Extesio Theorem) Let A be a algebra o X ad let µ 0 : A [0, ] be a coutably additive set fuctio. The there exists a measure µ o σ(a) such that µ(a) = µ 0 (A) for all A A. If µ 0 (X) <, the µ is the uique such measure. The uiqueess part follows immediately from Theorem 3.0. The existece part will be proved via a sequece of claims. These will also reveal some other useful facts, apart from the statemet of the theorem. Claim I. Let µ be a outer measure ad let M be the collectio of µ -measurable sets. The M is a σ-algebra. Moreover, the restrictio of µ to M is a complete measure. Proof. It is obvious that X M. From the symmetry betwee A ad A c i the defiitio of µ -measurability, it is also obvious that M is closed uder complemets. It remais to show that M is closed uder coutable uios. Suppose that A, B M ad let E be a arbitrary subset of X. The A B M sice µ (E) = µ (E A) + µ (E A c ) = µ (E A B) + µ (E A B c ) + µ (E A c B) + µ (E A c B c ) = µ (E (A B)) + µ (E (A B) c

12 where the last iequality follows from that A B = (A B) (A B c ) (A c B), so that the defiitio of outer measure implies that the first three terms i the middle expressio boud the first term i the last expressio, ad that (A B) c = A c B c. Moreover, if A B =, the (A B) A = A ad (A B) A c = B, so the applyig the defiitio of µ -measurability of A with E = A B gives µ (A B) = µ (A) + µ (B). I summary M is closed uder fiite uios ad µ is additive o M. Now suppose that A j M, j =, 2,... are disjoit sets. Write B = A j ad B = A j. Let E be a arbitrary subset of X. By the µ -measurability of A, µ (E B ) = µ (E B A ) + µ (E B A c ) so by iductio it follows that = µ (E A ) + µ (E B ) µ (E B ) = µ (E A j ). Above, we proved that M is closed uder fiite uios, so B M for each. Hece µ (E) = µ (E B ) + µ (E B c ) = µ (E A j ) + µ (E B c ). µ (E A j ) + µ (E B) c Lettig ad usig the defiitio of outer measure, it follows that µ (E) ( ) µ (A j ) + µ (E B c ) µ (E A j ) + µ (E B c ) = µ (E B) + µ (E B c ) µ (E). Hece all the iequalities must be equalities ad it follows that B M. This proves that M is closed uder disjoit coutable uios ad it is a easy exercise 2

13 to show that this etails that M is closed uder arbitrary coutable uios, i.e. M is a σ-algebra. Moreover, takig E = B gives µ (B) = µ (A j ) provig that the restrictio of µ to M is a measure. It remais to prove completeess. Assume that N M, µ (N) = 0 ad A N. The µ (A) = 0 by the defiitio of outer measure. Therefore µ (E) µ (E A) + µ (E A c ) = µ (E A c ) µ (E) provig that A M. Next assume that µ 0 is a coutably additive set fuctio o the algebra A. Defie µ : P(X) [0, ] by µ (E) = if{ µ 0 (A j ) : A j A, A j E}. () Claim II. µ is a outer measure. Proof. It is trivial that µ ( ) = 0 ad E F µ (E) µ (F ). It remais to prove coutable subadditivity. Fix ɛ > 0. If E j P(X), j =, 2,..., the for each j oe ca fid A j (k) A, k =, 2,... so that k A j(k) E j ad k µ 0(A j (k)) µ (E j ) + ɛ2 j. Sice j,k A j(k) j E j, we get µ ( j E j ) j,k µ 0 (A j (k)) j µ (E j ) + ɛ ad sice ɛ was arbitrary, µ ( j E j ) j µ (E j ) as desired. For the fial two claims, it is assumed that µ is defied by () ad M is the σ-algebra of µ -measurable sets. Claim III. µ (E) = µ 0 (A) for all E A. 3

14 Proof. If E A, take E = A ad E 2 = E 3 =... = i the defiitio of µ to see that µ (E) µ 0 (E). Provig the reverse iequality amouts to showig that µ 0 (A) j µ 0(A j ) wheever A j A ad j A j E. Let B = E (A \ A j. The the B s are disjoit ad B = E. By the coutable additivity of µ 0, it follows that µ 0 (E) µ 0 (B ) µ 0 (A ). Claim IV. A M. Proof. Pick A A ad arbitrary E X ad ɛ > 0. By the defiitio of µ, there exist B j A such that j B j E ad j µ 0(B j ) < µ (E) + ɛ. We get, by the additivity of µ 0 o A, µ (E) + ɛ > j µ 0 (B j A) + j µ 0 (B j A c ) µ (E A) + µ(e A c ) where the last equality follows from the defiitio of µ. Take together, these four claims prove Carathéodory s Theorem. 3.8 The Lebesgue measure ad Lebesgue-Stieltjes measures Up to ow, we have ot see ay cocrete examples of o-trivial measures. Whe X is a coutable space, X = {x, x 2,...}, the it is easy to costruct such measures. Take e.g. M = P(X), let {w(x } = be ay collectio of oegative umbers ad let µ be defied by µ(a) = x A w(x). We have also see that for X = (0, ] ad M = P(X), o sesible legth measure exists. We are ow equipped with the tools eeded to costruct a proper legth measure o R. Sice it is ot possible to do this for all subsets, we have to settle for a smaller σ- algebra. Clearly sets of the form costructed i Sectio 2 via the axiom of choice, are uatural to expect to be able to measure i terms of legth. O the other had, ay sesible legth measure must be able to measure the legth of a iterval. If we could also measure the legth of ay set that ca be costructed from a coutable umber of set operatios o itervals, the it is difficult eough to come up with a example of a set which would ot have a legth (such as the set A i 4

15 Sectio 2) ad eve harder to motivate why oe would eve wish to give such a set a legth if doig so causes problems. This poit of view is what we are goig to adopt. Now recall that the Borel σ-algebra is the σ-algebra geerated by all itervals ad hece, by virtue of beig a σ-algebra, cotais all sets we wish to assig a legth to. Hece the aim is to costruct a legth measure o B(R). It turs out to be slightly more comfortable to restrict to (0, ] ad B(0, ]. Havig doe so, we obviously also have legth measures o (, + ] for all Z by traslatio ad ca exted to the whole real lie by lettig, for E B(R), defiig the legth of E be the sum of the legths of E (, + ], Z. Let X = (0, ] ad let A be the algebra cosistig of fiite disjoit uios of itervals of the type (a, b], 0 a b. Hece ay A A ca be writte as (a j, b j ] for some Z + ad the (a j, b j ] s disjoit. Defie µ 0 : A [0, ] by µ 0 ( (a j, b j ]) = (b j a j ). Clearly the legth of ay set i A must be give by µ 0 (A), so we would like to exted µ to a measure o B(0, ] = σ(a). By Carathéodory s Extesio Theorem, there is a uique such extesio, provided that µ 0 is a coutably additive set fuctio o A. It is trivial that µ 0 ( ) = 0 ad that µ 0 is additive, but coutable additivity is ot so clear. It must be proved that µ 0 ( A ) = µ 0(A ) wheever A, A 2 are disjoit sets i A ad A. Sice µ 0 is fiitely additive, we may assume without loss of geerality that the A s ad A cosist of a sigle iterval: A = (a, b ] ad A = (a, b]. O oe had, by fiite additivity, µ 0 (A) = µ 0 (A \ A j ) + µ 0 ( ( ) A j ) µ 0 A j = µ 0 (A j ) for every, so lettig gives µ 0 (A) µ 0 (A j ). Now we focus o the reverse iequality. Fix ɛ > 0. The sets (a ɛ2, b ) form a ope cover of the set compact set [a, b ɛ] ad ca hece be reduced to a fiite subcover (a ɛ2, b ), =,..., N. Let c = a ɛ2 ad assume 5

16 without loss of geerality that c c 2... c N (otherwise just reorder). We may also assume without loss of geerality that b b 2... b N, otherwise discard those itervals that are cotaied i oe of the others; this caot icrease N (b j a j ). The, sice b j a j+ for all i =,..., N, b ɛ a b N c Hece N (b j c j ) µ 0 (A) = b a N (b j a j + ɛ2 j ) µ 0 (A j ) + 2ɛ. (b j a j ) + ɛ. This establishes that µ 0 is coutably additive. Hece µ 0 exteds to a uique legth measure µ o B(0, ]. This measure is kow as the Lebesgue measure ad the otatio we will use for it is m. Lookig back o the proof of Carathéodory s Extesio Theorem, we fid that for sets E B(0, ] that are ot i A, m(e) is explicitly expressed i terms of µ 0 by µ (E) = if{ µ 0 (A ) : A A, A E} (2) ad m the restrictio of the outer measure µ to B(0, ]. Moreover, we recall that µ 0 actually exteds to a complete measure o the σ-algebra M of µ -measurable sets. This σ-algebra cotais A ad hece B(0, ], but othig says that it could ot be larger. Ideed, it turs out that M equals the completio of B(0, ] with respect to m ad that this σ-algebra is strictly larger tha the Borel σ-algebra. The larger σ-algebra M is called the Lebesgue σ-algebra, deoted L(0, ]. Sice this extesio comes at o extra cost, it will be assumed throughout that the Lebesgue measure is the complete measure defied o L(0, ], uless otherwise stated. The costructio of the Lebesgue measure ca easily be geeralized i the followig way. Let F : R R be a o-decreasig right-cotiuous fuctio. Redefie the µ 0 above by µ 0,F ( A j ) = (F (b j ) F (a j )). A aalogous argumet shows that µ 0 is coutably additive o A ad hece exteds to a uique measure µ F o B(R). For sets E B(R) \ A, (2) becomes µ F (E) = if{ µ 0,F (A ) : A A, A E} (3) 6

17 ad µ F the restrictio of µ F to B(R). As for the Lebesgue measure, the σ-algebra M F of µ F -measurable sets is strictly larger tha B(R) ad the restrictio of µ F to M F coicides with the completio of µ F. I aalogy with the Lebesque measure, we will heceforth take the otatio µ F to deote this completio uless otherwise stated. The measure µ F thus costructed is called the Lebesgue-Stieltjes measure associated to F. From (3) it follows (exercise!) that a Lebesgue-Stieltjes measure satisfies the followig regularity properties, called outer regularity ad ier regularity respectively. Propositio 3.5 For all E M F, µ F (E) = if{µ F (U) : U ope, U E} = sup{µ F (K) : K compact, K E}. Aother property i the same vei is the followig. Propositio 3.6 For all E M F ad ɛ > 0, there exists a set A, which is a fiite uio of ope itervals, such that 3.9 The Cator Set µ F (A E) < ɛ. For ay x R, we have m({x}) = 0, so for ay coutable subset E R, m(e) = 0. Does the reverse implicatio also hold? I.e. are coutable sets the oly oes to have Lebesgue measure 0? The aswer is o. The most well-kow example is the Cator set. It is costructed the followig way. Let for =, 2,..., D = 3 j=0 ((3j + )3, (3j + 2)3 ). Let C = [0, ] \ D ad recursively C = C \ D. Let C = C. The set C is the Cator set. I words, the process is the followig. Start with the closed uit iterval with the ope mid third removed; this is C. From the two closed itervals that make up C, remove from each of them the ope mid third to get C 2. Now C 2 is the uio of four closed itervals. Remove from each of these the ope mid third to get C 3, 7

18 etc. The Cator set is the limitig set of this process. Clearly m(c ) = (2/3), so by the cotiuity of measures m(c) = 0. O the other had, C has the same cardiality as (0, ]. To see this, write each umber x [0, ] by its triary expasio: x = a (x)3 = where a (x) {0,, 2}. The expasio is uique for all x except those that are of the type x = j3, j Z +, for which oe ca either choose a expasio edig with a ifiite sequece of 0 s or oe edig with a ifiite sequece of 2 s. I such cases, we pick the latter expasio. The C = {x {0, } : a (x) {0, 2} for each }. Hece, by mappig each 2 to, we see that C is i a --correspodece with the set of all biary expasios b 2, i.e. with (0, ]. 4 Measurable fuctios / radom variables Let (X, M, µ) be a measure space ad let (Y, N ) be a measurable space. Defiitio 4. A fuctio f : X Y is said to be (M, N )-measurable if f (A) M for all A N. So f is (M, N )-measurable if {x X : f(x) A} is M-measurable wheever A is N -measurable. I words, this could be phrased as that f is measurable if statemets that make sese i terms of the values of f also make sese i terms of the values of x. See the probabilistic iterpretatio of this i the example below. Whe oe of the σ-algebras is uderstood, we may speak of f as simply M- measurable or N -measurable ad if M ad N are both uderstood, we may speak of f as simply measurable. If (X, M, µ) is a probability space, a (M, N )- measurable fuctio is usually called a (Y -valued) radom variable. Example. Let (X, M, P) be a probability space ad suppose Y = (R, B(R)). Let ξ : X R be a radom variable. This meas that ξ is a (M, B(R))-measurable fuctio, i.e. ξ (B) = {x X : ξ(x) B} M 8

19 wheever B B(R). Hece P(ξ (B)) = P(ξ B) is defied for all Borel sets B. I.e. measurability meas that it makes sese to speak of the probability that ξ belogs to B for ay give Borel set B. Clearly the compositio of two measurable fuctios is measurable. More specifically, if (Z, O) is a third measurable space, f : X Y is (M, N )- measurable ad g : Y Z is (N, O)-measurable, the, sice (g f) (A) = f (g (A)), g f is (M, O)-measurable. The followig result is a idispesable tool for provig that a give fuctio is measurable. Theorem 4.2 Let E be a class of subsets of Y ad assume that N = σ(e). The f : X Y is measurable if ad oly if f (A) M for all A E. Proof. The oly if directio is trivial. Let F = {A N : f (A) M}. Sice F E, it suffices to show that F is a σ-algebra. The key is the to recall that f commutes as a operator with the basic set operatios, i.e. f (A c ) = f (A) c ad ( ) f A α = ( ) f (A α ), f A α = f (A α ) α α α for all A ad A α ad α ragig over arbitrary idex sets. Hece X = f (Y ) ad X M (sice M is a σ-algebra), so Y F, A F f (A) M f (A) c M f (A c ) M A c F, A F, =, 2,... f (A ) M f (A ) M f ( A ) M A F. α Corollary 4.3 If X ad Y are topological spaces ad M ad N are the Borel σ-algebras, the ay cotiuous fuctio is measurable. Proof. Let f be cotiuous ad let T be the topology (i.e. the family of ope sets) of Y. By the defiitio of cotiuity, f (U) is ope for all U T ad hece measurable by the defiitio of the Borel σ-algebra o X. Sice B(Y ) = σ(t ) a applicatio of Theorem 4.2 with E = T gives the result. 9

20 Corollary 4.4 A map f : X R is (Borel)-measurable i either of the followig cases f [, a] M for all a R, f [, a) M for all a R, f [a, ] M for all a R, f (a, ] M for all a R. Sice either of the four classes geerate B(R), the proofs follow o mimickig the proof of Corollary 4.3. Of course aalogous statemets are valid if R is replaced with R, R + or R +. Example. Let X be the sample space of a radom experimet. The ξ : X R is a radom variable iff {ξ a} is a evet for all a R. This is sometimes take as the defiitio of a radom variable i courses which wat to preset the ecessary fudametals without ivolvig uecessary measure-theoretic detail. Theorem 4.5 Let f, g : X R be measurable ad λ R a costat. The f + g, λf ad fg are all measurable fuctios. The same is true for /f provided that f(x) > 0 for all x X. Proof. We do f + g ad leave the other cases as exercises. By Corollary 4.4 it suffices to show that {x : f(x) + g(x) < a} M for all a R. However {x : f(x) + g(x) < a} = ( ) {x : f(x) < q} {x : g(x) < a q} M q Q sice Q is coutable ad f ad g are measurable. Theorem 4.6 Assume that f, f 2,... are measurable. The sup f, if f, lim sup f ad lim if f are measurable. Moreover, the set {x : lim f (x), exists} is measurable ad if lim f (x) exists for all x, the lim f (x) is a measurable fuctio. 20

21 Proof. That sup f is measurable follows from the observatio that {x : sup f (x) a} = {x : f (x) a}, a coutable uio of measurable sets. Sice costat fuctios are trivially measurable, we get that if f = 0 sup ( f ) is measurable. Sice lim sup f = if m sup m f ad lim if f = sup m if m f, these are the also measurable. If lim f (x) exists for all x, the lim f = lim if f = lim sup f ad is hece measurable. Fially {x : lim f (x) exists} = {x : lim sup (x) lim if(x) = 0} is measurable by Theorem 4.5 (sice {0} B(R)). Example. Costructio of a uiform radom variable. Let (X, M, P) = ([0, ], B, m) ad ξ(x) = x, x X. The ξ is cotiuous ad hece a radom variable ad P(ξ a) = m{x : ξ(x) x} = m{x : x a} = m[0, a] = a. Example. Costructio of a radom variable with give distributio. Assume that F : R R is o-decreasig ad right cotiuous with lim F (x) = 0, lim F (x) =. x x We wat to costruct a radom variable ξ so that P(ξ a) = F (a). Recall the Lebesque-Stieltjes measure µ F. The coditios o F imply that µ F is a probability measure, so let (X, M, P) = (R, B, µ F ) ad ξ(x) = x, x R. The P(ξ a) = µ F (, a] = F (a). A alterative costructio is the followig, which is most coveietly described i the case whe F is cotiuous ad strictly icreasig. The F exists, so we ca take (X, M, P) = ([0, ], B, m) ad ξ(x) = F (x) ad get P(ξ a) = m{x : F (x) a} = m[0, F (a)] = F (a). I the geeral case, oe ca replace F with the geeralized iverse, which maps all poits i [F (x ), F (x+)] to x ad poits y [0, ] for which F ({y}) is a 2

22 iterval, which must have the form [c, d) or [c, d] sice F is right cotiuous, to c. Example. Costructio of a sequece of uiform radom variables. Agai take (X, M, P) = ([0, ], B, m). Represet each x [0, ] with its biary expasio x = a (x)2. Each a (x) is a {0, }-valued measurable fuctio of x, sice a ({}) is a uio of 2 itervals (of legth 2 ). Let { ij } j=, i =, 2,... be disjoit sequeces ad let ξ i (x) = a ij 2 j. j= The ξ is measurable for each i by Theorems 4.5 ad 4.6 (why do we eed them both?) ad clearly P(ξ i a) = a as i the first of the previous examples. Example. Costructio of a sequece of fair coi flips. With the same settig as i the previous example, let simply ξ i (x) = a i (x). We ed this sectio with a few otes o completeess. Suppose that g is M-measurable ad that f = g a.e. If µ is complete, the this implies that f is measurable. However if µ is ot complete, the this may ot be the case. O the other had, by the costructio of the completio µ of µ, it is clear that f is M- measurable. Similarly, if µ is complete, f, f 2,... measurable ad f f a.e., the f is measurable. (These facts make up Propositio 2. i Follad.) Vice versa, if f is M-measurable, the there exists a M-measurable fuctio such that f = g µ-a.e. (This last fact is Propositio 2.2 i Follad.) 4. Product-σ-algebras ad complex measurable fuctios Let (Y, N ) be a measurable space ad f : X Y. The the σ-algebra o X geerated by f is give by σ(f) := σ{f (A) : A N }. I other words, σ(f) is the smallest σ-algebra o X that makes f measurable. (I fact {f (A) : A N } is a σ-algebra (prove this!), so σ(f) equals this set.) 22

23 More geerally, if F is a family of fuctios from X to Y, the σ(f) := σ{f (A) : f F, A N }. Now let (X, M ) ad (X 2, M 2 ) be two measurable spaces. The projectio maps π ad π 2 are give by i =, 2. π i : X X 2 X i, π i (x, x 2 ) = x i Defiitio 4.7 The product σ-algebra of M ad M 2 is give by More geerally M M 2 := σ(π, π 2 ) = σ{e E 2 : E i M i, i =, 2}. M = σ{π : =, 2,...} = σ{ E : E M } ad for a geeral idex set I M α = σ{π α : α I} α I = σ{ α I E α : E α Mα ad E α = X α for all but coutably may α}. Make sure that you uderstad the equalities i the defiitios. Propositio 4.8 Let (X, M) ad (Y α, N α ), α I, be measurable spaces. A map h = (f α ) α I : X α I Y α is (M, α I N α)-measurable if ad oly if each f α is (M, N α )-measurable. Proof. Sice f α = π α h, a compositio of two measurable maps, the oly if directio holds. O the other had, if all f α are measurable, the for ay α ad A N α, h (πα (A)) = (π α h) (A) = fα (A) M. Sice α N a is geerated by π α, α I, the if directio ow follows from Theorem

24 Propositio 4.9 B(R 2 ) = B(R) B(R). Proof. Let A = {(a, b ) (a 2, b 2 ) : a, b, a 2, b 2 Q}. Sice ay ope set i R 2 ca be writte as a coutable uio of sets i A, we have B(R 2 ) = σ(a). By defiitio B(R) B(R) cotais A ad hece B(R) B(R) B(R 2 ). O the other had, B(R) B(R) is geerated by π (A), A B(R), i =, 2. We have π (A) = A R, so it suffices to show that A R B(R 2 ) for every A B(R). (The similar statemet for π 2 is of course aalogous.) Sice A R is ope i R 2 wheever A is ope i R, this holds for all ope A. Hece, the family {A B(R) : A R B(R 2 )} cotais all ope sets, so if we ca show that it is also a σ-algebra, we are doe. This, however, is obvious. Two immediate corollaries follow. Corollary 4.0 B(C) = B(R) B(R). Corollary 4. A fuctio f : X C is (M, B(C))-measurable if ad oly if Rf ad If are both measurable. 4.2 Idepedet radom variables I the ext sectios (X, M, P) will be a probability space. Defiitio 4.2 Let I be a arbitrary set ad let E α, α I, be subclasses of M. We say that {E α } α I is idepedet if P( j J E j ) = j J P(E j ) for all fiite J I ad all E j E j, j J. The family of radom variables {ξ α } α I said to be idepedet if {σ(ξ α )} α I is idepedet. The family of evets {E α } α I, is said to be idepedet if {χ Eα } α I is idepedet. The give defiitio is completely geeral i terms of the idex set I. Although havig I ucoutable ca be useful sometimes, e.g. whe defiig Gaussia white oise, it will ot be so here, so i the sequel I will be either fiite or coutably ifiite. 24

25 Lemma 4.3 Assume that I, J M are two π-systems ad let N = σ(i) ad O = σ(j ). The {N, O} is idepedet if ad oly if {I, J } is idepedet. Proof. The oly if directio is trivial. The if directio will be proved by a two-step procedure. First fix arbitrary I I ad defie two measures o O by, for each B O, settig µ (B) = P(I B) µ 2 (B) = P(I)P(B) By hypothesis µ ad µ 2 agree o J ad µ (X) = µ 2 (X) <, so by the Uiqueess Theorem for measures, µ = µ 2. Next fix arbitrary B O ad defie two measures o N by settig, for each A N, µ 3 (A) = P(A B) µ 4 (A) = P(A)P(B). By what we just proved, µ 3 ad µ 4 agree o I. They are also fiite ad agree o X, ad are hece equal. This proves idepedece. Clearly Lemma 4.3 exteds to all fiite collectios of π-systems ad their geerated σ-algebras. Sice idepedece of a ifiite family of σ-algebras is equivalet to idepedece of fiite subfamilies, Lemma 4.3 also exteds to: Corollary 4.4 Let I, I 2,... M be π-systems. If {I, I 2,...} is idepedet, the also {σ(i ), σ(i 2 ),...} is idepedet. The followig two examples are importat. First observe the followig useful fact. Let f : X (Y, N ) ad suppose that E P(Y ) geerates N. The {f (E) : E E} geerates σ(f); this is so sice {E Y : f (E) σ(f)} is a σ-algebra, by the commutativity of iverse images ad basic set operatios. Example. Let ξ ad η be two radom variables. The {ξ (, a] : a R} ad {η (, b] : b R} are π-systems ad geerate σ(ξ) ad σ(η) respectively. Hece by Lemma 4.3 {ξ, η} is idepedet iff P(ξ (, a] η (, b]) = P(ξ (, a])p(η (, b]) for all a, b, i.e. if P(ξ a, η b) = P(ξ a)p(η b) for all a, b R. More geerally, by Corollary 4.4, {ξ, ξ 2,...} is idepedet iff P(ξ i a,..., ξ i a ) = P(ξ ik a k ) 25 k=

26 for all =, 2,..., all i <... < i ad all a,..., a R. For trivial reasos, {f(ξ), g(η)} are idepedet wheever ξ ad η are idepedet. (Check that you uderstad why!). Aalogously, if {{ξ, ξ 2,...}, {η, η 2,...}} is a idepedet pair of families of radom variables (i.e. a idepedet pair of R -valued radom variables; there is othig i the above defiitios that prevets us from cosiderig radom variables takig o values i a arbitrary space), the f(ξ, ξ 2,...) ad g(η, η 2,...) are idepedet. It is ituitively clear that if {ξ, ξ 2,...} is idepedet, the, if we extract two disjoit subfamilies, these two should make a idepedet pair of R -valued radom variables. The ext example shows that this is ideed the case. Example. Let ξ, ξ 2,... be idepedet radom variables ad let I ad J be two disjoit idex sets (i.e. I, J N ad I J = ). The {{ξ i a,..., {ξ i a } : =, 2,..., i <... < i, a,..., a R} is a π-system that geerates σ(ξ i : i I) ad the aalogous π-system geerates σ(ξ j : j J). By the previous example, the two π-systems are idepedet. Hece the collectios (ξ i : i I) ad (ξ j : j J) are idepedet, by Corollary 4.4. To relax our laguage a bit, let us take the statemet ξ, ξ 2,... are idepedet to mea that the family {ξ, ξ 2,...} is idepedet. Note that it is actually importat to spell this out, sice aother iterpretatio of the statemet could have bee that the radom variables are all pairwise idepedet. This, however, is a much weaker statemet. Cosider for example the three {0, }-valued radom variables ξ, ξ 2, ξ 3 give by P(ξ = 0, ξ 2 = 0, ξ 3 = ) = P(ξ = 0, ξ 2 =, ξ 3 = 0) = P(ξ =, ξ 2 = 0, ξ 3 = 0) = P(ξ =, ξ 2 =, ξ 3 = ) = /4, which are pairwise idepedet, but clearly ot idepedet sice ay of them is the xor sum of the other two. Hece, i the sequel, sayig that a set of radom variables are idepedet meas somethig stroger tha sayig that the same radom variables are pairwise idepedet. Theorem 4.5 (Borel-Catelli s Secod Lemma) Let E, E 2,... be a sequece of idepedet evets. If P(E ) =, the P(lim sup E ) =. Proof. Note that (lim sup E ) c = ( m m E ) c = m m E c 26

27 so by the cotiuity of measures, it suffices to show that P( m Ec ) = 0 for all m. This i tur follows from the followig computatios P( r E) c = lim P( E c r ) = lim r m = m ( P (E )) m r P(E) c m m e P(E) = e m P(E) = 0 Example. Let ξ, ξ 2,... be idepedet radom variables with expoetial() distributio, i.e. P(ξ > x) = e x, x 0. The ( ξ ) P log > a = e a log = a. Hece P(ξ > a log ) is fiite for a > ad ifiite for a. By the Borel-Catelli Lemmas, this etails that if a, the almost surely ξ > a log for ifiitely may, if a >, the almost surely ξ > a log for oly fiitely may. 4.3 Kolmogorov s 0--law Let ξ, ξ 2,... be idepedet radom variables. For each, let T = σ(ξ +, ξ +2,...) ad T = T. The σ-algebra T is called the tail-σ-algebra (w.r.t. ξ, ξ 2,...). A set E T is called a tail evet ad a radom variable which is T -measurable is called a tail fuctio of the ξ s. 27

28 A tail evet does ot, for ay, deped o the first of the ξ k s, so at a first glace it may seem that T should be trivial. This, however, would be the wrog impressio, sice T actually cotais a lot of iterestig evets. E.g. the evet {x X : lim ξ (x) exists} is a tail evet ad η = lim sup ( ) ξ k is a tail fuctio; they are T -measurable of every ad hece T -measurable. Kolmogorov s 0--law states that the probability for a tail evet must be either 0 or ad that ay tail fuctio must be a costat a.s. Theorem 4.6 (Kolmogorov s 0--law) Let ξ, ξ 2,... be idepedet radom variables. (i) If E T, the P(E) {0, }, (ii) If η is T -measurable, the there exists a costat c R such that η = c a.s. Proof. (i) Let F = σ(ξ,..., ξ ), =, 2,.... By the above example, F ad T are idepedet. Sice T T, F ad T are idepedet for every. Hece F ad T are idepedet. Sice F is a π-system, it follows that σ( F ) ad T are idepedet. However T σ{ξ, ξ 2,...) = σ( F ), so T is idepedet of itself. This meas that for each E T, P(E) = P(E E) = P(E) 2 which etails that P(E) is either 0 or. (ii) For all a R, P(η a) {0, } by (i). Let c = if{a : P(η a) = }. The ( P(η c) = P {x : η(x) c + ) } = ad ( P(η < c) = P {x : η(x) c ) } = 0. Example. (Mokey typig Shakespeare) Suppose that a mokey is typig uiform radom keys o a laptop. There are, say, N keys o the laptop. The collected works of Shakespeare (to be abbr. CWS) 28

29 comprises, say, M symbols. Let E be the evet that the mokey happes to type CWS evetually. Will E occur? If we let F be the evet that the mokey types CWS ifiitely may times, the by Kolmogorov s 0--law, P(F ) is 0 or. Let F be the evet that the mokey types CWS with the M + th to ( + )M th symols it types. The P(F ) = /N m, so P(F ) = ad hece P(lim sup F ) = by Borel-Catelli. Hece P(E) P(F ) P(lim sup F ) =. So the aswer is yes, the mokey will evetually type CWS (but, of course, very much provided that it has a ifiite life ad ca be persuaded to sped a ifiite amout of time at the laptop). Note that they key i the example was really Borel-Catelli s Secod Lemma ad that the iformatio provided by Kolmogorov s 0--law was oly that P(F ) {0, }. I the ext example, the 0--law plays a more vital role. Example. (Percolatio) Cosider the two-dimesioal iteger lattice, i.e. the graph obtaied by placig a vertex at each iteger poit (, k) i the Euclidea plae ad placig a edge betwee (, k) ad (m, j) if either = m ad k j = or k = j ad m =. Now remove edges at radom by lettig each edge be kept (or ope ) with probability p ad removed (or closed) with probability p, idepedetly of other edges. The resultig radom graph will of course a.s. fall ito (ifiitely may) coected compoets. However, will there be a ifiitely large coected compoet? Let E be the evet that a ifiite coected compoet exists. Let ξ i be the status, i.e. kept or removed, of edge umber i; here assume that edges are umbered accordig to their distace from the origi ad arbitrarily amog those edges that are equally far away. Now observe that E is a tail evet. This is so sice the presece or absece of ifiite compoets caot be chaged by chagig the status of the first edges o matter the value of. (For a outcome where ifiite compoets exist, chagig a fiite umber of edges ca chage the umber of such compoets, but ever chage presece/absece.) Hece, by Kolmogorov s 0--law, P(E) is 0 or. Determiig for what p we have P(E) = 0 ad for what p we have P(E) = Percolatio theory has its origis i the study of water flow through porous materials. The edges the represet microscopic chaels which may or may ot be ope for water flow. 29

30 is a differet story. This is of course a geeral fact about applicatios of Kolmogorov s 0--law; it tells us that a tail evet has probability 0 or, but ever tells which it is. However, kowig that P(E) is 0 or is still very helpful sice if we ca also show that P(E) > 0, the it follows immediately that P(E) =. I the percolatio settig of this example, cosider the probability that o vertex i the 2 2-box cetered at the origi, is part of a ifiite path of kept edges. It ca be show that this probability is bouded by (3( p)). (This is doe by boudig the umber of ways that the box ca be cut off from ifiity.) This is less tha for large eough if p > 2/3. Hece P(E) = for p > 2/3. O the other had, by similar coutig, it is easy to see that P(E) = 0 for p < /3. I fact, the critical probability for whe P(E) switches from 0 to is p = /2. This a cetral ad highly o-trivial fact of percolatio theory. (Whe p = /2, the P(E) = 0.) 5 Itegratio of oegative fuctios Defiig the Lebesgue itegral is a stepwise procedure. It starts with oegative simple fuctios. Defiitio 5. A fuctio φ : (X, M, µ) C is said to be simple if it is of the form φ(x) = z j χ Ej (x) for some, where z j C ad {E,..., E } is a partitio of X such that E j M for all j. Let L + (X, M, µ) deote the set of all M-measurable fuctios f : X [0, ]. Depedig o the level of risk for cofusio, we ofte use shorthad otatios such as L + (X), L + (M) or simply L +. Defiitio 5.2 Let φ = a jχ Ej, a j R + be simple. The the itegral of φ with respect to µ is give by X φ(x)dµ(x) := a j µ(e j ). 30

31 Example. Let (X, M, µ) = ([0, ], L, m) ad φ = χ Q [0,]. Sice Q [0, ] is coutable, it is measurable, so φ is a simple fuctio ad φdm = 0. Compare this with what happes if we try to calculate the Riema itegral of this fuctio. Sice the Riema itegral is defied i terms of approximatios of φ from above ad from below by simple fuctios that are costat itervals, we fid that the Riema itegral of φ is ot defied. Thus, there are fuctios defied o a iterval of the real lie which the Lebesgue itegral ca hadle, but which the Riema itegral caot. Later, we will also see that ay Riema itegrable fuctio o a iterval is Lebesgue itegrable ad that for such fuctios, the two methods give the same result. Alterative ad/or shorthad otatios for the itegral are φ(x)µ(dx), φdµ X ad φ. The represetatio of a simple fuctio as a fiite liear combiatio of characteristic fuctios is of course ot uique, but it is easy to see that differet represetatios give the same result, so the itegral is well-defied. For A M, write φdµ := φχ A dµ. A This is well-defied sice φχ A = a jχ A Ej + 0 χ Ac is simple. A few basic facts follow. Propositio 5.3 Let c R + ad φ = a jχ Ej, ψ = m b jχ Fj L + be simple fuctios. The (a) cφ = c φ, (b) (φ + ψ) = φ + ψ, (c) φ ψ φ ψ, (d) The map A φ, A M, is a measure. A Proof. Part (a) is trivial. For part (b) observe that φ + ψ = (a i + b j )χ Ei F j. i j Hece (φ + ψ) = i = i X (a i + b j )µ(e i F j ) = a i µ(e i F j ) + j i j j a i µ(e i ) + b j µ(f j ) = φ + ψ. j b j µ(e i F j ) i 3

32 For part (c) use the represetatios φ = i j a iχe i F j ad ψ = i j b jχe i F j. O each E i F j we have a i b j, so the result follows immediately from the defiitio. To prove part (d), it must be show that if A, A 2,... are disjoit sets i M, the k A k φ = k A k φ. We have k= A k φ = = ( a j µ E j ( k a j µ(e j A k ) = j= k= j= ) A k ) = j= k= k= a j µ(e j A k ) A k φ where the secod equality is coutable additivity of µ. The ext step is to defie itegrals of arbitrary fuctios i L + by approximatig with simple fuctios. The followig approximatio result tells us that it makes sese to do so. Theorem 5.4 (a) Let f L +. There are simple fuctios φ L + such that φ (x) f(x) for every x X. (b) Let f : X C be measurable. The there are simple fuctios φ such that φ φ 2... f ad φ f poitwise. Proof. I (a), let A j = {x : f(x) [j2, (j+)2 )}, j = 0,..., 2 ad let A 2 = {x : f(x) }. Sice f is measurable, all these sets are measurable, so lettig 2 φ (x) = j2 χ Aj (x) 0 gives φ s of the desired form. For (b), apply the proof of (a) to all four parts of f; see below for defiitios. I the light of Theorem 5.4, we make the followig defiitio. Defiitio 5.5 Let f L +. The { } f(x)dµ(x) := sup φ(x)dµ(x) : 0 φ f, φ simple. X X 32

33 For A M, A fdµ := fχ A dµ. It is obvious that if c R +, the cf = c f ad if f g, f, g L +, the f g. The ext result is oe of the key results i itegratio theory. Theorem 5.6 (The Mootoe Covergece Theorem) Assume that f, f L + ad f f poitwise. The f dµ fdµ. Proof. Sice {f } is icreasig, { f } is icreasig ad hece lim f exists (but may be equal to ). Sice f f for all, lim f f. Now pick a arbitrary simple fuctio φ L + such that φ f ad a arbitrary a (0, ). Sice f f poitwise, the sets A := {x : f (x) > aφ(x)} are icreasig i ad A = X. Sice the map A φ is a measure, it follows A from the cotiuity of measures that A φ φ. Therefore lim f a lim if φ = a φ. A Sice a was arbitrary, lettig a etails that lim f φ. The result ow follows from the defiitio of f. The first cosequece of the MCT is that the itegral is additive. Ideed, it is i fact coutably additive: Theorem 5.7 Let f L +. The ( f )dµ = f dµ. Proof. First cosider fiite additivity. By Theorem 5.4, there are simple oegative fuctios φ ad ψ such that φ f ad ψ f 2 poitwise. By the MCT ad Propositio 5.3, (f + f 2 ) = lim (φ + ψ ) = lim φ + lim ψ = f + f 2. Now fiite additivity follows by iductio. Sice N f f as N, aother applicatio of the MCT ow shows that N N ( f ) = lim ( f ) = lim f = f. N N 33

34 Corollary 5.8 Let f L +. The the map A fdµ, A M, is a measure. A The hypothesis i the MCT is that f f poitwise. This ca be relaxed a bit; it suffices to have f f a.e. To see this, first observe that if f = 0, the we ca fid simple φ L + with φ f poitwise ad φ = 0. However, sice φ is simple, this trivially meas that φ = 0 a.e. Now if x is a poit such that f(x) > 0, the φ (x) > 0 for all sufficietly large. Hece ( ) µ{x : f(x) > 0} µ {x : φ (x) > 0} = 0. I summary Propositio 5.9 Let f L +. The fdµ = 0 if ad oly if f = 0 a.e. Suppose ow that f f a.e. ad let E = {x : f (x) f(x)}. The f χ E fχ E poitwise so by the MCT, f χ E fχ E. Sice f fχ E L + ad f fχ E = 0 a.e., Propositio 5.9 implies that fχ E = f. From the same argumet, f χ E = f. Puttig these facts together gives f f. (This result is Corollary 2.7 i Follad.) The MCT states that if f L + ad f f a.e., the f f, but what about whe f f without beig icreasig i? Does this also imply f f? The aswer is o, as the followig example shows. Let (X, M, µ) = ([0, ], L, m) ad f (x) = χ [0,/] (x). The f 0 a.e. (but ot poitwise, sice f (0) ), but f = for every. Hece some further assumptio is eeded to guaratee that f f a.e. etails that f f. Such a coditio will be give i the Domiated Covergece Theorem below. Before that, we will exted the itegral from oegative real fuctios to geeral complex fuctios. First however, we fiish the preset sectio with the importat Fatou s Lemma ad a ote o σ-fiiteess. Theorem 5.0 (Fatou s Lemma) If f L +, =, 2,..., the (lim if f )dµ lim if f dµ. 34

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014. Product measures, Toelli s ad Fubii s theorems For use i MAT3400/4400, autum 2014 Nadia S. Larse Versio of 13 October 2014. 1. Costructio of the product measure The purpose of these otes is to preset the

More information

Measure and Measurable Functions

Measure and Measurable Functions 3 Measure ad Measurable Fuctios 3.1 Measure o a Arbitrary σ-algebra Recall from Chapter 2 that the set M of all Lebesgue measurable sets has the followig properties: R M, E M implies E c M, E M for N implies

More information

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence Chapter 3 Strog covergece As poited out i the Chapter 2, there are multiple ways to defie the otio of covergece of a sequece of radom variables. That chapter defied covergece i probability, covergece i

More information

The Borel hierarchy classifies subsets of the reals by their topological complexity. Another approach is to classify them by size.

The Borel hierarchy classifies subsets of the reals by their topological complexity. Another approach is to classify them by size. Lecture 7: Measure ad Category The Borel hierarchy classifies subsets of the reals by their topological complexity. Aother approach is to classify them by size. Filters ad Ideals The most commo measure

More information

Lecture 3 The Lebesgue Integral

Lecture 3 The Lebesgue Integral Lecture 3: The Lebesgue Itegral 1 of 14 Course: Theory of Probability I Term: Fall 2013 Istructor: Gorda Zitkovic Lecture 3 The Lebesgue Itegral The costructio of the itegral Uless expressly specified

More information

Convergence of random variables. (telegram style notes) P.J.C. Spreij

Convergence of random variables. (telegram style notes) P.J.C. Spreij Covergece of radom variables (telegram style otes).j.c. Spreij this versio: September 6, 2005 Itroductio As we kow, radom variables are by defiitio measurable fuctios o some uderlyig measurable space

More information

If a subset E of R contains no open interval, is it of zero measure? For instance, is the set of irrationals in [0, 1] is of measure zero?

If a subset E of R contains no open interval, is it of zero measure? For instance, is the set of irrationals in [0, 1] is of measure zero? 2 Lebesgue Measure I Chapter 1 we defied the cocept of a set of measure zero, ad we have observed that every coutable set is of measure zero. Here are some atural questios: If a subset E of R cotais a

More information

The Boolean Ring of Intervals

The Boolean Ring of Intervals MATH 532 Lebesgue Measure Dr. Neal, WKU We ow shall apply the results obtaied about outer measure to the legth measure o the real lie. Throughout, our space X will be the set of real umbers R. Whe ecessary,

More information

Axioms of Measure Theory

Axioms of Measure Theory MATH 532 Axioms of Measure Theory Dr. Neal, WKU I. The Space Throughout the course, we shall let X deote a geeric o-empty set. I geeral, we shall ot assume that ay algebraic structure exists o X so that

More information

Integrable Functions. { f n } is called a determining sequence for f. If f is integrable with respect to, then f d does exist as a finite real number

Integrable Functions. { f n } is called a determining sequence for f. If f is integrable with respect to, then f d does exist as a finite real number MATH 532 Itegrable Fuctios Dr. Neal, WKU We ow shall defie what it meas for a measurable fuctio to be itegrable, show that all itegral properties of simple fuctios still hold, ad the give some coditios

More information

Advanced Stochastic Processes.

Advanced Stochastic Processes. Advaced Stochastic Processes. David Gamarik LECTURE 2 Radom variables ad measurable fuctios. Strog Law of Large Numbers (SLLN). Scary stuff cotiued... Outlie of Lecture Radom variables ad measurable fuctios.

More information

Infinite Sequences and Series

Infinite Sequences and Series Chapter 6 Ifiite Sequeces ad Series 6.1 Ifiite Sequeces 6.1.1 Elemetary Cocepts Simply speakig, a sequece is a ordered list of umbers writte: {a 1, a 2, a 3,...a, a +1,...} where the elemets a i represet

More information

Lecture Notes for Analysis Class

Lecture Notes for Analysis Class Lecture Notes for Aalysis Class Topological Spaces A topology for a set X is a collectio T of subsets of X such that: (a) X ad the empty set are i T (b) Uios of elemets of T are i T (c) Fiite itersectios

More information

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4. 4. BASES I BAACH SPACES 39 4. BASES I BAACH SPACES Sice a Baach space X is a vector space, it must possess a Hamel, or vector space, basis, i.e., a subset {x γ } γ Γ whose fiite liear spa is all of X ad

More information

Introduction to Probability. Ariel Yadin. Lecture 7

Introduction to Probability. Ariel Yadin. Lecture 7 Itroductio to Probability Ariel Yadi Lecture 7 1. Idepedece Revisited 1.1. Some remiders. Let (Ω, F, P) be a probability space. Give a collectio of subsets K F, recall that the σ-algebra geerated by K,

More information

Lecture 3 : Random variables and their distributions

Lecture 3 : Random variables and their distributions Lecture 3 : Radom variables ad their distributios 3.1 Radom variables Let (Ω, F) ad (S, S) be two measurable spaces. A map X : Ω S is measurable or a radom variable (deoted r.v.) if X 1 (A) {ω : X(ω) A}

More information

FUNDAMENTALS OF REAL ANALYSIS by. V.1. Product measures

FUNDAMENTALS OF REAL ANALYSIS by. V.1. Product measures FUNDAMENTALS OF REAL ANALSIS by Doğa Çömez V. PRODUCT MEASURE SPACES V.1. Product measures Let (, A, µ) ad (, B, ν) be two measure spaces. I this sectio we will costruct a product measure µ ν o that coicides

More information

Chapter 0. Review of set theory. 0.1 Sets

Chapter 0. Review of set theory. 0.1 Sets Chapter 0 Review of set theory Set theory plays a cetral role i the theory of probability. Thus, we will ope this course with a quick review of those otios of set theory which will be used repeatedly.

More information

Singular Continuous Measures by Michael Pejic 5/14/10

Singular Continuous Measures by Michael Pejic 5/14/10 Sigular Cotiuous Measures by Michael Peic 5/4/0 Prelimiaries Give a set X, a σ-algebra o X is a collectio of subsets of X that cotais X ad ad is closed uder complemetatio ad coutable uios hece, coutable

More information

Lecture 2. The Lovász Local Lemma

Lecture 2. The Lovász Local Lemma Staford Uiversity Sprig 208 Math 233A: No-costructive methods i combiatorics Istructor: Ja Vodrák Lecture date: Jauary 0, 208 Origial scribe: Apoorva Khare Lecture 2. The Lovász Local Lemma 2. Itroductio

More information

6.3 Testing Series With Positive Terms

6.3 Testing Series With Positive Terms 6.3. TESTING SERIES WITH POSITIVE TERMS 307 6.3 Testig Series With Positive Terms 6.3. Review of what is kow up to ow I theory, testig a series a i for covergece amouts to fidig the i= sequece of partial

More information

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function.

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function. MATH 532 Measurable Fuctios Dr. Neal, WKU Throughout, let ( X, F, µ) be a measure space ad let (!, F, P ) deote the special case of a probability space. We shall ow begi to study real-valued fuctios defied

More information

Solution. 1 Solutions of Homework 1. Sangchul Lee. October 27, Problem 1.1

Solution. 1 Solutions of Homework 1. Sangchul Lee. October 27, Problem 1.1 Solutio Sagchul Lee October 7, 017 1 Solutios of Homework 1 Problem 1.1 Let Ω,F,P) be a probability space. Show that if {A : N} F such that A := lim A exists, the PA) = lim PA ). Proof. Usig the cotiuity

More information

Chapter 6 Infinite Series

Chapter 6 Infinite Series Chapter 6 Ifiite Series I the previous chapter we cosidered itegrals which were improper i the sese that the iterval of itegratio was ubouded. I this chapter we are goig to discuss a topic which is somewhat

More information

7.1 Convergence of sequences of random variables

7.1 Convergence of sequences of random variables Chapter 7 Limit Theorems Throughout this sectio we will assume a probability space (, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite

More information

Chapter IV Integration Theory

Chapter IV Integration Theory Chapter IV Itegratio Theory Lectures 32-33 1. Costructio of the itegral I this sectio we costruct the abstract itegral. As a matter of termiology, we defie a measure space as beig a triple (, A, µ), where

More information

(A sequence also can be thought of as the list of function values attained for a function f :ℵ X, where f (n) = x n for n 1.) x 1 x N +k x N +4 x 3

(A sequence also can be thought of as the list of function values attained for a function f :ℵ X, where f (n) = x n for n 1.) x 1 x N +k x N +4 x 3 MATH 337 Sequeces Dr. Neal, WKU Let X be a metric space with distace fuctio d. We shall defie the geeral cocept of sequece ad limit i a metric space, the apply the results i particular to some special

More information

6 Integers Modulo n. integer k can be written as k = qn + r, with q,r, 0 r b. So any integer.

6 Integers Modulo n. integer k can be written as k = qn + r, with q,r, 0 r b. So any integer. 6 Itegers Modulo I Example 2.3(e), we have defied the cogruece of two itegers a,b with respect to a modulus. Let us recall that a b (mod ) meas a b. We have proved that cogruece is a equivalece relatio

More information

Advanced Analysis. Min Yan Department of Mathematics Hong Kong University of Science and Technology

Advanced Analysis. Min Yan Department of Mathematics Hong Kong University of Science and Technology Advaced Aalysis Mi Ya Departmet of Mathematics Hog Kog Uiversity of Sciece ad Techology September 3, 009 Cotets Limit ad Cotiuity 7 Limit of Sequece 8 Defiitio 8 Property 3 3 Ifiity ad Ifiitesimal 8 4

More information

Part II Probability and Measure

Part II Probability and Measure Part II Probability ad Measure Based o lectures by J. Miller Notes take by Dexter Chua Michaelmas 2016 These otes are ot edorsed by the lecturers, ad I have modified them (ofte sigificatly) after lectures.

More information

Sequences. Notation. Convergence of a Sequence

Sequences. Notation. Convergence of a Sequence Sequeces A sequece is essetially just a list. Defiitio (Sequece of Real Numbers). A sequece of real umbers is a fuctio Z (, ) R for some real umber. Do t let the descriptio of the domai cofuse you; it

More information

MAT1026 Calculus II Basic Convergence Tests for Series

MAT1026 Calculus II Basic Convergence Tests for Series MAT026 Calculus II Basic Covergece Tests for Series Egi MERMUT 202.03.08 Dokuz Eylül Uiversity Faculty of Sciece Departmet of Mathematics İzmir/TURKEY Cotets Mootoe Covergece Theorem 2 2 Series of Real

More information

Sequences and Series of Functions

Sequences and Series of Functions Chapter 6 Sequeces ad Series of Fuctios 6.1. Covergece of a Sequece of Fuctios Poitwise Covergece. Defiitio 6.1. Let, for each N, fuctio f : A R be defied. If, for each x A, the sequece (f (x)) coverges

More information

MA131 - Analysis 1. Workbook 2 Sequences I

MA131 - Analysis 1. Workbook 2 Sequences I MA3 - Aalysis Workbook 2 Sequeces I Autum 203 Cotets 2 Sequeces I 2. Itroductio.............................. 2.2 Icreasig ad Decreasig Sequeces................ 2 2.3 Bouded Sequeces..........................

More information

MA131 - Analysis 1. Workbook 3 Sequences II

MA131 - Analysis 1. Workbook 3 Sequences II MA3 - Aalysis Workbook 3 Sequeces II Autum 2004 Cotets 2.8 Coverget Sequeces........................ 2.9 Algebra of Limits......................... 2 2.0 Further Useful Results........................

More information

Lecture 2 Measures. Measure spaces. µ(a n ), for n N, and pairwise disjoint A 1,..., A n, we say that the. (S, S) is called

Lecture 2 Measures. Measure spaces. µ(a n ), for n N, and pairwise disjoint A 1,..., A n, we say that the. (S, S) is called Lecture 2: Measures 1 of 17 Course: Theory of Probability I Term: Fall 2013 Istructor: Gorda Zitkovic Lecture 2 Measures Measure spaces Defiitio 2.1 (Measure). Let (S, S) be a measurable space. A mappig

More information

Fall 2013 MTH431/531 Real analysis Section Notes

Fall 2013 MTH431/531 Real analysis Section Notes Fall 013 MTH431/531 Real aalysis Sectio 8.1-8. Notes Yi Su 013.11.1 1. Defiitio of uiform covergece. We look at a sequece of fuctios f (x) ad study the coverget property. Notice we have two parameters

More information

Sequences I. Chapter Introduction

Sequences I. Chapter Introduction Chapter 2 Sequeces I 2. Itroductio A sequece is a list of umbers i a defiite order so that we kow which umber is i the first place, which umber is i the secod place ad, for ay atural umber, we kow which

More information

Math 299 Supplement: Real Analysis Nov 2013

Math 299 Supplement: Real Analysis Nov 2013 Math 299 Supplemet: Real Aalysis Nov 203 Algebra Axioms. I Real Aalysis, we work withi the axiomatic system of real umbers: the set R alog with the additio ad multiplicatio operatios +,, ad the iequality

More information

Recitation 4: Lagrange Multipliers and Integration

Recitation 4: Lagrange Multipliers and Integration Math 1c TA: Padraic Bartlett Recitatio 4: Lagrage Multipliers ad Itegratio Week 4 Caltech 211 1 Radom Questio Hey! So, this radom questio is pretty tightly tied to today s lecture ad the cocept of cotet

More information

5 Many points of continuity

5 Many points of continuity Tel Aviv Uiversity, 2013 Measure ad category 40 5 May poits of cotiuity 5a Discotiuous derivatives.............. 40 5b Baire class 1 (classical)............... 42 5c Baire class 1 (moder)...............

More information

Math 61CM - Solutions to homework 3

Math 61CM - Solutions to homework 3 Math 6CM - Solutios to homework 3 Cédric De Groote October 2 th, 208 Problem : Let F be a field, m 0 a fixed oegative iteger ad let V = {a 0 + a x + + a m x m a 0,, a m F} be the vector space cosistig

More information

An Introduction to Randomized Algorithms

An Introduction to Randomized Algorithms A Itroductio to Radomized Algorithms The focus of this lecture is to study a radomized algorithm for quick sort, aalyze it usig probabilistic recurrece relatios, ad also provide more geeral tools for aalysis

More information

Distribution of Random Samples & Limit theorems

Distribution of Random Samples & Limit theorems STAT/MATH 395 A - PROBABILITY II UW Witer Quarter 2017 Néhémy Lim Distributio of Radom Samples & Limit theorems 1 Distributio of i.i.d. Samples Motivatig example. Assume that the goal of a study is to

More information

Real and Complex Analysis, 3rd Edition, W.Rudin

Real and Complex Analysis, 3rd Edition, W.Rudin Real ad Complex Aalysis, 3rd ditio, W.Rudi Chapter 6 Complex Measures Yug-Hsiag Huag 206/08/22. Let ν be a complex measure o (X, M ). If M, defie { } µ () = sup ν( j ) : N,, 2, disjoit, = j { } ν () =

More information

Seunghee Ye Ma 8: Week 5 Oct 28

Seunghee Ye Ma 8: Week 5 Oct 28 Week 5 Summary I Sectio, we go over the Mea Value Theorem ad its applicatios. I Sectio 2, we will recap what we have covered so far this term. Topics Page Mea Value Theorem. Applicatios of the Mea Value

More information

1 Introduction. 1.1 Notation and Terminology

1 Introduction. 1.1 Notation and Terminology 1 Itroductio You have already leared some cocepts of calculus such as limit of a sequece, limit, cotiuity, derivative, ad itegral of a fuctio etc. Real Aalysis studies them more rigorously usig a laguage

More information

Lesson 10: Limits and Continuity

Lesson 10: Limits and Continuity www.scimsacademy.com Lesso 10: Limits ad Cotiuity SCIMS Academy 1 Limit of a fuctio The cocept of limit of a fuctio is cetral to all other cocepts i calculus (like cotiuity, derivative, defiite itegrals

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS MASSACHUSTTS INSTITUT OF TCHNOLOGY 6.436J/5.085J Fall 2008 Lecture 9 /7/2008 LAWS OF LARG NUMBRS II Cotets. The strog law of large umbers 2. The Cheroff boud TH STRONG LAW OF LARG NUMBRS While the weak

More information

FUNDAMENTALS OF REAL ANALYSIS by

FUNDAMENTALS OF REAL ANALYSIS by FUNDAMENTALS OF REAL ANALYSIS by Doğa Çömez Backgroud: All of Math 450/1 material. Namely: basic set theory, relatios ad PMI, structure of N, Z, Q ad R, basic properties of (cotiuous ad differetiable)

More information

s = and t = with C ij = A i B j F. (i) Note that cs = M and so ca i µ(a i ) I E (cs) = = c a i µ(a i ) = ci E (s). (ii) Note that s + t = M and so

s = and t = with C ij = A i B j F. (i) Note that cs = M and so ca i µ(a i ) I E (cs) = = c a i µ(a i ) = ci E (s). (ii) Note that s + t = M and so 3 From the otes we see that the parts of Theorem 4. that cocer us are: Let s ad t be two simple o-egative F-measurable fuctios o X, F, µ ad E, F F. The i I E cs ci E s for all c R, ii I E s + t I E s +

More information

7.1 Convergence of sequences of random variables

7.1 Convergence of sequences of random variables Chapter 7 Limit theorems Throughout this sectio we will assume a probability space (Ω, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite

More information

Empirical Processes: Glivenko Cantelli Theorems

Empirical Processes: Glivenko Cantelli Theorems Empirical Processes: Gliveko Catelli Theorems Mouliath Baerjee Jue 6, 200 Gliveko Catelli classes of fuctios The reader is referred to Chapter.6 of Weller s Torgo otes, Chapter??? of VDVW ad Chapter 8.3

More information

Math 113 Exam 3 Practice

Math 113 Exam 3 Practice Math Exam Practice Exam will cover.-.9. This sheet has three sectios. The first sectio will remid you about techiques ad formulas that you should kow. The secod gives a umber of practice questios for you

More information

Discrete Mathematics for CS Spring 2007 Luca Trevisan Lecture 22

Discrete Mathematics for CS Spring 2007 Luca Trevisan Lecture 22 CS 70 Discrete Mathematics for CS Sprig 2007 Luca Trevisa Lecture 22 Aother Importat Distributio The Geometric Distributio Questio: A biased coi with Heads probability p is tossed repeatedly util the first

More information

Math 220A Fall 2007 Homework #2. Will Garner A

Math 220A Fall 2007 Homework #2. Will Garner A Math 0A Fall 007 Homewor # Will Garer Pg 3 #: Show that {cis : a o-egative iteger} is dese i T = {z œ : z = }. For which values of q is {cis(q): a o-egative iteger} dese i T? To show that {cis : a o-egative

More information

Lecture 19: Convergence

Lecture 19: Convergence Lecture 19: Covergece Asymptotic approach I statistical aalysis or iferece, a key to the success of fidig a good procedure is beig able to fid some momets ad/or distributios of various statistics. I may

More information

Probability for mathematicians INDEPENDENCE TAU

Probability for mathematicians INDEPENDENCE TAU Probability for mathematicias INDEPENDENCE TAU 2013 28 Cotets 3 Ifiite idepedet sequeces 28 3a Idepedet evets........................ 28 3b Idepedet radom variables.................. 33 3 Ifiite idepedet

More information

A Proof of Birkhoff s Ergodic Theorem

A Proof of Birkhoff s Ergodic Theorem A Proof of Birkhoff s Ergodic Theorem Joseph Hora September 2, 205 Itroductio I Fall 203, I was learig the basics of ergodic theory, ad I came across this theorem. Oe of my supervisors, Athoy Quas, showed

More information

Here are some examples of algebras: F α = A(G). Also, if A, B A(G) then A, B F α. F α = A(G). In other words, A(G)

Here are some examples of algebras: F α = A(G). Also, if A, B A(G) then A, B F α. F α = A(G). In other words, A(G) MATH 529 Probability Axioms Here we shall use the geeral axioms of a probability measure to derive several importat results ivolvig probabilities of uios ad itersectios. Some more advaced results will

More information

A gentle introduction to Measure Theory

A gentle introduction to Measure Theory A getle itroductio to Measure Theory Gaurav Chadalia Departmet of Computer ciece ad Egieerig UNY - Uiversity at Buffalo, Buffalo, NY gsc4@buffalo.edu March 12, 2007 Abstract This ote itroduces the basic

More information

Probability and Random Processes

Probability and Random Processes Probability ad Radom Processes Lecture 5 Probability ad radom variables The law of large umbers Mikael Skoglud, Probability ad radom processes 1/21 Why Measure Theoretic Probability? Stroger limit theorems

More information

Introduction to Probability. Ariel Yadin. Lecture 2

Introduction to Probability. Ariel Yadin. Lecture 2 Itroductio to Probability Ariel Yadi Lecture 2 1. Discrete Probability Spaces Discrete probability spaces are those for which the sample space is coutable. We have already see that i this case we ca take

More information

Math 2784 (or 2794W) University of Connecticut

Math 2784 (or 2794W) University of Connecticut ORDERS OF GROWTH PAT SMITH Math 2784 (or 2794W) Uiversity of Coecticut Date: Mar. 2, 22. ORDERS OF GROWTH. Itroductio Gaiig a ituitive feel for the relative growth of fuctios is importat if you really

More information

MAS111 Convergence and Continuity

MAS111 Convergence and Continuity MAS Covergece ad Cotiuity Key Objectives At the ed of the course, studets should kow the followig topics ad be able to apply the basic priciples ad theorems therei to solvig various problems cocerig covergece

More information

Math 341 Lecture #31 6.5: Power Series

Math 341 Lecture #31 6.5: Power Series Math 341 Lecture #31 6.5: Power Series We ow tur our attetio to a particular kid of series of fuctios, amely, power series, f(x = a x = a 0 + a 1 x + a 2 x 2 + where a R for all N. I terms of a series

More information

CHAPTER I: Vector Spaces

CHAPTER I: Vector Spaces CHAPTER I: Vector Spaces Sectio 1: Itroductio ad Examples This first chapter is largely a review of topics you probably saw i your liear algebra course. So why cover it? (1) Not everyoe remembers everythig

More information

Mathematical Methods for Physics and Engineering

Mathematical Methods for Physics and Engineering Mathematical Methods for Physics ad Egieerig Lecture otes Sergei V. Shabaov Departmet of Mathematics, Uiversity of Florida, Gaiesville, FL 326 USA CHAPTER The theory of covergece. Numerical sequeces..

More information

Beurling Integers: Part 2

Beurling Integers: Part 2 Beurlig Itegers: Part 2 Isomorphisms Devi Platt July 11, 2015 1 Prime Factorizatio Sequeces I the last article we itroduced the Beurlig geeralized itegers, which ca be represeted as a sequece of real umbers

More information

MATH301 Real Analysis (2008 Fall) Tutorial Note #7. k=1 f k (x) converges pointwise to S(x) on E if and

MATH301 Real Analysis (2008 Fall) Tutorial Note #7. k=1 f k (x) converges pointwise to S(x) on E if and MATH01 Real Aalysis (2008 Fall) Tutorial Note #7 Sequece ad Series of fuctio 1: Poitwise Covergece ad Uiform Covergece Part I: Poitwise Covergece Defiitio of poitwise covergece: A sequece of fuctios f

More information

Introductory Ergodic Theory and the Birkhoff Ergodic Theorem

Introductory Ergodic Theory and the Birkhoff Ergodic Theorem Itroductory Ergodic Theory ad the Birkhoff Ergodic Theorem James Pikerto Jauary 14, 2014 I this expositio we ll cover a itroductio to ergodic theory. Specifically, the Birkhoff Mea Theorem. Ergodic theory

More information

Lecture 3: August 31

Lecture 3: August 31 36-705: Itermediate Statistics Fall 018 Lecturer: Siva Balakrisha Lecture 3: August 31 This lecture will be mostly a summary of other useful expoetial tail bouds We will ot prove ay of these i lecture,

More information

University of Colorado Denver Dept. Math. & Stat. Sciences Applied Analysis Preliminary Exam 13 January 2012, 10:00 am 2:00 pm. Good luck!

University of Colorado Denver Dept. Math. & Stat. Sciences Applied Analysis Preliminary Exam 13 January 2012, 10:00 am 2:00 pm. Good luck! Uiversity of Colorado Dever Dept. Math. & Stat. Scieces Applied Aalysis Prelimiary Exam 13 Jauary 01, 10:00 am :00 pm Name: The proctor will let you read the followig coditios before the exam begis, ad

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 Fuctioal Law of Large Numbers. Costructio of the Wieer Measure Cotet. 1. Additioal techical results o weak covergece

More information

The Pointwise Ergodic Theorem and its Applications

The Pointwise Ergodic Theorem and its Applications The Poitwise Ergodic Theorem ad its Applicatios Itroductio Peter Oberly 11/9/2018 Algebra has homomorphisms ad topology has cotiuous maps; i these otes we explore the structure preservig maps for measure

More information

Math 216A Notes, Week 5

Math 216A Notes, Week 5 Math 6A Notes, Week 5 Scribe: Ayastassia Sebolt Disclaimer: These otes are ot early as polished (ad quite possibly ot early as correct) as a published paper. Please use them at your ow risk.. Thresholds

More information

Math 451: Euclidean and Non-Euclidean Geometry MWF 3pm, Gasson 204 Homework 3 Solutions

Math 451: Euclidean and Non-Euclidean Geometry MWF 3pm, Gasson 204 Homework 3 Solutions Math 451: Euclidea ad No-Euclidea Geometry MWF 3pm, Gasso 204 Homework 3 Solutios Exercises from 1.4 ad 1.5 of the otes: 4.3, 4.10, 4.12, 4.14, 4.15, 5.3, 5.4, 5.5 Exercise 4.3. Explai why Hp, q) = {x

More information

Introduction to Functional Analysis

Introduction to Functional Analysis MIT OpeCourseWare http://ocw.mit.edu 18.10 Itroductio to Fuctioal Aalysis Sprig 009 For iformatio about citig these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. LECTURE OTES FOR 18.10,

More information

Notes 27 : Brownian motion: path properties

Notes 27 : Brownian motion: path properties Notes 27 : Browia motio: path properties Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces:[Dur10, Sectio 8.1], [MP10, Sectio 1.1, 1.2, 1.3]. Recall: DEF 27.1 (Covariace) Let X = (X

More information

Sequences A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence

Sequences A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence Sequeces A sequece of umbers is a fuctio whose domai is the positive itegers. We ca see that the sequece 1, 1, 2, 2, 3, 3,... is a fuctio from the positive itegers whe we write the first sequece elemet

More information

Machine Learning Theory Tübingen University, WS 2016/2017 Lecture 12

Machine Learning Theory Tübingen University, WS 2016/2017 Lecture 12 Machie Learig Theory Tübige Uiversity, WS 06/07 Lecture Tolstikhi Ilya Abstract I this lecture we derive risk bouds for kerel methods. We will start by showig that Soft Margi kerel SVM correspods to miimizig

More information

Math Solutions to homework 6

Math Solutions to homework 6 Math 175 - Solutios to homework 6 Cédric De Groote November 16, 2017 Problem 1 (8.11 i the book): Let K be a compact Hermitia operator o a Hilbert space H ad let the kerel of K be {0}. Show that there

More information

MATH 413 FINAL EXAM. f(x) f(y) M x y. x + 1 n

MATH 413 FINAL EXAM. f(x) f(y) M x y. x + 1 n MATH 43 FINAL EXAM Math 43 fial exam, 3 May 28. The exam starts at 9: am ad you have 5 miutes. No textbooks or calculators may be used durig the exam. This exam is prited o both sides of the paper. Good

More information

The Growth of Functions. Theoretical Supplement

The Growth of Functions. Theoretical Supplement The Growth of Fuctios Theoretical Supplemet The Triagle Iequality The triagle iequality is a algebraic tool that is ofte useful i maipulatig absolute values of fuctios. The triagle iequality says that

More information

What is Probability?

What is Probability? Quatificatio of ucertaity. What is Probability? Mathematical model for thigs that occur radomly. Radom ot haphazard, do t kow what will happe o ay oe experimet, but has a log ru order. The cocept of probability

More information

Sets and Probabilistic Models

Sets and Probabilistic Models ets ad Probabilistic Models Berli Che Departmet of Computer ciece & Iformatio Egieerig Natioal Taiwa Normal Uiversity Referece: - D. P. Bertsekas, J. N. Tsitsiklis, Itroductio to Probability, ectios 1.1-1.2

More information

Exercises 1 Sets and functions

Exercises 1 Sets and functions Exercises 1 Sets ad fuctios HU Wei September 6, 018 1 Basics Set theory ca be made much more rigorous ad built upo a set of Axioms. But we will cover oly some heuristic ideas. For those iterested studets,

More information

Lecture XVI - Lifting of paths and homotopies

Lecture XVI - Lifting of paths and homotopies Lecture XVI - Liftig of paths ad homotopies I the last lecture we discussed the liftig problem ad proved that the lift if it exists is uiquely determied by its value at oe poit. I this lecture we shall

More information

In number theory we will generally be working with integers, though occasionally fractions and irrationals will come into play.

In number theory we will generally be working with integers, though occasionally fractions and irrationals will come into play. Number Theory Math 5840 otes. Sectio 1: Axioms. I umber theory we will geerally be workig with itegers, though occasioally fractios ad irratioals will come ito play. Notatio: Z deotes the set of all itegers

More information

McGill University Math 354: Honors Analysis 3 Fall 2012 Solutions to selected problems

McGill University Math 354: Honors Analysis 3 Fall 2012 Solutions to selected problems McGill Uiversity Math 354: Hoors Aalysis 3 Fall 212 Assigmet 3 Solutios to selected problems Problem 1. Lipschitz fuctios. Let Lip K be the set of all fuctios cotiuous fuctios o [, 1] satisfyig a Lipschitz

More information

Application to Random Graphs

Application to Random Graphs A Applicatio to Radom Graphs Brachig processes have a umber of iterestig ad importat applicatios. We shall cosider oe of the most famous of them, the Erdős-Réyi radom graph theory. 1 Defiitio A.1. Let

More information

Entropy Rates and Asymptotic Equipartition

Entropy Rates and Asymptotic Equipartition Chapter 29 Etropy Rates ad Asymptotic Equipartitio Sectio 29. itroduces the etropy rate the asymptotic etropy per time-step of a stochastic process ad shows that it is well-defied; ad similarly for iformatio,

More information

A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence

A sequence of numbers is a function whose domain is the positive integers. We can see that the sequence Sequeces A sequece of umbers is a fuctio whose domai is the positive itegers. We ca see that the sequece,, 2, 2, 3, 3,... is a fuctio from the positive itegers whe we write the first sequece elemet as

More information

Lecture 2: April 3, 2013

Lecture 2: April 3, 2013 TTIC/CMSC 350 Mathematical Toolkit Sprig 203 Madhur Tulsiai Lecture 2: April 3, 203 Scribe: Shubhedu Trivedi Coi tosses cotiued We retur to the coi tossig example from the last lecture agai: Example. Give,

More information

2.1. The Algebraic and Order Properties of R Definition. A binary operation on a set F is a function B : F F! F.

2.1. The Algebraic and Order Properties of R Definition. A binary operation on a set F is a function B : F F! F. CHAPTER 2 The Real Numbers 2.. The Algebraic ad Order Properties of R Defiitio. A biary operatio o a set F is a fuctio B : F F! F. For the biary operatios of + ad, we replace B(a, b) by a + b ad a b, respectively.

More information

2.4 Sequences, Sequences of Sets

2.4 Sequences, Sequences of Sets 72 CHAPTER 2. IMPORTANT PROPERTIES OF R 2.4 Sequeces, Sequeces of Sets 2.4.1 Sequeces Defiitio 2.4.1 (sequece Let S R. 1. A sequece i S is a fuctio f : K S where K = { N : 0 for some 0 N}. 2. For each

More information

} is said to be a Cauchy sequence provided the following condition is true.

} is said to be a Cauchy sequence provided the following condition is true. Math 4200, Fial Exam Review I. Itroductio to Proofs 1. Prove the Pythagorea theorem. 2. Show that 43 is a irratioal umber. II. Itroductio to Logic 1. Costruct a truth table for the statemet ( p ad ~ r

More information

Math F215: Induction April 7, 2013

Math F215: Induction April 7, 2013 Math F25: Iductio April 7, 203 Iductio is used to prove that a collectio of statemets P(k) depedig o k N are all true. A statemet is simply a mathematical phrase that must be either true or false. Here

More information

Random Variables, Sampling and Estimation

Random Variables, Sampling and Estimation Chapter 1 Radom Variables, Samplig ad Estimatio 1.1 Itroductio This chapter will cover the most importat basic statistical theory you eed i order to uderstad the ecoometric material that will be comig

More information

lim za n n = z lim a n n.

lim za n n = z lim a n n. Lecture 6 Sequeces ad Series Defiitio 1 By a sequece i a set A, we mea a mappig f : N A. It is customary to deote a sequece f by {s } where, s := f(). A sequece {z } of (complex) umbers is said to be coverget

More information