Best Constants in Kahane-Khintchine Inequalities in Orlicz Spaces
|
|
- Jeffery Cornelius Haynes
- 6 years ago
- Views:
Transcription
1 J. Multivariate Aal. Vol. 45, No., 993, (83-6) Prerit Ser. No. 0, 99, Math. Ist. Aarhus Best ostats i Kahae-Khitchie Iequalities i Orlicz Saces GORAN PESKIR Several iequalities of Kahae-Khitchie s tye i certai Orlicz saces are roved. For this the classical symmetrizatio techique is used ad four basically differet methods have bee reseted. The first two are based o the well-kow estimates for subormal radom variables, see [9], the third oe is a cosequece of a certai Gaussia-Jese s majorizatio techique, see [6], ad the fourth oe is obtaied by Haageru-Youg-Stechki s best ossible costats i the classical Khitchie iequalities, see [4]. Moreover, by usig the cetral limit theorem it is show that this fourth aroach gives the best ossible umerical costat i the iequality uder cosideratio: If f " i j i g is a Beroulli sequece, ad k k deotes the Orlicz orm iduced by the fuctio (x) e x 0 for x R ; the the best ossible umerical costat satisfyig the followig iequality: X a i " i X j a i j for all a;... ; a R ad all, is equal to 83. Similarly, the best ossible estimates of that tye are also deduced for some other iequalities i Orlicz saces, discovered i this aer.. Itroductio Let f " i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k deote the gauge orm o (; F ; P ), that is: k X k if f a > 0 j E[ (Xa) ] g for all real valued radom variables X o (; F ; P ), where (x) e x 0 for x R, ad with if ;. The the followig iequality is satisfied: () a i " i j a i j for all a;... ; a R ad all, where is a umerical costat, see for istace [9], [3], [4], [9], [], [30]. I this aer we shall show that the best ossible umerical costat AMS 980 subject classificatios. Primary 4A44, 4A50, 46E30, 60E5, 60G50. Secodary 44A0. Key words ad hrases: Orlicz orm, Frechet orm, Khitchie iequality, Haageru-Youg-Stechki s costats, subormal, symmetrizatio. gora@imf.au.dk
2 i iequality () is equal to 83, ad the reset work is devoted to the study of various ways for rovig (), as well as to the study of the aalogous questio for some other Orlicz orms, see sectio below. Moreover, by usig the classical symmetrizatio techique, iequality (), as well as some other iequalities of that tye which will be deduced later, will be exteded i a aroriate way to more geeral cases. Let us say that the iequality give i () has a umber of alicatios. I articular, usig that result together with certai modulus of cotiuity results of Presto s tye, see [5], [0] ad [30], oe ca obtai a coectio betwee the cetral limit theorem i a Baach sace ad the uiform law of large umbers o the uit ball of its dual sace, see [7] ad [8]. For alicatios to the law of iterated logarithm i a Baach sace, see [30].. Prelimiary facts Orlicz fuctioals, orms ad saces. Let (X; A; ) be a measure sace, ad let (B; k k) be a Baach sace. Let ' be a icreasig left cotiuous fuctio from [0; [ ito [0; [ such that '(0) lim t#0 '(t) 0 : The the Orlicz fuctioals ' ; T ' ad 7 ' geerated by ' are defied as follows: ' (R f ) if f a > 0 j T ' (R f ) if f a > 0 j 7 ' (R f ) R f (t) '(dt) R f (at) '(dt) g R f (at) '(dt) a g where R f (t) 3 f k f k > t g for f B X. Recall that R 0 f (t) '(dt) deotes the itegral of a fuctio f with resect to the Lebesgue-Stieltjes measure ' defied by ' ([0; x[) '(x) for all x 0, that 3 deotes the outer -measure, ad that B X deotes the set of all fuctios from X ito B. Also the followig formula is valid, see [5]: 3 ' d 0 3 f > t g '(dt) for every fuctio from [0; ] ito [0; ], where we ut '() lim t! '(t). The the fuctio orm ad the fuctio sace, see [5], iduced by fuctioals ', T ' ad 7 ' are give resectively by: k f k ' if f a > 0 j 3 '( a k f k ) d L ' ( 3 ; B) f f B X j lim "#0 k "f k ' 0 g k f k T' if f a > 0 j 3 '( g a k f k ) d a g L T' ( 3 ; B) f f B X j lim "#0 k "f k T' 0 g
3 k f k 7' 3 '( k f k ) d L 7' ( 3 ; B) f f B X j lim k "f k 7' 0 g "#0 R 3 f d for all f B X. Recall that deotes the uer -itegral of a fuctio f. It is show i [5] that the followig statemets are satisfied: () ( B X ; k k T' ) is a seudo-f 3 -sace, ad cosequetly ( L T' ( 3 ; B) ; k k T' ) is a seudo-frechet sace. () If ' is covex, the ( B X ; k k' ) is a seudo-b3 -sace, ad cosequetly ( L ' ( 3 ; B) ; k k ' ) is a seudo-baach sace. (3) If ' is subadditive ( for istace, cocave ), the ( B X ; k k 7' ) is a seudo-f 3 -sace, ad thus ( L 7' ( 3 ; B) ; k k 7' ) is a seudo-frechet sace. Moreover, the we have: r for all f B X. q k f k 7' ^ k f k 7' k f k T' k f k 7' _ k f k 7' (4) If ' is covex at 0, i.e. '(t) '(t) for all [0; ] ad all t 0, the ( B X ; q q k k' ) for all f B X. is a seudo-f 3 -sace, ad we have: k f k ' ^ k f k ' k f k T' q k f k ' _ k f k ' (5) If ' is cocave at 0, i.e. '(t) '(t) for all [0; ] ad all t 0, the ( B X ; q q k k 7' ) for all f B X. is a seudo-f 3 -sace, ad we have: k f k 7' ^ k f k 7' k f k T' q k f k 7' _ k f k 7' I the ext cosideratios we shall maily work with the fuctio defied by (x) e x 0 for x R. The Baach sace B will be equal to R, ad the measure will be a robability measure, that is (X). We shall write L () ; L T () ad L 7 () to deote the saces of all -measurable fuctios i L ( 3 ;R) ; L T ( 3 ;R) ad L 7 ( 3 ;R), resectively. Oe ca verify that the give saces are closed ad the atural ijectios are cotiuous. The fuctio orm k k will be shortly deoted by k k. The Orlicz sace (L (); k k ) will be called the gauge sace, ad the Orlicz orm k k will be called the gauge orm. For more iformatios i this directio we shall refer the reader to []. Subormal radom variables. Let X be a real valued radom variable defied o a robability sace (; F ; P ). The the Lalace trasform of X is give by: 3
4 L X (z) E(e zx ) for all z D(L X ), where D(L X ) f z j E je zx j < g is the comlex domai of L X. The real domai of L X is give by R(L X ) D(L X ) \ R. Let us recall that a ormal distributed radom variable N N (; ) with mea ad variace 0 has the Lalace trasform give by: L N (z) e z + z for all z D(L N ). Ad a real radom variable X is called subormal, if its Lalace trasform is domiated o the real lie by the Lalace trasform of some ormally distributed radom variable. I other words, X is subormal, if there exists R ad 0 such that: (6) L X (t) e t + t for all t R. It is well-kow that for a subormal radom variable X we have: (7) EX ad VarX. Ad if (6) is satisfied with 0 ad, the X is said to be a stadard subormal radom variable. The by (7) we have EX 0 ad VarX. A sequece of ( stadard ) subormal radom variables will be called a ( stadard ) subormal sequece. For more iformatios i this directio we shall refer the reader to [9] (.6). Gaussia-Jese s majorizatio techique. Recall that a fiite or ifiite sequece of ideedet ad idetically distributed radom variables " ; " ;... takig values 6 with the same robability is called a Beroulli sequece. ertai iequalities ivolvig Beroulli sequeces ca be easily obtaied by usig a techique which will be called the Gaussia-Jese s majorizatio techique ad which may be described as follows, see [6] (.-): Suose that we have a covex ad sig-symmetric fuctio g : R! ] 0 ; ], i.e. g is covex ad g (6x ;... ; 6x ) g (x ;... ; x ) for all choices of sigs 6 ad all (x ;... ; x ) R ; ad let X ;... ; X L (P ) be real valued radom variables. The we evidetly have: g (x ;... ; x ) g (jx j ;... ; jx j) for all (x ;... ; x ) R, ad thus by Jese s iequality we get: (8) g (x ;... ; x ) Eg (X ;... ; X ) where EjX i j jx i j, for i ;... ;. Moreover, suose that we have: (9) g (x ;... ; x ) E h x i " i is a Beroulli sequece ad h : R! R is a give fuctio such that the right where ";... ; " side i (9) is well-defied ad belogs to ] 0 ; ] for all (x;... ; x ) R. The takig X i;xi N (0; jx i j ) to be mutually ideedet for i ;... ;, as well as ideedet of 4
5 ";... ; ", by (8), Fubii s theorem ad the fact that E j X i;xi j j x i j for i ;... ;, we get: (0) g (x;... ; x ) E h X i;xi " i for (x;... ; x ) R. A usefuless of above iequality lies i the fact that we obviously have P X i;x i " i N (0 ; P jx ij ), so it might be a "good" chace to comute exactly the right side i (0), or at least to make a "good" estimate for it, ad i this way to obtai a estimate for the left side i (0), that is for the fuctio g defied by (9). For a alicatio of this techique see roof 3 of theorem 3. below. lassical Khitchie iequalities. Let f " i j i g be a Beroulli sequece, ad let 0 < ; q < ad. The the Khitchie costat K (; q) is defied to be the smallest umber satisfyig the followig iequality: () E j " i x i j K (; q) for all x ;... ; x () K(; q) su jx i j qq R. Let us ut (; q) ( 0 q )+, ad defie: K (; q) (;q) for all 0 < ; q <. The we have, see [6] (.0): (3) K (; q) (0 q )+, for all ; q > 0 ad all (4) K (; q) ad [K (; q)] are icreasig i (; ; q) (5) K (; q) ad log K (; q) are covex i, for all q > 0 ad all (6) K (; q) ( r 0 q )+ K (; r), for all ; q; r > 0 ad all (7) K(; q) 8 >< >: (8) K(; q) 0( + ) if ; q < if 0 < or 0 < q 0( + ), for all < q < < Note that K(; q) EjNj for ; q, where N N (0; ) is a stadard ormal radom variable. U. Haageru i [4], R.M.G. Youg i [3] ad S.B. Stechki i [4] have show that the costats K(; q) i (7) are ideed the best ossible i () for ; q. Let us i additio remid that Stirlig s formula states: (9)! e 0 e r, + < r < for. Recall that a cocave fuctio ' : R +! R + is subadditive, i.e. we have '(x + y) '(x) + '(y) for all x; y 0. Usig this fact ad Jese s iequality we may easily 5
6 obtai the followig two iequalities: (0) () jx i j jx i j ( 0 ), jx i j for all 0 < beig valid for all x;... ; x R ad all., jx i j for all < < 3. Basic results o the gauge sace L (P ) This sectio is devoted to the study of a iequality of Kahae-Khitchie s tye i the gauge sace (L (P ) ; k k ), where we recall that (x) e x 0 for x R. The first ste i that directio is doe i the ext theorem, see iequality () i theorem below. The result is well-kow ad our attetio is maily directed to its roof i order to fid the best ossible umerical costat uder cosideratio, as well as to obtai aroriate tools for closely related questios o some other Orlicz saces i the ext two sectios. Usig these results we reach this first ambitio i corollary 4 below. The, usig the classical symmetrizatio techique, we exted the give results to more geeral cases, see remark 7 ad corollary 9 below. Theorem. ( A Kahae-Khitchie iequality i the gauge sace ). Let f " i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k deote the gauge orm o (; F ; P ). The there exists a umerical costat > 0 such that the followig iequality is satisfied: () a i " i X ja i j for all a ;... ; a R ad all. Proof. Give a ;... ; a R for some, we deote S P a i" i ; T (S ) P ad A ja ij. The by the defiitio of the gauge orm k k, it is eough to show the followig iequality: () ex T A dp I order to illustrate various ways for rovig (), as well as to make ossible their comarisos, we shall reset four basically differet roofs of (): Proof. The first roof is based o a classical Kahae-Khitchie iequality for subormal sequeces of radom variables ad Stirlig s formula, ad we shall show that (), ad thus () also, holds with + e 56 :0.... For this ut 0 + e 56, ad let us cosider the Lalace trasform L S of S. Sice " ;... ; " are ideedet ad cosh (x) e x for x R, the we have: 6
7 (3) L S (t) E [ ex (t S ) ] Y cosh (ta i ) Y E [ ex (ta i " i ) ] Y t ja i j t A ex ex for all t R. Thus if we ut D S A, the for its Lalace trasform L D we get: L D (t) e t, for all t R, or i other words: (4) 8 S A j 9 is a stadard subormal sequece Now we aly a idea reseted i Kahae s book [9], see. 43. Sice S is symmetric, the so is (S ) k+ ad thus E (S ) k+ 0 for all k 0; ;..., ad we have: for all L S (t) t R. Hece by (3) we have: X k0 t k (k)! E(S ) k E(S ) k (k)! ex t k t A for all k ; ;..., ad all t > 0. I that way we may obtai the followig estimate for the Lalace trasform L T of T : L T (u) E [ ex (u T ) ] E [ ex (u (S ) ) ] X k0 u k k! E (S ) k + X k u k k! (k)! (t k ) (tk ex ) A k where t k > 0 are arbitrary umbers for k ; ;.... I order to fid a suitable choice for t k s let us ut (t k ) A k for k ; ;.... The we get: L T (u) + + X X k k u k k! (k)! Usig Stirlig s formula (.9), oe ca easily obtai: (k)! k k! (k) k ex (t k ) k ex ( k) (k)! k k! ( k ) (ua ) k ex ( k k ) k 0 k + 0 k for all k ; ;.... Thus uttig k k for k ; ;..., we get: 7
8 L T (u) X k (ua ) k ex k 0 k + e 56 e 56 X k (ua ) k 0 ua 0 rovided that j ua j <. Puttig u 0 A we may coclude: 0 ex T dp L T 0 A + Thus () is satisfied ad roof is comlete. e 56 0 A Proof. The secod roof is based o a classical Kahae-Khitchie iequality for tail robabilities of symmetric subormal radom variables ad o the real reresetatio of the P - itegral, ad we shall show that (), ad thus () also, holds with 6 : Put for this 0 6 ad E T A ; ad let us cosider the Lalace trasform L E of E. By the real reresetatio of the P -itegral we have: (5) L E (u) E [ ex (u E ) ] ex (u E ) dp 0 + P f ex (u E ) > t g dt + 0 e t P 8 T A > t u 9 dt + 0 P 8 ex u T A e t P 8 jd j > for all u > 0, where D is give as above by D S A. By (4) we have: q t u > t 9 dt 9 dt E [ ex (v D ) ] e v for all v R, ad sice S is symmetric, hece by Markov s iequality we get: P f jd j t g P f D t g P f ex (v D ) ex (vt) g v ex (0vt) E [ ex (v D ) ] ex 0 vt for all t; v 0. Now it is easy to check that the fuctio S(v) v 0 vt attais its miimal value o R + at the oit v mi t, ad S(v mi ) 0t. I this way we may coclude: P f jd j t g e 0t 8
9 for all t > 0. Isertig this last iequality ito (5) we fid that: L E (u) e (0 u ) t dt u 0 + 4u 0 u for all 0 < u <. Sice 0 < <, thus we may coclude: 0 0 ex dp L E 0 0 A T Thus () is satisfied ad roof is comlete. Proof 3. The third roof is based o a alicatio of the Gaussia-Jese s majorizatio techique which is described i sectio, ad we show that (), ad thus () also, holds with 43 : Put 0 43 ad defie: g(x ;... ; x ) E h 0 x for (x ;... ; x ) R, where h(x) ex 0 A for x R. The evidetly g is a covex fuctio from R ito R +, ad oe may easily observe that g is sig-symmetric. Thus by (.0) we get: ] g(a ;... ; a ) E h E [ ex 0 A T E [ ex x i " i 0 A X i " i X i " i ] where X i N (0; ja i j ) for i ;... ; are ideedet, as well as ideedet of " ;... ; " : Sice " i X i X i for i ;... ;, the by ideedece of X ;... ; X ; " ;... ; " we have P X i" i N (0; A ) ad thus: 0 A X i " i where Y N (0; ). Hece we get: 0 E [ ex T 0 A A X i " i ] E [ ex (Y ) ] 9 0 ex Y x 0 x 0 ex h 0 0 dx x i dx
10 where 0. Sice R 0 e 0ax dx a ex 0 0 rovided that roof 3 is comlete. 0 A T dp for a > 0, the we may coclude: s 0 q 0 0 < 0, or i other words that 0 >. Thus () is satisfied ad Proof 4. The idea of the fourth roof is very simle. Namely, we shall exad the itegrad i () ito Taylor s series ad the we shall aly the classical Khitchie iequalities (.) with (.7). Thus there is a good reaso to believe that the give result will bevery dee, see [4]. Ideed, we shall see i corollary 4 below that the soo give costat 83 :63... is really the best ossible. Let us say that durig author s comutatios o the subject, this fact was cojectured by J. Hoffma-Jørgese. So, we shall show that (), ad thus () also, holds with 83. Let us cosider the left side i (). The we have: ex T dp E [ ex A X k0 A (S ) ] k! ( A E (S ) k ) k By the classical Kahae-Khitchie iequalities (.) with (.7) we fid: E j where K(k; ) k 0(k + ) (6) ex A T dp a i " i j k K(k; ) X ja i j k for k ; ;.... Sice 0( ), the we have: X k0 k0 k! ( A k 0(k + ) (A ) k ) k X k0 k k! 0(k + ) Now oe ca easily check that 0(k + ) (k 0 )!! for k, where (k 0 )!! k (k 0 ) (k 0 3) Sice evidetly j j <, thus we may coclude: X (k 0 )!! k (7) ex T dp A k k! 0 0 0
11 Thus () is satisfied ad roof 4 is comlete. I order to rove that 83 is the best ossible costat i iequality () i theorem, we shall first tur out the ext two auxiliary results which are also of iterest i themselves. Lemma. Let f X i j i g be a sequece of ideedet idetically distributed radom X i ad j g is a symmetric stadard subormal variables such that E (X) <, ad let (S 0 ES ), where S P VarX, for. Suose that f sequece, that is, is symmetric ad we have L (t) e t for all t R ad all. The for every >, the sequece 8 ex [ ] j 9 is uiformly itegrable. Proof. Let > be give, the it is eough to show that for some > we have: su E [ ex ] < Thus let us deote I() su 0 E [ ex ]. The by the real reresetatio of the P -itegral we have: E f ex [ ] g I() su su 0 + su P f ex [ P f j j > ] > t g dt log t g dt Sice by our assumtio f j g is a symmetric stadard subormal sequece, the by the estimate established i roof of theorem we have: I() + ex 0 log t dt + t Thus I() <, if >. Sice by our assumtio >, the we see that there exists ] ; [ for which I() <, ad this fact comletes the roof. dt Proositio 3. Let f X i j i g be a sequece of ideedet idetically distributed
12 radom variables defied o a robability sace ( ; F; P ) such that E (X) <, let (S 0 ES ) ; where S P X i ad VarX for, let N N (0; ) be a stadard ormal radom variable, ad let k k deote the gauge orm o (; F; P ). If f j g is a symmetric stadard subormal sequece, the () k k 0! k N k as!, where k N k 83. Proof. Let us ut k k for, ad k N k. The oe ca easily comute that 83. I order to rove () we shall first rove: () lim su! Ideed suose < lim su!. Thus + " < k for some 0 < <... ad some " > 0. Sice >, the by lemma the sequece f ex [ +" ] j g is uiformly itegrable, ad therefore by the cetral limit theorem ad the defiitio of the gauge orm k k we get: lim! lim k! N N ex [ ] dp > ex [ ] dp + " ex [ ] dp + " k ex [ ] dp + " Thus < lim su! leads to a cotradictio, ad hece () is roved. Secod we show: (3) lim if! Agai suose > lim if!. Thus 0 " > k for some < <... ad some " > 0 with 0 " >. Hece by lemma, the cetral limit theorem, ad the defiitio of the gauge orm k k we get: k lim if ex [ ] dp k! k k lim if ex [ ] dp k! 0 " lim ex [ ] dp! 0 " N ex [ ] dp > 0 "
13 Thus > lim if! leads to a cotradictio, ad cosequetly (3) is roved. But the () ad (3) evidetly comlete the roof. orollary 4. The best ossible umerical costat to 83. i iequality () i theorem is equal Proof. We have show i roof 4 of theorem that iequality () i theorem is satisfied with 83. Thus the best ossible costat i iequality () i theorem, say D, is less tha 83. To rove D 83, let us take a... a for. The () i theorem gets the followig form: () " i D beig valid for all. Ad i order to aly roositio 3 we should kow that f P ( ) " i j g is a symmetric stadard subormal sequece. But let us ote that this fact is established i roof of theorem, see its relatio (4). Thus lettig! i (), by roositio 3 we get 83 D, ad hece we may coclude that the best ossible costat D is ideed equal to 83. This fact comletes the roof. Problem 5. I order to set u a atural questio related to the result i roositio 3, let us recall some basic facts from [5] which are i the backgroud of our sectio o Orlicz fuctioals, orms ad saces. Let L deote the set of all icreasig left cotiuous fuctios ' from [0; [ ito [0; [ with '(0) lim t!0+ '(t) 0 ; ad let R deote the set of all decreasig right cotiuous fuctios R from [0; [ ito [0; ]. Let Q deote the set of all fuctios q from R ito [0; ] such that q(0) 0 ad q(r) su q(r ) wheever R lim if! R for R; R ; R... R. Recall that q Q is said to be: (i) subadditive, if q(r) q(r ) + q(r ) wheever R; R ; R R ad R R 8 R, i.e. 8'; ; L with ' 8, which meas '(x + y) (x) + (y) 8x; y 0 ; we have: 0 R(x) '(dx) 0 R (x) (dx) + 0 R (x) (dx) (ii) strogly subadditive, if q(r) q(r ) + q(r ) wheever R; R ; R R ad R R 8 R, i.e. R(x + y) R (x) + R (y) 8x; y 0 (iii) weakly subadditive, if q(r) q(r ) + q(r ) wheever R; R ; R R ad R R + R, i.e. R(x) R (x) + R (x) 8x 0 (iv) homogeeous, if q(r () ) q(r) for all > 0, wheever R R (v) o-degeerate, if q(r) 0 imlies R 0 (vi) moderated, if q(r () ) q(r) for all 0 <, wheever R R Recall that R () (t) R(t) for t 0, wheever R R. Ad a atural questio related to 3
14 the result i roositio 3, see also roositio 4.3 ad () i theorem 5. below, may be stated as follows: What are ( ecessary ad ) sufficiet coditios for R (t)! R(t) (8t S) to imly q(r )! q(r), where R; R; R... R ; q Q ad S [0; [ is a give subset? Sice coditio (i) lays a imortat role i cases whe fuctio orms ad fuctio saces are iduced by a measure sace, see [5] (.8), from that oit of view, it is ot a big restrictio to assume that q i our questio satisfies that coditio also. Let us say that the aswer to that geerally stated roblem will have iterestig cosequeces related to Orlicz saces, as well as Lauret saces, ad for more iformatios i that directio we refer the reader to [5]. Usig the classical symmetrizatio techique we shall ow exted the result from theorem i a aroriate way to a more geeral case. First we shall look at the sig-symmetric case i the ext theorem, ad the we shall ass to the geeral case i theorem 8 below. Theorem 6. Let f X i j i g be a sequece of ideedet a:s: bouded symmetric real valued radom variables defied o a robability sace (; F ; P ), let k k deote the gauge orm, ad let k k deote the usual su-orm o (; F ; P ). The for every the followig iequality is satisfied: () X X i X 83 k X i k Moreover, the umerical costat 83 is the best ossible i (). Proof. Give, we deote X (X;... ; X ), ad let ";... ; " be a Beroulli sequece such that " (";... ; " ) is ideedet of X (X;... ; X ). It is o restrictio to assume that X ad " are defied o the same robability sace (; F ; P ). Put 0 83 ad let us defie a fuctio f from R R ito R as follows: " # f (x;... ; x ; ;... ; ) ex x i i 0 jx i j for (x;... ; x ; ;... ; ) R R. Sice X (X;... ; X ) ad (";... ; " ) are ideedet, the by Fubii s theorem we may coclude: where E ex " 0 jx i j g(x) E f (x; ") E ex # o X i " i " 0 jx i j E g(x) # o x i " i for x (x;... ; x) R. By roof 4 of theorem oe directly fids that g(x) for all 4
15 x R, ad thus we may obtai Eg(X). Sice by our assumtios X (X;... ; X ) is sig-symmetric, the by the iequality just established we may coclude: () E But the we have: E ex " E ex Thus we may coclude: " E 0 ex " jx i j 0 X # o X i 0 k X i k ex " 0 jx i j X # o X i " i X # o X i jx i j X X i X 0 X # o X i k X i k ad the roof of () is comlete. The last statemet follows directly by corollary 4. These facts comlete the roof. Remark 7. Note that () i the roof of theorem 6 states: If X;... ; X are ideedet symmetric real valued radom variables, the the followig iequality is satisfied: X X i 83 jx i j Moreover, the umerical costat 83 is the best ossible ( the smallest ) with that roerty. Theorem 8. Let f X i j i g be a sequece of ideedet a:s: bouded real valued radom variables defied o a robability sace (; F ; P ), let k k deote the gauge orm, ad let k k deote the usual su-orm o (; F ; P ). The for every the followig iequality is satisfied: () X (X i 0 EX i ) X 33 k X i 0 EX i k 5
16 Proof. Give, we deote X (X;... ; X ), ad let Y (Y;... ; Y ) be a radom vector such that X ad Y are ideedet ad idetically distributed. It is o restrictio to assume that X ad Y are defied o the same robability sace (; F ; P ), as well as that EX i 0 for i ;... ;. Put S P X i ad T P Y i, the we have: () k S k k S 0 T k For this it is eough to show that E ex h S () i o where () k S 0 T k. Thus defie: f (s; t) ex h s 0 t () i for s; t R. The evidetly t 7! f (s; t) is a covex fuctio o R, for all s R, ad moreover by our assumtios we have T L (P ) with ET 0. Therefore by Fubii s theorem ad Jese s iequality we may easily obtai: Ef (S ; ET ) Ef (S ; T ) Sice ET 0, the by the defiitio of the Orlicz orm k k we get: E 8 ex h S () i 9 Ef (S ; ET ) Ef (S ; T ) h S 0 T E 8 ex () i 9 Thus () is roved. Now sice X (X;... ; X ) ad Y (Y;... ; Y ) are ideedet ad idetically distributed, ad X;... ; X are ideedet, the obviously X 0 Y (X 0 Y;... ; X 0 Y ) is sig-symmetric, ad thus by () ad theorem 6 we may coclude: k S k k S 0 T k X h X Thus () is showed ad the roof is comlete. X (X i 0 Y i ) k X i 0 Y i k 0 k X i k + k Y i k i k X i k 6
17 orollary 9. Let f X i j i g be a sequece of ideedet a:s: bouded real valued radom variables defied o a robability sace (; F ; P ), let k k deote the gauge orm, ad let k k deote the usual su-orm o (; F ; P ) : The for every > 0 ad all we have: () X where () is give by: () (X i 0 EX i ) 8 < : () X 33 ; if 0 < k X i 0 EX i k 33 0 ; if < < : Moreover, if X; X;... are symmetric, the for every > 0 ad all we have: () X X i X D () where D () is give by: D () 8 < : k X i k 83 ; if 0 < 83 0 ; if < < : Proof. Iequality () follows by theorem 8, (.0) ad (.), ad iequality () follows by theorem 6, (.0) ad (.). 4. Basic results o the Orlicz sace L T (P ) Recall that the gauge orm k k from the revious sectio is the orm k k iduced by the Orlicz fuctioal, as defied i sectio, where (x) e x 0 for x R. osequetly, oe ca be iterested to fid out iequalities ivolvig the orms k k T ad k k7 which corresod to the iequalities reseted i corollary 3.9 ad remark 3.7. Note that the iequality i (.4) ca be used for this urose as well as the iequalities i (.3) ad (.5) for related oes, but we shall try to rove it directly usig the facts obtaied i revious aroaches. The startig oit should obviously be the iequality reseted i theorem 3. for which we have tured out four basically differet roofs, or i other words four differet techiques. However, ote that the Orlicz orm k k T is ot ecessarily homogeeous, but oe ca easily check that we have: () k c X k T j c j k X k T, for all j c j () k c X k T j c j k X k T, for all j c j where X is a give radom variable, see also (.). Ad accordig to the result i corollary 3.4 we may coclude that the fourth aroach i the roof of theorem gives the best estimate for the left side uder cosideratio, so it is reasoable to aly that techique i order to get as otimal 7
18 result as ossible. Ad this is really doe i the roof of the ext theorem. Theorem. Let f "i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio. The the followig iequality is satisfied: () jaij ai"i T for all a;... a R ad all, where the umerical costat s () s r 3 08 Moreover, that umerical costat is the best ossible i (). + 3 s , with is give by: Proof. Give a;... P ; a R for some, we deote as before P S a i"i, T (S) ad A ja ij. The by the defiitio of the Orlicz orm k k T, it is eough to show the followig iequality: (3) ex T dp + A For this ote that by (7) i roof 4 of theorem 3. we have: (4) ex T dp 0 0 x x A for all x >. Put (x) ( 0 x ) 0 ad (x) + x for x >. The oe ca easily check that > o ] ; [, < o ] ; [ ad () (). Moreover, give satisfies the followig equatio: Thus by usig Ferrari s formulas, see [6], it is a matter of routie to check that is give by () above. Hece (3) follows by (4) ad the roof of () is comlete. To rove that the give umerical costat is the best ossible i (), we shall follow the idea reseted i the roof of corollary 3.4. So, let D deote the best ossible umerical costat i (). If we take a... a for ; the () gets the followig form: (5) " i T D Now oe eeds a result similar to that reseted i roositio 3.3 but with k k T istead of k k. Ad this fact will be established i roositio 3 below, so lettig! i (5), by r
19 roositio 3 below we may coclude that D. Sice the iequality D follows by (), this fact comletes the roof. Remark. We have see i the roof of theorem that the umerical costat give by () i theorem is a uique solutio of the equatio: x 4 + x 3 0 x 0 4x 0 0 for x >. By the well-kow criterio for ratioal solutios for algebraic equatios with ratioal coefficiets, see [6], each ratioal solutio of the above equatio belogs to the set f 6; 6 g. Ad oe ca easily check that 6, as well as 6, does ot satisfy the above equatio, ad thus we may coclude that the above equatio has o ratioal solutios at all. Therefore the umerical costat give by () i theorem is ot a ratioal umber. But oe ca easily check that we have: : ; as well as that the followig ratioal aroximatios are valid: () < < ad q 7 30 < < q 9 8 where :003..., ad : Thus iequality () i theorem is satisfied with 7750, as well as with 98. Note that 83, see corollary < Proositio 3. Let f X i j i g be a sequece of ideedet idetically distributed radom variables defied o a robability sace ( ; F ; P ) such that E (X ) <, let (S 0 ES ) ; where S P X i ad VarX for, let N N (0; ) be a stadard ormal radom variable, ad let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio. If f j g is a symmetric stadard subormal sequece, the () k k T 0! k N k T as!, where k N k T is equal to the umerical costat give by () i theorem. Proof. Let us ut k k T for, ad k N k T. The by the defiitio of the Orlicz orm k k T oe ca easily check that > ad : Thus by the roof of theorem we see that is give by () i theorem. Ad i order to rove that! as!, oe ca reeat the roof of () ad (3) reseted i the roof of roositio 3.3, where the umerical costat should be relaced by the umerical costat + o the right laces. These facts easily comlete the roof. Theorem 4. Let f X i j i g be a sequece of ideedet symmetric real valued radom variables defied o a robability sace (; F ; P ), ad let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio. The for every the followig iequality is satisfied: () jx i j X i T 9
20 where is the umerical costat give by () i theorem. Moreover, that umerical costat is the best ossible i (). Proof. The roof of () is comletely the same as the roof of () i theorem 3.6, where the umerical costat should be relaced by the umerical costat + o the right laces, with 0, ad oe should use theorem istead of roof 4 of theorem 3. to obtai the desired iequality. Also ote that the last statemet follows directly by the last statemet i theorem. These facts easily comlete the roof. We shall cotiue our cosideratios by searchig for a aalogous iequality to that reseted i theorem 3. where the gauge orm k k should be relaced by the Orlicz orm k k T as defied i sectio. For this, let us cosider a Beroulli sequece f "i j i g which is defied o a robability sace (; F ; P ). Desite the fact that the Orlicz orm k k T o (; F ; P ) is ot homogeeous, accordig to (4.) ad iequality () i theorem, we may coclude that the followig iequality is satisfied: (3) X ai"i for all a;... ; a R ad all T jaij for which P ja ij, where is the umerical costat give by () i theorem. Ad oe ca ask is this iequality true i geeral, or i other words, does (3) hold for all a;... ; a R ad all? We shall ow show that the aswer to this questio is egative. For this, suose that (3) is satisfied for all a;... ; a R ad all. The takig a... a for, we get: "i T for all. Hece by the defiitio of the Orlicz orm k k T ex " i dp + we may coclude: for all. Lettig!, by lemma 3. ad the cetral limit theorem we easily obtai: ex! N dp lim ex " i dp where N N (0; ) is a stadard ormal radom variable. Sice this iequality obviously does ot hold for ay real umber, thus (3) does ot hold i that case. By the way, let us ote that usig the fact established i lemma 3. together with the classical Hartma-Witer law of iterated logarithm oe ca easily coclude: lim! ex " i dp 0
21 for ay > ad ay > 0. However, ote that by the geeral iequality give i (.4) ad theorem 3. we get: X h a i " T i 83 ja i j i h _ 83 ja i j i for all a;... ; a R ad all. Hece we ca deduce the followig iequality with a uique costat: (4) X a i " T h i 83 ja i j _ for all a;... ; a R ad all. I articular, we have: (5) X a i " T i 83 ja i j 4 ja i j 4 i for all a;... ; a R ad P for which ja ij. Now oe ca ask, first of all, is the umerical costat 83 the best ossible i (4)? I other words we may ask, is it the P best ossible i (5), sice by (3) we kow that this is ot true for ja ij? I order to get some relimiary iformatios i this directio, let D deote the best ossible umerical costat i (5), that is, let (5) be satisfied for all a;... ; a R ad all for which P ja ij ; if we relace 83 by D, ad let D be the smallest umber with that roerty. Puttig a... a for all, by roositio 3 ad (5) we may easily coclude that D ; where deotes the umerical costat give by () i theorem. Thus D 83 ; ad i the ext lemma we show that the first iequality is actually equality. These facts aswer our first questio. Lemma 5. The best ossible umerical costat i iequality (4) above is equal to by () i theorem. give Proof. Give a;... ; a R for some, we deote S P P a i" i, T (S ) ad A ja ij. Accordig to (3) ad coclusios which recede lemma 5, the oly fact which remais to rove is iequality (5) with the umerical costat give by () i theorem, istead of 83. For this we follow roof 4 of theorem 3., ad usig the same argumets as for relatios (6) ad (7) there, we may obtai: () ex T dp A X X k0 k0 X k0 k! A k 0(k + ) (k 0 )!! k k! k! ( k 0(k + ) (A ) k A ) k A k 0 A 0
22 sice evidetly by our assumtios j A j <, where is the umerical costat give by () i theorem. Now oe ca easily check that: x x + 0 x 0 0 for all () 0 x, ad thus the followig iequality is satisfied: 0 x 0 + x for all 0 x. Ad by () ad () we get: ex A T dp + 4 A Hece by the defiitio of the Orlicz orm k k T we may coclude that k S k T 4 A, ad this fact comletes the roof. Note that a mai reaso for exoet 4 aears i iequalities (4) ad (5) above is comig from the geeral iequality give i (.4). Ad oe ca ask is this exoet ideed the best ossible i that case? I order to aswer this questio, let f "i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio. Give a ;... ; a R for some, we deote S P a i" i, T (S ) P ad A ja ij. Suose that A, let > be a give umber, ad let q be the cojugate exoet of, that is: + q. The the same argumet as i () i the roof of lemma 5, yields the followig estimate: (6) ex (A ) X k0 T dp k! k 0(k + ) (A ( ) k (A ) ) k X q k! (A ) k 0(k + ) k0 X (k 0 )!! (A ) q k k k! 0 (A ) q 0 k0 sice evidetly j (A ) q j <, where is the umerical costat give by () i theorem. Let us ow defie: q 3 su 8 q j 0 xq 0 + x 0q ; 8x [0; ] 9 Recall that by the defiitio of, see the roof of theorem, we have:
23 0 0 + Thus the iequality i defiitio of q 3 is always satisfied for x, as well as for x 0. Let 3 be the cojugate exoet of q 3, that is: 3 + q 3. The we evidetly have: for all 0 xq3 0 + x 0q3 + x 3 0 x, ad thus by (6) we may obtai: ex (A) 3 T dp + (A) 3 By the defiitio of the Orlicz orm k k T hece we may coclude: (7) k S kt (A ) 3 Note that by lemma 5 we have 3, or i other words it is o restrictio to assume that q 3. Moreover, it is easy to check that the iequality i defiitio of q 3, for q 4, is equivalet to the followig iequality: x 6 0 x x3 0 x + 0 for all 0 x, which is obviously ot satisfied, sice the left side takes the value > 0 at x 0. Thus q 3 < 4, or i other words 3 > 43. osequetly, we may coclude: (8) 8 3 < 3 8 Ad the aim of the ext lemma is to establish that q 3 3, or i other words that 3 3. Lemma 6. The largest umber q satisfyig the followig iequality: () 0 xq 0 + x 0q for all 0 x, where is the umerical costat give by () i theorem, is equal to 3. Proof. By (8) above we kow that the largest umber q satisfyig (), say q 3, is strictly less tha 4. Furthermore, oe ca easily verify that for q 3, iequality () is equivalet to the followig easy to check iequality: x x for all 0 x. Thus we may deduce that 3 q 3 < 4. Let us therefore take 0 < " <, 3
24 ad let us cosider iequality () for q 3 + ". The oe ca easily verify that iequality () is equivalet to the followig iequality: x +" 0 x +" + 4 x+" 0 x " + 0 for all 0 < x. But ote that for every " > 0, the left side above takes the value > 0 at x 0, ad therefore we may coclude that () is ot satisfied for ay q 3 + " with some " > 0. This fact shows that the largest q satisfyig iequality () is equal exactly to 3, ad the roof is comlete. Theorem 7. Let f "i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio. The the followig iequality is satisfied: () X ai"i h 3 i T jaij _ jaij for all a ;... ; a R ad all, where is the umerical costat give by () i theorem. Moreover, that umerical costat is the best ossible i (). Proof. Iequality () follows directly by (3), (7) ad lemma 6, ad the last statemet follows straight forward by lemma 5. Problem 8. Note that the exoet 3 i iequality () i theorem 7 is otimal i the framework of our best estimate established i roof 4 of theorem 3. which i tur lies o the best ossible costats i classical Khitchie iequalities, see (6) ad (7) i roof 4 of theorem 3.. However, ote that we did ot rove that it is ideed the best ossible, maily because of the fact that somethig "wrog" may hae for "small" A s for which our basic iequality emloyed above, see (6), ossibly does ot work i the best way. Hece, we ca set u a atural questio: What is the best ossible exoet which ca take the lace of 3 i iequality () i theorem 7? Note that by results established above this umber is greater or equal to 3, ad strictly less tha. As usual, by usig the classical symmetrizatio techique, we shall exted the result of theorem 7 to more geeral cases i the ext two theorems. Theorem 9. Let f X i j i g be a sequece of ideedet a:s: bouded symmetric real valued radom variables defied o a robability sace (; F ; P ), let k k T deote the Orlicz orm as defied i sectio, ad let k k deote the usual su-orm o (; F ; P ). The for every the followig iequality is satisfied: () where X X T i h k X i k _ k X i k 3 i is the umerical costat give by () i theorem. Moreover, that umerical costat 4
25 is the best ossible i (). Proof. The roof of () is comletely the same as the 0P roof of () i theorem 3.6, where the umerical costat should be relaced by + jx ij 0P _ jx ij 3 3 0P ; ad the by + k X i k 0P _ k X i k 3 3 P o the right laces, as well as exressios jx ij P ad jx ij 0P by jx ij 0P _ jx ij 3 ad 0P jx ij 0P _ jx ij 3, ad where theorem 7 should be used istead of theorem 3. o the right laces. Also ote that the last statemet follows directly by the last statemet give i theorem 7. These facts easily comlete the roof. Theorem 0. Let f X i j i g be a sequece of ideedet a:s: bouded real valued radom variables defied o a robability sace (; F ; P ), let k k T deote the Orlicz orm as defied i sectio, ad let k k deote the usual su-orm o (; F ; P ). The for every the followig iequality is satisfied: () X h (X i 0 EX i ) T k X i 0 EX i k k X i 0 EX i k where is the umerical costat give by () i theorem. 3 i Proof. The roof of () is comletely the same to the roof of () i theorem 3.8, where the umerical costat should be relaced by the umerical costat + () + k S 0T k T o the right laces. I this way we may coclude: k S k T k S 0 T k T for all, ad i the rest oe should use theorem 9 istead of theorem 3.6 to deduce the fial coclusio. These facts easily comlete the roof. orollary. Let f X i j i g be a sequece of ideedet a:s: bouded real valued radom variables defied o a robability sace (; F ; P ), let k k T deote the Orlicz orm o (; F ; P ) as defied i sectio, let k k deote the usual su-orm o (; F ; P ), ad let be the umerical costat give by () i theorem. The for every > 0 ad all we have: () X (X i 0 EX i ) where () is give by: T () _ h X k X i 0 EX i k X k X i 0 EX i k 3 i _ 5
26 () 8 < : ; if 0 < 0 ; if < < : Moreover, if X; X;... are symmetric, the for every > 0 ad all we have: () X X T i D () where D () is give by: D () 8 < : h k X i k _ ; if 0 < 0 ; if < < : k X i k. 3 i Proof. Iequality () follows by theorem 0, (.0) ad (.), ad iequality () follows by theorem 9, (.0) ad (.). 5. Basic results o the Orlicz sace L 7 (P ) The cosideratios i this sectio are devoted to the study of the questios reseted i the last two sectios, but ow for the Orlicz orm k k7 as defied i sectio, where (x) ex 0 for x R. Similarly to the revious aroach we shall essetially use the estimate established i roof 4 of theorem 3.. The first result i this directio may be stated as follows: Theorem. Let f "i j i g be a Beroulli sequece defied o a robability sace (; F ; P ), ad let k k 7 deote the Orlicz orm o (; F ; P ) as defied i sectio. The for every > the followig iequality is satisfied: () jaij ai"i for all a ;... ; a R ad all. Moreover, the estimate give by () is the best ossible i the sese described i the roof below. (S) Proof. Give a ;... ; a R for some, we deote as before S P a i"i ; T P ad A ja ij. Accordig to (7) i roof 4 of theorem 3. we may coclude: ex jaij T A ai"i dp T
27 0 0 for all ad all >. Thus () is satisfied ad the first art of the roof is comlete. Ad for the last statemet, let us take a... a for, the () gets the followig form: " i 7 for all ad all > () as! P, for all > S " i, we have: (3) " i 7 ex 0 0. Ad we shall ow show that: 0! 0 0. For this, it is eough to show that for give > h S i dp 0! 0, with as!. I order to rove (3), by the cetral limit theorem it is eough to verify that the sequece f ex ( ) ] j g is uiformly itegrable, where S for. Ad by lemma 3. we may otice that this fact is satisfied, if f j g is a symmetric stadard subormal sequece. But this last fact is established i roof of theorem 3., see its relatio (4). Thus (3) follows, ad therefore () is satisfied also. Hece we may coclude that the estimate give by () is ideed the best ossible, i the sese that give > ad " > 0 with 0 " > ; we ca fid " ad a ;... ; a " R such that the followig two iequalities are satisfied: ( 0 ") " X " X ja i j ja i j X " X " a i " i 7 a i " i 7 > > 0 + " ( + ") 0 Note that the fuctio 7! 0 is decreasig o ] ; [. These facts comlete the roof. 0 0 Theorem. Let f X i j i g be a sequece of ideedet symmetric real valued radom variables defied o a robability sace (; F ; P ), ad let k k 7 deote the Orlicz orm o (; F ; P ) as defied i sectio. The for every > ad all the followig iequality is satisfied: () jx i j X i
28 Moreover, the estimate give by () is the best ossible i the sese described i the roof of theorem. Proof. Give, we deote X (X ;... ; X ), ad let " ;... ; " be a Beroulli sequece such that " (" ;... ; " ) is ideedet of X (X ;... ; X ). It is o restrictio to assume that X ad " are defied o the same robability sace (; F ; P ). Exactly as i the roof of theorem 3.6 we may coclude: for all > 8 h E ex jx i j X i 9 X i " i, where the fuctio g is give by: 8 h g(x; ) E ex for x (x ;... ; x ) R ad > g(x; ) jx i j Eg(X; ) X i 9 x i " i. By () i theorem we have: 0 for all x R ad all >. Sice by our assumtios X (X ;... ; X ) is sigsymmetric, the by the iequality above we may coclude: 8 h E ex E 8 ex h jx i j jx i j X i 9 X i X i 9 X i " i 0 for all >. Hece () follows directly, ad the first art of the roof is comlete. The last statemet follows straight forward by the last statemet i theorem. These facts comlete the roof. Remark 3. By (4.6) we may easily deduce the followig "dual" estimate which exteds the result of theorem : () a X i " i 0 ja i j 7 ja i j q 0 0 for all a;... a R, all, ad all > 0 for which 0 P ja ij < ;
29 where > ad + q. Now it is a matter of routie, see the roof of theorem 3.6, to coclude that the followig iequality exteds iequality () give i theorem : If X ; X ;... are ideedet symmetric a:s: bouded real valued radom varibles, the we have: () jx i j X i 7 0 k X i k q 0 0 for all > 0 ad all for which 0 P k X i k < 0 0, where > ad + q. Moreover, uttig a i for i ;... ; i (), oe ca easily establish that the estimates () ad () are the best ossible, for >, i the sese described i the roof of theorem. We shall leave the details i this directio to the reader. Ackowledgmet. The author would like to thak his suervisor, Professor J. Hoffma- Jørgese, for istructive discussios ad valuable commets. REFERENES [] ARAUJO, A. P. (978). O the cetral limit theorem i Baach saces. J. Multivariate Aal. 8 (598-63). [] BOLLOBAS, B. (980). Martigale iequalities. Math. Proc. amb. Philos. Soc. 87 (377-38). [3] GLUSKIN, E. D. PIETSH, A. ad PUHL, J. (980). A geeralizatio of Khitchie s iequality ad its alicatio i the theory of oerator ideals. Studia Math. 67 (49-55). [4] HAAGERUP, U. (978-98). The best costats i the Khitchie iequality. Oerator algebras, ideals, ad their alicatios i theoretical hysics, Proc. It. of. Leizig (69-79). Studia Math. 70 (3-83). [5] HOFFMANN-JØRGENSEN, J. (99). Fuctio orms. Math. Ist. Aarhus, Prerit Ser. No. 40, (9 ). [6] HOFFMANN-JØRGENSEN, J. (99). Iequalities for sums of radom elemets. Math. Ist. Aarhus, Prerit Ser. No. 4, (7 ). [7] HOFFMANN-JØRGENSEN, J. (994). Probability with a view toward statistics. hama ad Hall. [8] JOHNSON, W. B. SHEHTMAN, G. ad INN, J. (985). Best costats i momet iequalities for liear combiatios of ideedet ad exchageable radom variables. A. Probab. 3 (34-53). [9] KAHANE, J. P. ( ). Some radom series of fuctios. D.. Heath & o. (first editio). ambridge Uiversity Press (secod editio). [0] KOMOROWSKI, R. (988). O the best ossible costats i the Khitchie iequality for 3. Bull. Lodo Math. Soc. 0 (73-75). 9
30 [] KRASNOSEL SKII, M. A. ad RUTIKII, Ya. B. (96). ovex fuctios ad Orlicz saces. P. Noordhoff, Ltd. Groige. [] LEDOUX, M. (985). Sur ue iégalité de H. P. Rosethal et le théorème limite cetral das les esaces de Baach. Israel J. Math. 50 (90 38). [3] MARUS, M. B. ad PISIER, G. (984). haracterizatios of almost surely cotiuous -stable radom Fourier series ad strogly statioary rocesses. Acta Mathematica 5 (45-30). [4] MARUS, M. B. ad PISIER, G. (985). Stochastic rocesses with samle aths i exoetial Orlicz saces. Proc. Probab. Baach Saces V, Lecture Notes i Math. 53 (38-358). [5] NANOPOULOS,. ad NOBELIS, P. (978). Régularité et roriétés limites des foctios aléatoires. Sém. Probab. XII, Lecture Notes i Math. 649 ( ). [6] NEWMAN,. M. (975). A extesio of Khitchie s iequality. Bull. Amer. Math. Soc. 8 (93-95). [7] PESKIR, G. (99). Note o the coectio betwee the cetral limit theorem ad the uiform law of large umbers i Baach saces. Not aeared i writte form. [8] PESKIR, G. ad WEBER, M. (99). Necessary ad sufficiet coditios for the uiform law of large umbers i the statioary case. Math. Ist. Aarhus, Prerit Ser. No. 7, (6 ). Proc. Fuct. Aal. IV (Dubrovik 993), Various Publ. Ser. Vol. 43, 994 (65-90). [9] PISIER, G. (98). De ouvelles caractérisatios des esembles de Sido. Adv. i Math. Sul. Stud. 7b, Academic Press, New York (686-75). [0] PRESTON,. (97). Baach saces arisig from some itegral iequalities. Idiaa Uiv. Math. Joural 0 (997-05). [] RAO, M. M. ad REN,. D. (99). Theory of Orlicz saces. Marcel Dekker Ic., New York. [] RODIN, V. A. ad SEMYONOV, E. M. (975). Rademacher series i symmetric saces. Aalysis Mathematica (07-). [3] SAWA, J. (985). The best costat i the Khitchie iequality for comlex Steihaus variables, the case. Studia Math. 8 (07-6). [4] STEHKIN, S. B. (96). O the best lacuary systems of fuctios. Izv. Akad. Nauk SSSR Ser. Mat. 5 (i Russia) ( ). [5] SAREK, S. J. (978). O the best costat i the Khitchie iequality. Studia Math. 58 (97-08). [6] TIGNOL, J. P. (988). Galois theory of algebraic equatios. Istitut de Mathématique Pure et Aliquée, UL, Louvai-la-Neuve, Belgium. [7] TOMASEWSKI, B. (98). Two remarks o the Khitchie-Kahae iequality. olloq. Math. 46 (83-88). [8] WANG G. (99). Shar square-fuctio iequalities for coditioally symmetric martigales. Tras. Amer. Math. Soc. 38 (393-49). [9] WEBER, M. (983). Aalyse ifiitesimale de foctios aleatoires. Ecole d Eté Probabilités de Sait-Flour XI, Lecture Notes i Math. 976 ( ). 30
31 [30] WEBER, M. (99). New sufficiet coditios for the law of the iterated logarithm i Baach saces. Sém. Probab. XXV, Lecture Notes i Math. 495 (30-35). [3] YOUNG, R. M. G. (976). O the best ossible costats i the Khitchie iequality. J. Lodo Math. Soc. 4 ( ). Gora Peskir Deartmet of Mathematical Scieces Uiversity of Aarhus, Demark Ny Mukegade, DK-8000 Aarhus home.imf.au.dk/gora gora@imf.au.dk 3
Maximal Inequalities of Kahane-Khintchine s Type in Orlicz Spaces
Math. Proc. Cambridge Philos. Soc. Vol. 5, No., 994, (75-90) Prerit Ser. No. 33, 992, Math. Ist. Aarhus Maximal Iequalities of Kahae-Khitchie s Tye i Orlicz Saces GORAN PESKIR Several imal iequalities
More informationBest Constants in Kahane-Khintchine Inequalities for Complex Steinhaus Functions
Proc. Amer. Math. Soc. Vol. 3, No. 0, 995, (30-3) Prerit Ser. No. 4, 993, Math. Ist. Aarhus Best Costats i Kahae-Khitchie Iequalities for Comlex Steihaus Fuctios GORAN PESKIR Let f' k g k be a sequece
More informationA Note on Sums of Independent Random Variables
Cotemorary Mathematics Volume 00 XXXX A Note o Sums of Ideedet Radom Variables Pawe l Hitczeko ad Stehe Motgomery-Smith Abstract I this ote a two sided boud o the tail robability of sums of ideedet ad
More informationECE534, Spring 2018: Final Exam
ECE534, Srig 2018: Fial Exam Problem 1 Let X N (0, 1) ad Y N (0, 1) be ideedet radom variables. variables V = X + Y ad W = X 2Y. Defie the radom (a) Are V, W joitly Gaussia? Justify your aswer. (b) Comute
More informationECE534, Spring 2018: Solutions for Problem Set #2
ECE534, Srig 08: s for roblem Set #. Rademacher Radom Variables ad Symmetrizatio a) Let X be a Rademacher radom variable, i.e., X = ±) = /. Show that E e λx e λ /. E e λx = e λ + e λ = + k= k=0 λ k k k!
More informationConvergence 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 informationA REFINEMENT OF JENSEN S INEQUALITY WITH APPLICATIONS. S. S. Dragomir 1. INTRODUCTION
TAIWANESE JOURNAL OF MATHEMATICS Vol. 14, No. 1,. 153-164, February 2010 This aer is available olie at htt://www.tjm.sysu.edu.tw/ A REFINEMENT OF JENSEN S INEQUALITY WITH APPLICATIONS FOR f-divergence
More information7.1 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 informationThe log-behavior of n p(n) and n p(n)/n
Ramauja J. 44 017, 81-99 The log-behavior of p ad p/ William Y.C. Che 1 ad Ke Y. Zheg 1 Ceter for Applied Mathematics Tiaji Uiversity Tiaji 0007, P. R. Chia Ceter for Combiatorics, LPMC Nakai Uivercity
More informationIf 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 informationFinal Solutions. 1. (25pts) Define the following terms. Be as precise as you can.
Mathematics H104 A. Ogus Fall, 004 Fial Solutios 1. (5ts) Defie the followig terms. Be as recise as you ca. (a) (3ts) A ucoutable set. A ucoutable set is a set which ca ot be ut ito bijectio with a fiite
More informationChapter 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 informationNotes on the prime number theorem
Notes o the rime umber theorem Keji Kozai May 2, 24 Statemet We begi with a defiitio. Defiitio.. We say that f(x) ad g(x) are asymtotic as x, writte f g, if lim x f(x) g(x) =. The rime umber theorem tells
More informationSYMMETRIC POSITIVE SEMI-DEFINITE SOLUTIONS OF AX = B AND XC = D
Joural of Pure ad Alied Mathematics: Advaces ad Alicatios olume, Number, 009, Pages 99-07 SYMMERIC POSIIE SEMI-DEFINIE SOLUIONS OF AX B AND XC D School of Mathematics ad Physics Jiagsu Uiversity of Sciece
More informationAn operator equality involving a continuous field of operators and its norm inequalities
Available olie at www.sciecedirect.com Liear Algebra ad its Alicatios 49 (008) 59 67 www.elsevier.com/locate/laa A oerator equality ivolvig a cotiuous field of oerators ad its orm iequalities Mohammad
More information1 Convergence in Probability and the Weak Law of Large Numbers
36-752 Advaced Probability Overview Sprig 2018 8. Covergece Cocepts: i Probability, i L p ad Almost Surely Istructor: Alessadro Rialdo Associated readig: Sec 2.4, 2.5, ad 4.11 of Ash ad Doléas-Dade; Sec
More informationMAT1026 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 informationEquations and Inequalities Involving v p (n!)
Equatios ad Iequalities Ivolvig v (!) Mehdi Hassai Deartmet of Mathematics Istitute for Advaced Studies i Basic Scieces Zaja, Ira mhassai@iasbs.ac.ir Abstract I this aer we study v (!), the greatest ower
More informationWeak and Strong Convergence Theorems of New Iterations with Errors for Nonexpansive Nonself-Mappings
doi:.36/scieceasia53-874.6.3.67 ScieceAsia 3 (6: 67-7 Weak ad Strog Covergece Theorems of New Iteratios with Errors for Noexasive Noself-Maigs Sorsak Thiawa * ad Suthe Suatai ** Deartmet of Mathematics
More information7.1 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 informationMachine 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 informationTR/46 OCTOBER THE ZEROS OF PARTIAL SUMS OF A MACLAURIN EXPANSION A. TALBOT
TR/46 OCTOBER 974 THE ZEROS OF PARTIAL SUMS OF A MACLAURIN EXPANSION by A. TALBOT .. Itroductio. A problem i approximatio theory o which I have recetly worked [] required for its solutio a proof that the
More informationA Central Limit Theorem for Belief Functions
A Cetral Limit Theorem for Belief Fuctios Larry G. Estei Kyougwo Seo November 7, 2. CLT for Belief Fuctios The urose of this Note is to rove a form of CLT (Theorem.4) that is used i Estei ad Seo (2). More
More informationLecture 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(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 informationSelf-normalized deviation inequalities with application to t-statistic
Self-ormalized deviatio iequalities with applicatio to t-statistic Xiequa Fa Ceter for Applied Mathematics, Tiaji Uiversity, 30007 Tiaji, Chia Abstract Let ξ i i 1 be a sequece of idepedet ad symmetric
More informationRiesz-Fischer Sequences and Lower Frame Bounds
Zeitschrift für Aalysis ud ihre Aweduge Joural for Aalysis ad its Applicatios Volume 1 (00), No., 305 314 Riesz-Fischer Sequeces ad Lower Frame Bouds P. Casazza, O. Christese, S. Li ad A. Lider Abstract.
More informationLecture 10 October Minimaxity and least favorable prior sequences
STATS 300A: Theory of Statistics Fall 205 Lecture 0 October 22 Lecturer: Lester Mackey Scribe: Brya He, Rahul Makhijai Warig: These otes may cotai factual ad/or typographic errors. 0. Miimaxity ad least
More information6.3 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 informationSingular 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 informationDANIELL AND RIEMANN INTEGRABILITY
DANIELL AND RIEMANN INTEGRABILITY ILEANA BUCUR We itroduce the otio of Riema itegrable fuctio with respect to a Daiell itegral ad prove the approximatio theorem of such fuctios by a mootoe sequece of Jorda
More informationTHE INTEGRAL TEST AND ESTIMATES OF SUMS
THE INTEGRAL TEST AND ESTIMATES OF SUMS. Itroductio Determiig the exact sum of a series is i geeral ot a easy task. I the case of the geometric series ad the telescoig series it was ossible to fid a simle
More informationSequences 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 informationON SUPERSINGULAR ELLIPTIC CURVES AND HYPERGEOMETRIC FUNCTIONS
ON SUPERSINGULAR ELLIPTIC CURVES AND HYPERGEOMETRIC FUNCTIONS KEENAN MONKS Abstract The Legedre Family of ellitic curves has the remarkable roerty that both its eriods ad its suersigular locus have descritios
More informationSolutions to Problem Sheet 1
Solutios to Problem Sheet ) Use Theorem. to rove that loglog for all real 3. This is a versio of Theorem. with the iteger N relaced by the real. Hit Give 3 let N = [], the largest iteger. The, imortatly,
More informationProposition 2.1. There are an infinite number of primes of the form p = 4n 1. Proof. Suppose there are only a finite number of such primes, say
Chater 2 Euclid s Theorem Theorem 2.. There are a ifiity of rimes. This is sometimes called Euclid s Secod Theorem, what we have called Euclid s Lemma beig kow as Euclid s First Theorem. Proof. Suose to
More informationSolutions to HW Assignment 1
Solutios to HW: 1 Course: Theory of Probability II Page: 1 of 6 Uiversity of Texas at Austi Solutios to HW Assigmet 1 Problem 1.1. Let Ω, F, {F } 0, P) be a filtered probability space ad T a stoppig time.
More informationAdvanced 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 informationAn 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 informationInclusion Properties of Orlicz and Weak Orlicz Spaces
J. Math. Fud. Sci., Vol. 48, No. 3, 06, 93-03 93 Iclusio Properties of Orlicz ad Weak Orlicz Spaces Al Azhary Masta,, Hedra Guawa & Woo Setya Budhi Aalysis ad Geometry Group, Faculty of Mathematics ad
More informationCommon Coupled Fixed Point of Mappings Satisfying Rational Inequalities in Ordered Complex Valued Generalized Metric Spaces
IOSR Joural of Mathematics (IOSR-JM) e-issn: 78-578, p-issn:319-765x Volume 10, Issue 3 Ver II (May-Ju 014), PP 69-77 Commo Coupled Fixed Poit of Mappigs Satisfyig Ratioal Iequalities i Ordered Complex
More informationLecture 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 informationThe Inequalities of Khintchine and Expanding Sphere of Their Action
Usekhi Mat Nauk Vol 5 No 5 995 (3-6) (Russia) Russia Math Surveys Vol 5 No 5 995 (849-94) (glish) Prerit Ser No 4 994 Math Ist Aarhus Research Reort No 34 (995) Det Theoret Statist Aarhus To the ceteary
More informationRandom Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.
Radom Walks o Discrete ad Cotiuous Circles by Jeffrey S. Rosethal School of Mathematics, Uiversity of Miesota, Mieapolis, MN, U.S.A. 55455 (Appeared i Joural of Applied Probability 30 (1993), 780 789.)
More informationIntroduction to Probability. Ariel Yadin
Itroductio to robability Ariel Yadi Lecture 2 *** Ja. 7 ***. Covergece of Radom Variables As i the case of sequeces of umbers, we would like to talk about covergece of radom variables. There are may ways
More informationTHE GREATEST ORDER OF THE DIVISOR FUNCTION WITH INCREASING DIMENSION
MATHEMATICA MONTISNIGRI Vol XXVIII (013) 17-5 THE GREATEST ORDER OF THE DIVISOR FUNCTION WITH INCREASING DIMENSION GLEB V. FEDOROV * * Mechaics ad Matheatics Faculty Moscow State Uiversity Moscow, Russia
More informationMeasure 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 informationMath 155 (Lecture 3)
Math 55 (Lecture 3) September 8, I this lecture, we ll cosider the aswer to oe of the most basic coutig problems i combiatorics Questio How may ways are there to choose a -elemet subset of the set {,,,
More informationVECTOR SEMINORMS, SPACES WITH VECTOR NORM, AND REGULAR OPERATORS
Dedicated to Professor Philippe G. Ciarlet o his 70th birthday VECTOR SEMINORMS, SPACES WITH VECTOR NORM, AND REGULAR OPERATORS ROMULUS CRISTESCU The rst sectio of this paper deals with the properties
More informationSOME SEQUENCE SPACES DEFINED BY ORLICZ FUNCTIONS
ARCHIVU ATHEATICU BRNO Tomus 40 2004, 33 40 SOE SEQUENCE SPACES DEFINED BY ORLICZ FUNCTIONS E. SAVAŞ AND R. SAVAŞ Abstract. I this paper we itroduce a ew cocept of λ-strog covergece with respect to a Orlicz
More informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 3 9/11/2013. Large deviations Theory. Cramér s Theorem
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/5.070J Fall 203 Lecture 3 9//203 Large deviatios Theory. Cramér s Theorem Cotet.. Cramér s Theorem. 2. Rate fuctio ad properties. 3. Chage of measure techique.
More informationPRIME RECIPROCALS AND PRIMES IN ARITHMETIC PROGRESSION
PRIME RECIPROCALS AND PRIMES IN ARITHMETIC PROGRESSION DANIEL LITT Abstract. This aer is a exository accout of some (very elemetary) argumets o sums of rime recirocals; though the statemets i Proositios
More informationFall 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 informationk-generalized FIBONACCI NUMBERS CLOSE TO THE FORM 2 a + 3 b + 5 c 1. Introduction
Acta Math. Uiv. Comeiaae Vol. LXXXVI, 2 (2017), pp. 279 286 279 k-generalized FIBONACCI NUMBERS CLOSE TO THE FORM 2 a + 3 b + 5 c N. IRMAK ad M. ALP Abstract. The k-geeralized Fiboacci sequece { F (k)
More informationDefinition 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 informationA REMARK ON A PROBLEM OF KLEE
C O L L O Q U I U M M A T H E M A T I C U M VOL. 71 1996 NO. 1 A REMARK ON A PROBLEM OF KLEE BY N. J. K A L T O N (COLUMBIA, MISSOURI) AND N. T. P E C K (URBANA, ILLINOIS) This paper treats a property
More informationConfidence Intervals
Cofidece Itervals Berli Che Deartmet of Comuter Sciece & Iformatio Egieerig Natioal Taiwa Normal Uiversity Referece: 1. W. Navidi. Statistics for Egieerig ad Scietists. Chater 5 & Teachig Material Itroductio
More information1 Probability Generating Function
Chater 5 Characteristic Fuctio ad Probablity Covergece Theorem Probability Geeratig Fuctio Defiitio. For a discrete radom variable X which ca oly achieve o-egative itegers, we defie the robability geeratig
More informationDistribution 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 informationMath 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 informationON SOME NEW SEQUENCE SPACES OF NON-ABSOLUTE TYPE RELATED TO THE SPACES l p AND l I. M. Mursaleen and Abdullah K. Noman
Faculty of Scieces ad Mathematics, Uiversity of Niš, Serbia Available at: htt://www.mf.i.ac.rs/filomat Filomat 25:2 20, 33 5 DOI: 0.2298/FIL02033M ON SOME NEW SEQUENCE SPACES OF NON-ABSOLUTE TYPE RELATED
More informationMath 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 informationInfinite 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 informationMATH301 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 informationNotes 5 : More on the a.s. convergence of sums
Notes 5 : More o the a.s. covergece of sums Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: Dur0, Sectios.5; Wil9, Sectio 4.7, Shi96, Sectio IV.4, Dur0, Sectio.. Radom series. Three-series
More informationProduct 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 informationChapter 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 informationConcavity of weighted arithmetic means with applications
Arch. Math. 69 (1997) 120±126 0003-889X/97/020120-07 $ 2.90/0 Birkhäuser Verlag, Basel, 1997 Archiv der Mathematik Cocavity of weighted arithmetic meas with applicatios By ARKADY BERENSTEIN ad ALEK VAINSHTEIN*)
More informationAsymptotic distribution of products of sums of independent random variables
Proc. Idia Acad. Sci. Math. Sci. Vol. 3, No., May 03, pp. 83 9. c Idia Academy of Scieces Asymptotic distributio of products of sums of idepedet radom variables YANLING WANG, SUXIA YAO ad HONGXIA DU ollege
More information13.1 Shannon lower bound
ECE598: Iformatio-theoretic methods i high-dimesioal statistics Srig 016 Lecture 13: Shao lower boud, Fao s method Lecturer: Yihog Wu Scribe: Daewo Seo, Mar 8, 016 [Ed Mar 11] I the last class, we leared
More informationMi-Hwa Ko and Tae-Sung Kim
J. Korea Math. Soc. 42 2005), No. 5, pp. 949 957 ALMOST SURE CONVERGENCE FOR WEIGHTED SUMS OF NEGATIVELY ORTHANT DEPENDENT RANDOM VARIABLES Mi-Hwa Ko ad Tae-Sug Kim Abstract. For weighted sum of a sequece
More informationThe value of Banach limits on a certain sequence of all rational numbers in the interval (0,1) Bao Qi Feng
The value of Baach limits o a certai sequece of all ratioal umbers i the iterval 0, Bao Qi Feg Departmet of Mathematical Scieces, Ket State Uiversity, Tuscarawas, 330 Uiversity Dr. NE, New Philadelphia,
More informationLecture 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 informationThe natural exponential function
The atural expoetial fuctio Attila Máté Brookly College of the City Uiversity of New York December, 205 Cotets The atural expoetial fuctio for real x. Beroulli s iequality.....................................2
More information7 Sequences of real numbers
40 7 Sequeces of real umbers 7. Defiitios ad examples Defiitio 7... A sequece of real umbers is a real fuctio whose domai is the set N of atural umbers. Let s : N R be a sequece. The the values of s are
More informationLocal and global estimates for solutions of systems involving the p-laplacian in unbounded domains
Electroic Joural of Differetial Euatios, Vol 20012001, No 19, 1 14 ISSN: 1072-6691 UR: htt://ejdemathswtedu or htt://ejdemathutedu ft ejdemathswtedu logi: ft ocal global estimates for solutios of systems
More informationMASSACHUSETTS 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 informationABSOLUTE CONVERGENCE OF THE DOUBLE SERIES OF FOURIER HAAR COEFFICIENTS
Acta Mathematica Academiae Paedagogicae Nyíregyháziesis 6, 33 39 www.emis.de/jourals SSN 786-9 ABSOLUTE CONVERGENCE OF THE DOUBLE SERES OF FOURER HAAR COEFFCENTS ALEANDER APLAKOV Abstract. this aer we
More informationA Note on Bilharz s Example Regarding Nonexistence of Natural Density
Iteratioal Mathematical Forum, Vol. 7, 0, o. 38, 877-884 A Note o Bilharz s Examle Regardig Noexistece of Natural Desity Cherg-tiao Perg Deartmet of Mathematics Norfolk State Uiversity 700 Park Aveue,
More informationProbability 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 informationSequences and Limits
Chapter Sequeces ad Limits Let { a } be a sequece of real or complex umbers A ecessary ad sufficiet coditio for the sequece to coverge is that for ay ɛ > 0 there exists a iteger N > 0 such that a p a q
More informationThe random version of Dvoretzky s theorem in l n
The radom versio of Dvoretzky s theorem i l Gideo Schechtma Abstract We show that with high probability a sectio of the l ball of dimesio k cε log c > 0 a uiversal costat) is ε close to a multiple of the
More informationA NOTE ON BOUNDARY BLOW-UP PROBLEM OF u = u p
A NOTE ON BOUNDARY BLOW-UP PROBLEM OF u = u p SEICK KIM Abstract. Assume that Ω is a bouded domai i R with 2. We study positive solutios to the problem, u = u p i Ω, u(x) as x Ω, where p > 1. Such solutios
More informationOn equivalent strictly G-convex renormings of Banach spaces
Cet. Eur. J. Math. 8(5) 200 87-877 DOI: 0.2478/s533-00-0050-3 Cetral Europea Joural of Mathematics O equivalet strictly G-covex reormigs of Baach spaces Research Article Nataliia V. Boyko Departmet of
More informationMath 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 informationA 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 informationChapter 7 Isoperimetric problem
Chapter 7 Isoperimetric problem Recall that the isoperimetric problem (see the itroductio its coectio with ido s proble) is oe of the most classical problem of a shape optimizatio. It ca be formulated
More informationLECTURE 8: ASYMPTOTICS I
LECTURE 8: ASYMPTOTICS I We are iterested i the properties of estimators as. Cosider a sequece of radom variables {, X 1}. N. M. Kiefer, Corell Uiversity, Ecoomics 60 1 Defiitio: (Weak covergece) A sequece
More informationIntegrable 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 informationACO Comprehensive Exam 9 October 2007 Student code A. 1. Graph Theory
1. Graph Theory Prove that there exist o simple plaar triagulatio T ad two distict adjacet vertices x, y V (T ) such that x ad y are the oly vertices of T of odd degree. Do ot use the Four-Color Theorem.
More informationLecture 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 information6 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 informationSeunghee 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 informationECE 330:541, Stochastic Signals and Systems Lecture Notes on Limit Theorems from Probability Fall 2002
ECE 330:541, Stochastic Sigals ad Systems Lecture Notes o Limit Theorems from robability Fall 00 I practice, there are two ways we ca costruct a ew sequece of radom variables from a old sequece of radom
More informationElliptic Curves Spring 2017 Problem Set #1
18.783 Ellitic Curves Srig 017 Problem Set #1 These roblems are related to the material covered i Lectures 1-3. Some of them require the use of Sage; you will eed to create a accout at the SageMathCloud.
More informationAdvanced 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<, if ε > 0 2nloglogn. =, if ε < 0.
GLASNIK MATEMATIČKI Vol. 52(72)(207), 35 360 THE DAVIS-GUT LAW FOR INDEPENDENT AND IDENTICALLY DISTRIBUTED BANACH SPACE VALUED RANDOM ELEMENTS Pigya Che, Migyag Zhag ad Adrew Rosalsky Jia Uversity, P.
More informationMath 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 informationMath 113 Exam 3 Practice
Math Exam Practice Exam 4 will cover.-., 0. ad 0.. Note that eve though. was tested i exam, questios from that sectios may also be o this exam. For practice problems o., refer to the last review. This
More informationlim 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