The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria

Size: px
Start display at page:

Download "The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria"

Transcription

1 ESI The Erwin Schrodinger International Boltzmanngasse 9 Institute for Mathematical Physics A-090 Wien, Austria Large Deviations for Independent, Stationary, and Martingale Dierence Sequences Emmanuel Lesigne Dalibor Volny Vienna, Preprint ESI 667 (999) February 9, 999 Supported by Federal Ministry of Science and Transport, Austria Available via http//

2 LARGE DEVIATIONS FOR INDEPENDENT, STATIONARY, AND MARTINGALE DIFFERENCE SEQUENCES Emmanuel Lesigne, Dalibor Volny 3 February, 999 P n Abstract. Let ( i ) be a martingale dierence sequence and S n = i= i. We prove that if sup i E(e jij ) < then there exists c > 0 such that (S n > n) e?cn=3 ; this bound is optimal for the class of martingale dierence sequences which are also strictly stationary and ergodic. If the sequence ( i ) is bounded in L p, 2 p <, we get the estimation (S n > n) c n?p=2 which is again optimal for strictly stationary and ergodic sequences of martingale dierences. These results are compared with those for iid sequences; we give a simple proof that the estimate of Nagaev, Baum and Katz, (S n > n) = o(n?p ) for i 2 L p, p <, cannot be improved and that, reciprocally, it implies the integrability of j i j p? for all > 0. For an ergodic automorphism T of the probability space (; A; ) we show that there exists a dense set of f 2 L 0 for which the probabilities of (S n > n) go to zero arbitrarily slowly when n tends to innity (here i = f T i ). We study the asymptotic behaviour of (S n > n) for a generic f 2 L p 0.. Introduction. Let ( i ) be a sequence of real integrable random P variables with n zero means, dened on a probability space (; A; ) and S n = i= i. We shall study the asymptotic behaviour of the probabilities () (S n > nx); x > 0; n! The classical Cramer's theorem (cf. [C]) states that for independent and identically distributed (iid) random variables with nite exponential bounds the probabilities () are exponentially small. More precisely, Theorem. (Cramer). Let ( i ) be a sequence of iid random variables, E i = 0. There is equivalence between (i) There exists c > 0 such that Ee cj j <. (ii) For every x > 0 there exists c x > 0 such that (js n j > nx) = o(e?c xn ).? ( S n If the variables i are bounded, the estimates remain of the same type. Under suitable conditions, the convergence p n < x) =? (x)! where x = n and is a normal distribution function, holds. A survey of such results for independent random variables can be found e.g. in [P]. A nice result of the same type for bounded martingale dierence sequences has been given by Rackauskas ([Ra]). Typeset by AMS-TE

3 In this paper, we study the asymptotic behaviour of () with x constant. We are mostly concerned in the case when ( i ) is a sequence of martingale dierences, and in the case when ( i ) is a strictly stationary and ergodic sequence. Remark that we do not use the rate function to express this behaviour (cf., e.g. [Dem-Z]). It is well known (cf. e.g. [Ha-He]) that for strictly stationary and ergodic martingale dierence sequences classical limit theorems like the central limit theorem, the law of iterated logarithm, or the invariance principle remain valid in the same form as for iid sequences. As we shall see, this is not the case for large deviations estimates, except for bounded martingale dierences where a bound for probabilities () is given by Azuma's theorem ([A]). The case of unbounded martingale dierence sequences has remained up to now relatively neglected. Large deviations estimates for martingale dierence sequences are studied in the third section. We give new results under nite exponential moments or nite p-th moments ( p < ) hypothesis. In order to give a comparison we recall in the second section a classical result of Nagaev, Baum and Katz for iid random variables with nite p-th moments; we prove that their estimate cannot be substantially improved and that the asymptotic behaviour of probabilities () characterizes the integrability of j i j p?. In the third section, we also give large deviations estimates for stationary mixing sequences, which follow from martingale approximation. A strictly stationary and ergodic sequence of random variables ( i ) can always be represented under the form i = f T i where T is an ergodic bimeasurable and measure preserving automorphism of a probability space (; A; ), and f is a measurable function on. If f 2 L p (), then classical ergodic theorems of von P n? Neumann and Birkho say that the averages (=n)s n (f), S n (f) = f T i i=0, converge in L p and almost surely to E(f). U. Krengel ([Kr]) showed that the convergence can be arbitrarily slow and A. del Junco with J. Rosenblatt ([dj-ro]) showed that this event is generic. R. Burton and M. Denker ([Bu-Den], see also [V3]) proved that for a dense set of f 2 L 2 0 the central limit theorem holds; in [V], D. Volny showed that for any admissible sequence a n! there is a generic set of f 2 L p 0 such that the distributions of (=a n)s n (f) converge along subsequences to all probability laws. (We denote by L p 0 the space of zero mean functions in Lp ().) In the fourth section we deal with large deviations for ergodic sequences (f T i ) where f belongs to a generic subset of L p 0, p. 2

4 2. Independent random variables. Let ( i ) be a sequence of iid (independent identically distributed) random variables, E i = 0. We denote S n = n. By the weak law of large numbers we have, for any x > 0, lim n! (js nj > nx) = 0 The speed of convergence has been studied by many authors. The classical result concerning the case when random variables i have nite exponential moments has been recalled in Introduction. Here we shall consider the case when the random variables have a nite moment of order p. Theorem 2. (Nagaev, Baum and Katz). Assume E(j p j) < where p. Then (js n j > nx) = o(n?p ) This theorem can be found in [N], [Ba-Ka] and in the complements in Chapter 4 in [P] or in Chapter 2 of [Re]. Conversly, the speed of the convergence of (js n j > nx) characterizes the moments of i Theorem 2.2. Let p >. If there exists x > 0 such that (js n j > nx) = O(n?p ), then 8 > 0; E(j j p? ) < Theorem 2.2 is a strengthening of a result of P. Revesz who proved that if p > 2 and (js n j > nx) = O(n?p ) for some x > 0, then E(j j p?? ) < for all > 0. We shall also show that the estimate given in Theorem 2. cannot be essentially improved Theorem 2.3. Let p and (c n ) be a real positive sequence going to zero. For any increasing sequence of integers (n k ) with P k= c n k < ; there exists a probability law on the real line such that for iid random variables i with the distribution we have E(j i j p ) <, E i = 0 and as k!. n p? k c nk (js nk j > n k )! In the proofs of Theorem 2.2 and Theorem 2.3 we shall use the following lemma which occurs as an exercise in [Du]. Lemma 2.4. Let ( i ) be a sequence of integrable iid random variables with E i = 0. Then lim inf n! (js n j > n) n(j j > 2n) Proof. Let us denote A k = f k > 2n and S n? k >?ng. We have (S n > n) n k= n [ k= (A k )? A k jkn (A j \ A k ) = n(a )? 3 n(n? ) (A \ A 2 ) 2

5 We denote by a random variable distributed as the i 's. >From and (A ) = ( > 2n) (S n? >?n) ( > 2n) (js n? j < n) (A \ A 2 ) ( > 2n and 2 > 2n) = (( > 2n)) 2 ((jj > 2n)) 2 we deduce (js n j > n) n(jj > 2n)P (js n? j < n)? n 2 ((jj > 2n)) 2 The statement of the Lemma now follows from the last inequality, and lim n! (js n?j < n) = lim n(jj > 2n) = 0 n! Proof of Theorem 2.2. Let x > 0 be the x from the statement of the Theorem. By Lemma 2.4 we then have Therefore, for any > 0, n (jj > 2nx) = O(n?p ) n p?? P (jj > 2nx) < ; hence E(jj p? ) <. The proof shows that the the claim E(jj p? ) < can be strengthened; we can, e.g. get E(jj p =(log + jj) + ) <, where log + x is dened as log(max(x; e)). Proof of Theorem 2.3. By Lemma 2.4 it suces to prove the existence of a probability law with zero mean and nite p-th moment, for which lim k! c nk n p k ([2n k; +[) = Let ( k ) be a sequence of positive numbers such that c k! nk 0 and < k and dene k a = k c nk k n p k ; a k = c n k 2a k n p ; k = k a k ( 2nk +?2nk ) ( x denotes the Dirac measure P at point x). By its denition, is a symmetric probability law and, since k a kn p k <, has a nite p-th moment. But we have n p k c ([2n k; +[) > n p k nk c a k?! nk 4

6 3. Martingale dierence sequences. By ( i ) we denote a (not necessarily stationary) martingale dierence sequence and we set S n = n i= We study the speed of convergence of (S n nx) to zero in three dierent cases. The case when variables i are uniformly bounded is well known Azuma's inequality gives a speed similar to the one in Cramer's theorem on iid sequences. But in the case of exponential nite moments, the results are dierent from those in the iid case, and we prove that there exists a constant c such that i (S n nx) exp(?cn =3 ) In the case of nite p-th moments, we obtain also a new estimate of the type (S n nx) cn?p=2 In these three cases, we show that the estimates are optimal, even in the restricted class of strictly stationary and ergodic sequences of martingale dierences. Strictly stationary sequences will be always represented under the form ( i ) = (f T i ) where T is a measurable and measure preserving transformation of onto itself. Theorem 3. (Azuma [A]). Let ( i ) be a sequence of martingale dierences. If j i j < a < for all i, we have (2) (S n nx) e?nx2 =2a 2 ; n = ; 2; For the reader's convenience, Azuma's proof of this useful inequality is recalled at the end of this section. The fact that this estimation is optimal is already known for iid sequences. Theorem 3.2. Let ( i ) be a martingale dierence sequence such that Ee j ij K < for all i. For every x; > 0, for any large enough n,? 2 (? )x2=3 n =3 (3) (js n j > nx) < exp In every ergodic dynamical system of positive entropy there exists c > 0 and a function f, Ee jfj <, such that (f T i ) is a martingale dierence sequence and (4) (js n (f)j > n) > e?cn=3 for innitely many n (S n (f) = P n i= f T i ). The proof of the rst part of this theorem is based on Azuma's inequality and truncation arguments. In this proof we shall give a precise and non asymptotic upper bound for (js n j > nx). In the proof of the second part we use an estimation of \moderate variation" for sums of iid random variables which is due to Cramer. We give this estimation as it appears in Feller's book ([F], second edition, volume II, VI.7). 5

7 Theorem 3.3 (Cramer). Let (Y i ) be sequence of iid random variables, with EY i = 0 and E(e P ty i ) < for t real in a neighbourhood of zero. Denote 2 = E(Yi 2 ), and n R n = Y i= i. There exists a function dened and analytic on a neighbourhood of zero such that, for any real sequence (x n ), if x n! and x n = o( p n), then p (R n > x n n) = Z + p 2 x n exp? u2 2 2 du exp x 3 pn n xn xn pn + O pn Corollary 3.4. Under the hypotheses of Theorem 3.3, for any > 0, for all large enough n, exp?? 2 ( + p 2 )x2 n < R n > x n n < exp? 2 (? 2 )x2 n Let us consider now the case of martingale dierence sequences with nite moment of order p. Theorem 3.5. Let ( i ) be a martingale dierence sequence where i 2 L p, 2 p <, k i k p < M < for all i. Let x > 0. Then (5) (js n j > nx) (8pq =2 ) p M p x p n p=2 where q is the real number for which =p + =q =. In every ergodic dynamical system (; A; ; T ) of positive entropy and for every sequence (b n ) of positive numbers going to zero there exists f 2 L 2 0 such that (f T i ) is a martingale dierence sequence and (6) lim sup n! n p=2 b n (js n (f)j > n) = The proof of the rst part of this theorem is based on classical norm estimates for martingale dierence sequences. In the proof of the second part we use the Central Limit Theorem for iid sequences and Azuma's inequality. Via the so-called Gordin's decomposition, Theorem 3.5 have wide applications to the study of speed in Ergodic Theorem (see Corollary 3.8 below). Remark. Theorem 2.3 shows that for p = there exist iid sequences for which the convergence (js n (f)j > n)! 0 can be arbitrarily slow (from the point of view of subsequences). We rst prove (3) and (5). argument. In the proofs of (4) and (6) we use a common Proof of (3). For a martingale dierence sequence ( i ) with j i j, the Azuma inequality gives, for any x > 0, (7) (js n j > nx) 2 exp(?nx 2 =2) 6

8 Let ( i ) be a martingale dierence sequence satisfying the assumptions of Theorem 3.2 and denote by (F i ) its ltration. Let us x a > 0. For n = ; 2; and i n dene Y n;i = i (ji jan =3 )? E( i (ji jan =3 )jf i? ) ; Z n;i = i (ji j>an =3 )? E( i (ji j>an =3 )jf i? ) ; S 0 n = S 00 n = n n i= i= Y n;i ; Z n;i (Y n;i ) and (Z n;i ) are martingale dierence arrays and, because ( i ) is a martingale dierence sequence, i = Y n;i + Z n;i ( i n). Let us x t 2 (0; ). For every x > 0, (8) (js n j > nx) (js 0 nj > nxt) + (js 00 nj > nx(? t)) We have jy n;i j 2an =3 for i n, hence by using (7) we get (9) (jsnj 0 js 0 > nxt) = n j 2an > nxt 2 exp? t2 x 2 =3 2an =3 8a 2 n=3 Let F i (x) = (j i j > x). >From Ee j ij K it follows that F i (x) Ke?x for all x 0. Then? EZn;i 2 = E ( i (ji j>an )) 2?? =3 E (E( i (ji jan )jf =3 i? )) 2? E ( i (ji j>an )) 2 Z =? =3 x 2 df i (x) = (an =3 ;+) Z!? lim 2xF i (x) dx (an =3 ;M ] xe?x dx = K a 2 n 2=3 + 2an =3 + 2 e?an=3 an =3 M! M 2 F i (M)? a 2 n 2=3 F i (an =3 )? Ka 2 n 2=3 e?an=3 + 2K Z Thus we have E(Sn) 00 2? nk a 2 n 2=3 + 2an =3 + 2 e?an=3 and hence (0) (js 00 nj > nx(? t)) n 2 x 2 (? t) nk a 2 n 2=3 + 2an = e?an=3 = K a x 2 (? 2 n?=3 + 2an?2=3 + 2n? e?an=3 t) 2 In view of (9) and (0) we choose a so that t2 x 2 8a 2 () (js 00 nj > nx(? t)) K 4 x 2 (? t) 2 (tx) 4=3 n?=3 + (tx) 2=3 n?2=3 + 2n? exp 7 = a = 2 (tx)2=3. We obtain? 2 (tx)2=3 n =3

9 >From (8), (9), and () we deduce that for every x > 0 and every t 2 (0; ), (js n j > nx) < 2 + K (? t) ( 2 4 t4=3 x?2=3 n?=3 + t 2=3 x?4=3 n?2=3 + 2x?2 n? ) e? 2 t2=3 x 2=3 n =3 We can choose the parameter t arbitrarily close to, and (3) follows. Proof of (5). By Burkholder's inequality (cf. [Ha-He], Theorem 2.0), we have E i p (8pq =2 ) p E n? i=0 Set Y i = 2 i. By a convexity inequality we have therefore It follows E n? i=0 Y i (!) n? 2 p n? i n? p (8pq =2 ) p n p=2? n? i=0 (S n > nx) Z js n j>nx i=0 i=0 n? i 2 i=0 p=2 2=p Y p=2 i (!) Ej i j p n p=2 (8pq =2 ) p M p js n j p x p n p d (8pq=2 ) p M p x p n p=2 This nishes the proof of (5). Remark that for p = 2 we can use, instead of Burkholder's inequality, the only orthogonality of i. In the proof of (4) and (6) we shall use the following lemma. Lemma 3.6. Let (; A; ; T ) be an ergodic probability measure preserving dynamical system of positive entropy. There exist two T -invariant sub--algebras B and C of A and a function g on such that - the -algebras B and C are independent; - the function g is B-measurable, takes values -, 0 and, has zero mean and the process (g T n ) is independent; - the dynamical system (; C; ; T ) is aperiodic. Proof. Let h be the entropy of (; A; ; T ). Let a 2 (0; ] be such that h 0 =?(? a) log 2 (? a)? a log 2 (a=2) < h ; and let us consider the full shift on three letters (?; 0; ) with the innite product measure (a=2;? a; a=2) Z. This dynamical system will be denoted by S. It is a Bernoulli measure preserving system of entropy h 0. Consider another Bernoulli 8

10 measure preserving system of entropy smaller than h?h 0, and denote it by S 2. The product system S S 2 is Bernoulli and its entropy is smaller than h. Thus, by Sinai's Theorem (see e.g. [Sh]), it is a factor of the system (; A; ; T ). So we have in this system a copy of S and a copy of S 2 which are independent. This gives the -algebras B and C. The function g is obtained by lifting on the zero coordinate function of the shift S. Proof of (4). Let (n k ) k be an increasing sequence of integers satisfying some quick growth condition that will be specied in the sequel. Following Lemma 3.6 we consider two sub--algebras B and C and a function g. The fact that the dynamical system (; C; ; T ) is aperiodic guarantees the existence of a sequence (A k ) k in C such that the A k 's are two by two disjoint, (2) (A k ) < k?2 exp(?n =3 k ); and (3) n k \ j= T?j (A k ) > 2 k?2 exp(?n =3 k ) (It is always possible to construct such A 0 ks, using for example thin vertical slices in some Rokhlin tower.) We dene and we claim that f = g k= n =3 k A k ; (4) E(e jfj ) < ; (5) E f T n jf n? = 0 where F k is the -algebra generated by f T i, i k, (6) (js nk (f)j > n k ) > exp(?cn =3 k ) for some constant c. >From the denition of f and from (2) we get E(e jfj ) k= exp(n =3 k )k?2 exp(?n =3 k ) which proves (4). 9

11 Denote f = P k= n=3 k A k. By the assumptions, the random variable g T n is independent of the -algebra generated by C and the functions g T i, i < n, hence E(f T n jc _ F n? ) = ( f T n ) E(g T n jc _ F n? ) = 0. This implies (5). In order to prove (6) we dene f k = gn =3 k A k ; f + k = g j>k n =3 j Aj and f k? = g n =3 j Aj j<k We have (js nk (f)j > n k ) (js nk (f k )j > 3n k )? (js nk (f + k )j > n k)? (js nk (f? k )j > n k) From js nk (f k )j n =3 k js n k (g)j Yn k j= T?j A k we deduce, by using independence of the -algebras B and C, that (js nk (f k )j > 3n k )? \ n k j= T?j A k jsnk (g)j > 3n 2=3 k We denote by 2 the variance of g. From (3) and Corollary 3.4 (applied with x n = 3n =6 and = ), we get (7) (js nk (f k )j > 3n k ) 2k 2 exp(?n=3 k ) exp? 9 2 n=3 k Let us x a positive number c > + 9= 2. Now, we impose two conditions on the growth of the numbers n k (8) (9) j=k+ n k 2n 2=3 k? > cn =3 k ; n =3 j j?2 exp(?n =3 j ) < exp(?cn =3 k ) By noticing that jf? k j n =3 k? and using Azuma's inequality and (8), we obtain (20) (js nk (f? k )j > n k) 2 exp(?n k =2n 2=3 k? ) 2 exp(?cn=3 k ) We have EjS nk (f + k )j n kejf + k j = n k and thanks to (9), this implies that j=k+ n =3 j j?2 exp(?n =3 j ) (2) (js nk (f + k )j > n k) EjS n k (f + k )j n k < exp(?cn =3 k ) 0

12 >From (7), (20) and (2) we conclude that, for all large enough k, which concludes the proof of (4). (js nk (f)j > n k ) exp(?cn =3 k ) ; Proof of (6). As in the preceding proof, we use Lemma 3.6 and consider -algebras B, C and a function g. Let us x a sequence (c k ) k of positive numbers such that Pk c=p k <. By the Central Limit Theorem we know that there exists c > 0 such that, for all large enough n, (22)? js n (g)j > 2 p n > c We consider an increasing sequence of integers (n k ) k satisfying the following growth conditions (23) exp? 2 n k j<k p nj?2 = o(c k n?p=2 k ) ; (24) j>k c j n?p=2 j = o(c k n??p=2 k ) ; (25) b nk = o(c k ) By the Rokhlin lemma there exists a sequence (F k ) of elements of C, such that, for each k, the sets F k ; T? F k ; ; T?2n k+ F k are two by two disjoints and (F k ) = c k 2n +p=2 k Denote A k = 2n [ k i= T?i F k ; f k = gn =2 k A k ; f = k= f k We have kf k k p c =p k hence f 2 L p. The same argument as in the proof of (4) shows that (f T i ) is a martingale dierence sequence. We have? \ n k i= T?i A k 2 (A k) = c k 2n p=2 k

13 and js nk (f k )j p n k js nk (g)j Yn k i= The independence of C and process (g T i ) implies T?i A k (js nk (f k )j > 2n k ) (js nk (g)j > 2 p n k )? \ n k i= T?i A k Under condition (22) we conclude that (26) (js nk (f k )j > 2n k ) c c k 2n p=2 k The sequence (( P k? j= gf j) T i ) i0 is a martingale dierence sequence and j P k? gf j= jj P k? p j= nj. By Azuma's inequality we have Snk k? k? j= and, by (23), we conclude that (27) We have (28) S nk gf j nk exp? 2 n k j= j= Snk k? gf j nk = o(c k n?p=2 k ) j=k+ gf j 6= 0 nk by (24). >From (26), (27) and (28) we deduce that lim inf k! Because of (25), this implies and this concludes the proof of (6). j=k+ n p=2 k c k (js nk (f)j > n k ) > 0 n p=2 k b nk (js nk (f)j > n k )! p nj?2 ; (A j ) = o(c k n?p=2 k ) ; We explain now how, via martingale approximation, Theorem 3.5 have applications to the study of large deviations for some classes of stationary processes. This type of martingale approximation rst appears in [G], and the following theorem can be found in [V2]. Let (; T ; ) be a probability space and T a one-to-one, bimeasurable and measure preserving transformation of this space. Let M be a sub--algebra such that M T? M. Let us denote M _ = T i M i2z T i M and M? = \ i2z Let p and f 2 L p () be such that f is M -measurable and E(fjM? ) = 0. (Remark that if (; T ; ; T ) is a K-system and if M is well chosen, then any f 2 L p 0 () satises these two conditions.) 2

14 Theorem 3.7. The condition (29) n=0 E(f T n jm) and n=0 [f T?n? E(f T?n jm)] converge in L p is equivalent to the existence of u; m 2 L p such that (m T i ) is a martingale dierence sequence and f = m + u? u T (In [V2], the proof is formulated for p = ; 2 but in fact it works for all p.) >From Theorem 3.5 and Theorem 3.7 we deduce the following result. Corollary 3.8. Let p <. Under condition (29), we have (js n (f)j > n) = O n p=2 Proof of Corollary 3.8. We start from the decomposition f = m + u? u T given by Theorem 3.7. >From the integrability of juj p we deduce that hence lim x! x(jujp > x) = 0 ; (juj > n=3) = o n p Theorem 3.5 gives us (js n (m)j > n=3) = O n p=2 But we have js n (f)j js n (m)j + juj + ju T n j, hence (js n (f)j > n) (js n (m)j > n=3) + 2(juj > n=3) This concludes the proof of Corollary 3.8. Remark. As a special case let us consider a -mixing or an -mixing sequence (f T i ), f 2 L p, p > (for the denition see e.g. [I-L] or [Ha-He] or [Y]). We shall denote M k = (f T i i k) and M k = (f T i i k). Let (ft i ) be -mixing. By [I-L] (cf. also [Ha-He] or [Y]), for q > 0, =p+=q =, g 2 L p 0 measurable w.r.t. M 0, and h 2 L q 0, measurable w.r.t. Mk, k, we have Notice that q = p Ejghj 2(k) =p kgk p khk q p? hence jfjp? 2 L q. Thus, for k we have ke(fjm?k )k p p = EjE(fjM?k )j p = E E(fjM?k ) f p? 3 2(k) =p ke(fjm?k )k p kf p? k q

15 hence Therefore, if f 2 L p and ke(fjm?k )k p C(f)(k) k= (k) p(p?) < ; p(p?) then (29) is guaranteed with M = M 0 and consequently (js n (f)j > n) = O n p=2 Let (f T i ) be an -mixing sequence. By e.g. [Hall-Heyde, Corollary A.2] or [Y] for q > 0, =p + =q <, there exists a c > 0 such that for g 2 L p 0 measurable w.r.t. M 0, and h 2 L q 0, measurable w.r.t. Mk, k, we have Ejghj c(k)? p? q kgk p khk q Suppose that f 2 L p+h, p 2, h > 0, and q = p+h. We then have p? ke(fjm?k )k p p = EjE(fjM?k )j p c ke(fjm?k )k p kjfj p? k q [(k)] h(p?) p(p+h) (note that h(p?) =?? p(p+h) p q ), hence ke(fjm?k )k p C(f)[(k)] Therefore, if f 2 L p+h, p 2 and h > 0, k= h [(k)] p(p+h) < h p(p+h) guarantees (29), hence (js n (f)j > n) = O n p=2 Two comments on related results.. In [V4] a nonstationary version of Theorem 3.7 is shown. As a consequence one can prove the preceding estimates for or mixing sequences ( i ) where the assumption of stationarity is replaced by uniform boundedness of the L p norms. 2. Let us notice that for any -mixing stationary sequence (f T i ) of random variables (with an arbitrarily slow decay of (k)) where f is integrable and bounded from above, an exponential bound for the probabilities (S n (f) > nx) was found by R. H. Schonmann [Sc] (cf. also [Br]). W. Bryc and A. Dembo [Br-Dem] proved the Varadhan's version of the large deviation principle for and -mixing sequences with \hyperexponential" decay of (k) and (k); a counter-example shows that an exponential decay is not sucient. In [Br-Dem], one can also nd more references on large deviations for mixing processes. Now we recall the proof of Azuma's inequality. 4

16 Proof of Theorem 3.. Let ( n ) be a martingale dierence sequence uniformly bounded by. If t 2 (?; +) et x 2 [?; ], we have tx = + x 2 t +? x (?t) 2 and, by convexity of the exponential function, this gives e tx cosh t + x sinh t Therefore E? e ts n E n Y k=(cosh t + k sinh t)! We develope the product, use the martingale property and naly observe that all the terms vanish except one and But cosh t e t2 =2 so we have Now, for any x > 0, P (S n nx) E E? e ts n (cosh t) n E? e ts n e nt 2 =2? e x(s n?nx) = e?nx2 E e n xs e?nx 2 e nx2 =2 = e?nx2 =2 4. Genericity in dynamical systems. Let (; A; ) be a Lebesgue probability space and T a bimeasurable and measure preserving one-to-one transformation of. We shall suppose that the dynamical system (; A; ; T ) is ergodic and aperiodic. For a measurable function f, the sequence (f T i ) is a strictly stationary process. We set S n (f) = n? i=0 f T i ; n = ; 2; If f is integrable and has zero mean, then, by the Ergodic Theorem, we know that n S n(f)! 0. For an arbitrary sequence of real numbers (b n ) going to innity, we shall study the behaviour of the sequence? b n (S n (f) > n). We shall see that, in a great variety of functional spaces, for a generic f, the sequence? b n (S n (f) > n) does not go to zero; and we shall see that this sequence goes to innity for a wide class of functions f. In Theorems 4. and 4.2 we recall some known results on a related problem, namely the behaviour of (S n (f) > nc n ) where c n! 0, for a generic function f. 5

17 Theorem 4.. Let (c n ) n>0 be a sequence of real numbers, c n! 0. Then for every p there exists a dense G set G L p 0 such that for every f 2 G lim sup (js n (f)j > nc n ) = n! Theorem 4. is well known and can be proved by the following argument denote by H n the set of functions f in L p 0 for which there exists k > n such that (js k(f)j > kc k ) >? n ; this set H n is open in L p 0, and by using Rokhlin lemma, it can be shown that this set is dense; the intersection of these sets is a dense G satisfying the announced property. Next theorem, from [V], is a renement of Theorem 4.. Theorem 4.2. Let c n! 0, nc n!. a dense G set G L p 0 Then for every p there exists such that for every f 2 G and every probability measure on the real line there exists a sequence n k! such that the distributions of (=n k c nk )S nk (f) weakly converge to. Let us now come back to our original question. P Theorem 4.3. Let (a n ) be a sequence of positive real numbers such that a n =. Let be a strictly positive continuous increasing function on ]0; ]. For every p < there exists a dense G set G L p 0 such that for every f 2 G This result implies n= a n ((S n (f) > n)) = Theorem 4.4. Let (b n ) n>0 be a sequence of positive real numbers, b n!. For every p < there exists a dense G set G L p 0 such that for every f 2 G lim sup n! b n (S n (f) > n) = Proof of Theorem 4.4. If b n!, there exists an increasing function satisfying the hypotheses of Theorem 4.3 and such that n= p bn < We choose a n =, n = ; 2;. By Theorem 4.3 there exists a dense G set G L p 0 such that for every f 2 G n= If f 2 G then, innitely often, (S n (f) > n) = ((S n (f) > n) > p bn 6

18 Lemma 4.5. For any real 2]0; [ 4 and any integer k > 0 there exists h 2 L 0 such that (S j (h) 2j; j k) and Z jhj p d 2 p+2 Proof of Lemma 4.5. We consider a Rokhlin tower A; T? A; ; T?2k+ A and a measurable subset B disjoint from the tower such that (A) = =k and (B) = 2. We consider the function h equal to 2 on the tower, to?2 on B and 0 everywhere else. The event fs j (h) 2j; j kg is contained in the union of the T?j A for k j < 2k. Proof of Theorem 4.3. There exists a sequence ( n ) of positive numbers such that n! 0 and By H n we denote the set of all f 2 L p 0 n= k? a j (S j (f) > j) > j= a n n = for which there exists a k > n such that k j= a j j Each set H n is open in L p 0. We shall prove that it is dense. The intersection of all the H n will be the set G satisfying the property described in the Theorem. The functions g? g T with g 2 L form a dense subset of L p 0. Let us show that H n intersects each neighbourhood of any of them. We consider an integer n, a bounded measurable function g and a positive real number. We x an integer ` such that ` > 2kgk and j > ` =) j < 2 () and an integer k such that k > n ; ` and ` a j j < k j= j=`+ a j j By Lemma 4.5 we know that there exists a function h such that Z jhj p d 2 p+2 ; and (S j (h) 2j) for all j k Let For j > ` we have f = h + g? g T? (S j (f) j)? (S j (h) 2j) () 7

19 Therefore k? a j (S j (f) j) j= k j=`+ a j () > 2 k j=`+ a j j > k j= a j j Thus f 2 H n and kf? (g? g T )k p = khk p (2 p+2 ) =p. It is an easy observation that Theorem 4.4 cannot be extended to L 0. However the two next theorems give results of the same type for bounded functions. Remark. In the proof of Theorem 4.3 we used the fact that in the spaces L p 0, p <, the coboundaries g? g T with g 2 L form a dense subset. This, however, is not true for p = in any ergodic and aperiodic dynamical system and for each p > 0 there exists a function f 2 L p 0 and > 0 such that for each coboundary g? g T 2 L 0 with g 2 L p we have kf? (g? g T )k > (cf., e.g., [K] or [V-W]). In [V-W] it is shown that for each p <, the closure (in L 0 ) of the set of all g? g T 2 L 0, g 2 L p, can be characterized by the rate of decay of the probabilities of (js n (f)j > nc). Let B be the set of real measurable functions f bounded by one and of zero mean, endowed with the distance d(f; g) = inffa > 0j(jf? gj > a) < ag The metric space (B; d) is complete. Let M be the set of elements of B taking only values? and. This set is closed in (B; d). Theorem 4.6. Let (b n ) n>0 be a sequence of positive real numbers, b n!. For every c <, in the space (B; d) there is a dense G set of functions f satisfying lim supb n (js n (f)j > cn) = n! In the space (M; d) there is a dense G set of functions f satisfying lim sup b n (js n (f)j = n) = n! The proof of Theorem 4.6 can be done by using the ideas of Theorem 4. or [V] and is left to the reader. Theorem 4.7. Let (b n ) n>0 be a sequence of positive real numbers, b n!. In the Banach space L 0, there exists a dense G subset G such that for every f 2 G (30) lim sup n! b n(js n (f)j > cn) = holds for some c > 0. Proof of Theorem 4.7. For c > 0 denote V c = ff lim sup n! b n(js n (f)j > c) = g ; V c;n = ff 9k n; b k (js k (f)j > c) > ng 8

20 We have V c = T V c;n and, every f 2 S n= `= S Each set V c;n is open in L 0. Therefore Let f 2 L 0 V =` satises (30). `= V =` is G. and > 0. Theorem 4.6 guarantees the existence of a function g in V 2 bounded by 3. If f =2 V, we have f + g 2 V. This proves that V is 3-dense S in L 0. Therefore V =` is dense. `= We now present the \liminf" type result. Theorem 4.8. Let (b n ) n>0 be a sequence of realr numbers, b n!. There exists a bounded measurable function f on such that f d = 0 and lim inf n! b n(s n (f) n) = Remarks.. The set of functions f satisfying the preceding property is in fact dense in every L p 0 (), p <. 2. However, in each L p (), there is a dense G 0 set of functions f such that lim inf n! b n(s n (f) n) = 0 Both statements follow from the fact that the set of coboundaries g?g T with g measurable and bounded is dense in each L p 0, p <. In the proof of the second remark we need a Baire argument like in the proof of Theorem 4. or Theorem 4.3. Lemma 4.9. Let n; m be positive integers with n < m. Let A; T A; ; T 4n? A and B; T B; ; T 2m? B be two Rokhlin towers. Denote A = A [ T A [ [ T 4n? A and B = B [ T B [ [ T 2m? B. There exists A 0 A such that and, for each n k < 3n, (A 0 ) (A) 4(B) (A) B \ T k A T k A 0 Lemma 4.9 says that the intersection of the tower B with the 2n middle oors of the tower A is contained in a subtower of A of measure smaller than 4(B). Here, by \subtower of A", we mean a tower A 0 ; T A 0 ; ; T 4n? A 0 with A 0 A. A proof of this lemma is given at the end of the paper. Proof of Theorem 4.8. The sequence (b n ) is given. Let (n k ) k be an increasing sequence of positive integers such that b n > 4 ; 00b nk < b nk+ and b nk = inf nn k b n To each k p we associate a Rokhlin tower A k ; T A k ; ; T 8n k+? A k such that (A p k ) = = b nk and a set B k in A S k disjoint from A j= j such that (B k ) = S = b nk. (As before we use the notation A k = T j A k ). 0j<8n k+ 9

21 and We recursively dene a sequence (f k ) of functions by f = f k = 8 >< > 8 >< > on S 4n 2? j=0 T j A ;? on S 8n 2? j=4n 2 T j A ; 0 otherwise ; on S 4n k+? j=0 T j A k ;? on S 8n k+? j=4n k+ T j A k ; f k? on (A k [ B k ) c ; c k on B k ; where the constant value c k of the function f k on B k is chosen so that the mean value of f k on is zero. Because (f k 6= f k? ) p 2= b nk is a summable sequence, the sequence (f k ) converges almost surely to a function f. Notice that all these functions are bounded by one. By construction (recall that, for ` > k; B` \ A k = ;), we have? ff 6= fk g \ A k [ Let k < `. By Lemma 4.9 we know that the trace of the tower A` on the 4n k+ middle oors of the tower A k is contained in a subtower of A k of measure lower than 4(A`). This implies that the set ff 6= f k g \ 6n k+? [ is contained in a subtower A 0 k of A k such that `>k A` j=2n k+ T j A k (A 0 k) 4 `>k (A`) We have (A 0 k) 4 `>k 0 k?`(a k ) < 2 (A k) Let n be a positive integer and k be such that n k n < n k+. We have On the set S n (f k ) = n on the set 3n k+? [ 4n k+?n? [ j=0 j=2n k+ T j (A k n A 0 k); which is included in the preceding one, we have S n (f) = S n (f k ) = n 20 T j A k

22 Thus (S n (f) = n) n k+ (A k n A 0 k) 2 n k+(a k ) = 6 p b nk 6 p b n which proves the theorem. Proof of Lemma 4.9. Dene A 0 = 3n? [ k=n? A \ T?k B In the tower A; T A; ; T 4n? A, above each point of A 0 there is at least n points of B. Indeed, if! 2 A 0, there exists k between n and 3n? such that T k! 2 B. Then, either T k+j! 2 A \ B for each 0 j n or T k?j! 2 A \ B for each 0 j n. (Here we used two facts if! 0 2 B, then T j! 0 2 B either for any 0 j n or for any 0 j?n; and if! 0 2 T k A for one k between n and 3n?, then T j! 0 2 A for any j between n and?n.) We claim that this implies n(a 0 ) (B) Here is a proof of this claim for each subset K of E = f0; ; ; 4n? g, dene A 0 K = f! 2 A 0 j T k (!) 2 B for k 2 K and T k (!) =2 B for k 2 E n Kg The sets A 0 K form a partition of A0. We have A 0 K = ; if card(k)< n. Sets T k A 0 K, for K E and k 2 K, are two by two disjoint and included in B. Thus (B) (T k A 0 K) n KE k2k KE References (A 0 K) = n(a) [A] Azuma, K., Weighted sums of certain dependent random variables, T^ohoku Math. J. 9 (967), [Ba-Ka] [Br] Baum, L.E. and Katz, M., Convergence rates in the law of large numbers, Trans. Amer. Math. Soc. 20 (965), Bryc, On large deviations for uniformly strong mixing sequences, Stochastic Processes and their Applications 4 (992), [Br-Dem] Bryc, W. and Dembo, A., Large deviations and strong mixing, Ann. Inst. Henri Poincare 32 (996), [Bu-Den] [C] [dj-ro] Burton, R. and Denker, M., On the central limit theorem for dynamical systems, Trans. Amer. Math. Soc. 302 (987,), Cramer, H., Sur un nouveau theoreme-limite de la theorie des probabilites, Actualites Scientiques et Industrielles (Hermann, Paris) 736 (938), del Junco, A. and Rosenblatt, J., Counterexamples in ergodic theory and in number theory, Math. Ann. 245 (979),

23 [Dem-Z] [Du] [F] [G] [Ha-He] [I-L] [K] Dembo, A. and Zeitouni, O., Large Deviations Techniques and Applications, Jones and Bartlett, Boston, 993 (Or Applications of Mathematics 38, Springer, 998). Durrett, R., Probability Theory and Examples, Wadsworth and Brooks/Cole, Pacic Grove, California, 99. Feller, W., An Introduction to Probability Theory and its Applications, Vol. II, second edition, John Wiley, New York, 97. Gordin, M.I., The central limit theorem for stationary processes, Soviet Math. Doklady 0 (969), Hall, P. and Heyde, C.C., Martingale Limit Theory and its Applications, Academic Press, New York, 980. Ibragimov, I.A. and Linnik Y.V., Independant and Stationary Sequences of Random Variables, Wolters-Noordho, Groningen, 97. Katok A., Constructions in Ergodic Theory, An unpublished manuscript. [Kr] Krengel, U., On the speed of convergence in the ergodic theorem, Monatsh. Math. 86 (978), 3-6. [N] Nagaev, S.V., Some limit theorems for large deviations, Theory of Probab. and its Appl. 0 (965), [P] Petrov, V.V., Limit Theorems of Probability Theory, Oxford Science Public., 995. [Ra] Rackauskas, A., On probabilities of large deviations for martingales, Lietuvos Matematikos Rinkinys 30 (990), [Re] Revesz, P., The Laws of Large Numbers, Academic Press, New York and London, 968. [Sc] Schonmann, R.H., Exponential convergence under mixing, Prob. Th. Rel. Fields 8 (989), [Sh] Shields, P., The Theory of Bernoulli Shifts, University of Chicago Press, 973. [V] [V2] [V3] [V4] [V-W] [Y] Volny, D., On limit theorems and category for dynamical systems, Yokohama Math. J. 38 (990), Volny, D., Approximating martingales and the central limit theorem for strictly stationary processes, Stochastic Processes and their Applications 44 (993), Volny, D., Invariance principles and Gaussian approximation for strictly stationary processes (to appear(trans. Amer. Math. Soc.)). Volny, D., Martingale approximation of stochastic processes and limit theorems, In preparation. Volny, D. and Weiss, B., Coboundaries in L 0, In preparation. Yoshihara, K., Summation Theory for Weakly Dependent Sequences, Sanseido, Tokyo, 992. Lesigne Laboratoire de Mathematiques et Physique Theorique UPRES-A 6083 CNRS, Universite Francois Rabelais, Parc de Grandmont F Tours, France address lesigne@univ-tours.fr Volny Laboratoire d'analyse et Modeles Stochastiques, UPRES-A 6085 CNRS, Universite de Rouen, F-7682 Mont-Saint-Aignan Cedex, France address volny@univ-rouen.fr 22

Large deviations for martingales

Large deviations for martingales Stochastic Processes and their Applications 96 2001) 143 159 www.elsevier.com/locate/spa Large deviations for martingales Emmanuel Lesigne a, Dalibor Volny b; a Laboratoire de Mathematiques et Physique

More information

Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon Measures

Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon Measures 2 1 Borel Regular Measures We now state and prove an important regularity property of Borel regular outer measures: Stanford Mathematics Department Math 205A Lecture Supplement #4 Borel Regular & Radon

More information

BERNOULLI DYNAMICAL SYSTEMS AND LIMIT THEOREMS. Dalibor Volný Université de Rouen

BERNOULLI DYNAMICAL SYSTEMS AND LIMIT THEOREMS. Dalibor Volný Université de Rouen BERNOULLI DYNAMICAL SYSTEMS AND LIMIT THEOREMS Dalibor Volný Université de Rouen Dynamical system (Ω,A,µ,T) : (Ω,A,µ) is a probablity space, T : Ω Ω a bijective, bimeasurable, and measure preserving mapping.

More information

An Indicator Function Limit Theorem in Dynamical Systems

An Indicator Function Limit Theorem in Dynamical Systems arxiv:080.2452v2 [math.ds] 5 Nov 2009 An Indicator Function Limit Theorem in Dynamical Systems Olivier Durieu & Dalibor Volný October 27, 208 Abstract We show by a constructive proof that in all aperiodic

More information

A CLT FOR MULTI-DIMENSIONAL MARTINGALE DIFFERENCES IN A LEXICOGRAPHIC ORDER GUY COHEN. Dedicated to the memory of Mikhail Gordin

A CLT FOR MULTI-DIMENSIONAL MARTINGALE DIFFERENCES IN A LEXICOGRAPHIC ORDER GUY COHEN. Dedicated to the memory of Mikhail Gordin A CLT FOR MULTI-DIMENSIONAL MARTINGALE DIFFERENCES IN A LEXICOGRAPHIC ORDER GUY COHEN Dedicated to the memory of Mikhail Gordin Abstract. We prove a central limit theorem for a square-integrable ergodic

More information

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria ESI The Erwin Schrodinger International Boltzmanngasse 9 Institute for Mathematical Physics A-19 Wien, Austria The Negative Discrete Spectrum of a Class of Two{Dimentional Schrodinger Operators with Magnetic

More information

The Pacic Institute for the Mathematical Sciences http://www.pims.math.ca pims@pims.math.ca Surprise Maximization D. Borwein Department of Mathematics University of Western Ontario London, Ontario, Canada

More information

2 Section 2 However, in order to apply the above idea, we will need to allow non standard intervals ('; ) in the proof. More precisely, ' and may gene

2 Section 2 However, in order to apply the above idea, we will need to allow non standard intervals ('; ) in the proof. More precisely, ' and may gene Introduction 1 A dierential intermediate value theorem by Joris van der Hoeven D pt. de Math matiques (B t. 425) Universit Paris-Sud 91405 Orsay Cedex France June 2000 Abstract Let T be the eld of grid-based

More information

3 Integration and Expectation

3 Integration and Expectation 3 Integration and Expectation 3.1 Construction of the Lebesgue Integral Let (, F, µ) be a measure space (not necessarily a probability space). Our objective will be to define the Lebesgue integral R fdµ

More information

Exercise Solutions to Functional Analysis

Exercise Solutions to Functional Analysis Exercise Solutions to Functional Analysis Note: References refer to M. Schechter, Principles of Functional Analysis Exersize that. Let φ,..., φ n be an orthonormal set in a Hilbert space H. Show n f n

More information

B. Appendix B. Topological vector spaces

B. Appendix B. Topological vector spaces B.1 B. Appendix B. Topological vector spaces B.1. Fréchet spaces. In this appendix we go through the definition of Fréchet spaces and their inductive limits, such as they are used for definitions of function

More information

Mathematical Institute, University of Utrecht. The problem of estimating the mean of an observed Gaussian innite-dimensional vector

Mathematical Institute, University of Utrecht. The problem of estimating the mean of an observed Gaussian innite-dimensional vector On Minimax Filtering over Ellipsoids Eduard N. Belitser and Boris Y. Levit Mathematical Institute, University of Utrecht Budapestlaan 6, 3584 CD Utrecht, The Netherlands The problem of estimating the mean

More information

Asymptotic Properties of Kaplan-Meier Estimator. for Censored Dependent Data. Zongwu Cai. Department of Mathematics

Asymptotic Properties of Kaplan-Meier Estimator. for Censored Dependent Data. Zongwu Cai. Department of Mathematics To appear in Statist. Probab. Letters, 997 Asymptotic Properties of Kaplan-Meier Estimator for Censored Dependent Data by Zongwu Cai Department of Mathematics Southwest Missouri State University Springeld,

More information

MARIA GIRARDI Fact 1.1. For a bounded linear operator T from L 1 into X, the following statements are equivalent. (1) T is Dunford-Pettis. () T maps w

MARIA GIRARDI Fact 1.1. For a bounded linear operator T from L 1 into X, the following statements are equivalent. (1) T is Dunford-Pettis. () T maps w DENTABILITY, TREES, AND DUNFORD-PETTIS OPERATORS ON L 1 Maria Girardi University of Illinois at Urbana-Champaign Pacic J. Math. 148 (1991) 59{79 Abstract. If all bounded linear operators from L1 into a

More information

1 Introduction This work follows a paper by P. Shields [1] concerned with a problem of a relation between the entropy rate of a nite-valued stationary

1 Introduction This work follows a paper by P. Shields [1] concerned with a problem of a relation between the entropy rate of a nite-valued stationary Prexes and the Entropy Rate for Long-Range Sources Ioannis Kontoyiannis Information Systems Laboratory, Electrical Engineering, Stanford University. Yurii M. Suhov Statistical Laboratory, Pure Math. &

More information

ON THE ERDOS-STONE THEOREM

ON THE ERDOS-STONE THEOREM ON THE ERDOS-STONE THEOREM V. CHVATAL AND E. SZEMEREDI In 1946, Erdos and Stone [3] proved that every graph with n vertices and at least edges contains a large K d+l (t), a complete (d + l)-partite graph

More information

Some functional (Hölderian) limit theorems and their applications (II)

Some functional (Hölderian) limit theorems and their applications (II) Some functional (Hölderian) limit theorems and their applications (II) Alfredas Račkauskas Vilnius University Outils Statistiques et Probabilistes pour la Finance Université de Rouen June 1 5, Rouen (Rouen

More information

Markov processes Course note 2. Martingale problems, recurrence properties of discrete time chains.

Markov processes Course note 2. Martingale problems, recurrence properties of discrete time chains. Institute for Applied Mathematics WS17/18 Massimiliano Gubinelli Markov processes Course note 2. Martingale problems, recurrence properties of discrete time chains. [version 1, 2017.11.1] We introduce

More information

Constrained Leja points and the numerical solution of the constrained energy problem

Constrained Leja points and the numerical solution of the constrained energy problem Journal of Computational and Applied Mathematics 131 (2001) 427 444 www.elsevier.nl/locate/cam Constrained Leja points and the numerical solution of the constrained energy problem Dan I. Coroian, Peter

More information

Entropy dimensions and a class of constructive examples

Entropy dimensions and a class of constructive examples Entropy dimensions and a class of constructive examples Sébastien Ferenczi Institut de Mathématiques de Luminy CNRS - UMR 6206 Case 907, 63 av. de Luminy F3288 Marseille Cedex 9 (France) and Fédération

More information

arxiv: v1 [math.pr] 9 Apr 2015 Dalibor Volný Laboratoire de Mathématiques Raphaël Salem, UMR 6085, Université de Rouen, France

arxiv: v1 [math.pr] 9 Apr 2015 Dalibor Volný Laboratoire de Mathématiques Raphaël Salem, UMR 6085, Université de Rouen, France A CENTRAL LIMIT THEOREM FOR FIELDS OF MARTINGALE DIFFERENCES arxiv:1504.02439v1 [math.pr] 9 Apr 2015 Dalibor Volný Laboratoire de Mathématiques Raphaël Salem, UMR 6085, Université de Rouen, France Abstract.

More information

2 Makoto Maejima (ii) Non-Gaussian -stable, 0 << 2, if it is not a delta measure, b() does not varnish for any a>0, b() a = b(a 1= )e ic for some c 2

2 Makoto Maejima (ii) Non-Gaussian -stable, 0 << 2, if it is not a delta measure, b() does not varnish for any a>0, b() a = b(a 1= )e ic for some c 2 This is page 1 Printer: Opaque this Limit Theorems for Innite Variance Sequences Makoto Maejima ABSTRACT This article discusses limit theorems for some stationary sequences not having nite variances. 1

More information

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria ESI The Erwin Schrodinger International Boltzmanngasse 9 Institute for Mathematical Physics A-1090 Wien, Austria Common Homoclinic Points of Commuting Toral Automorphisms Anthony Manning Klaus Schmidt

More information

Introduction and Preliminaries

Introduction and Preliminaries Chapter 1 Introduction and Preliminaries This chapter serves two purposes. The first purpose is to prepare the readers for the more systematic development in later chapters of methods of real analysis

More information

Central Limit Theorem for Non-stationary Markov Chains

Central Limit Theorem for Non-stationary Markov Chains Central Limit Theorem for Non-stationary Markov Chains Magda Peligrad University of Cincinnati April 2011 (Institute) April 2011 1 / 30 Plan of talk Markov processes with Nonhomogeneous transition probabilities

More information

Comparison between criteria leading to the weak invariance principle

Comparison between criteria leading to the weak invariance principle Annales de l Institut Henri Poincaré - Probabilités et Statistiques 2008, Vol. 44, No. 2, 324 340 DOI: 10.1214/07-AIHP123 Association des Publications de l Institut Henri Poincaré, 2008 www.imstat.org/aihp

More information

Ole Christensen 3. October 20, Abstract. We point out some connections between the existing theories for

Ole Christensen 3. October 20, Abstract. We point out some connections between the existing theories for Frames and pseudo-inverses. Ole Christensen 3 October 20, 1994 Abstract We point out some connections between the existing theories for frames and pseudo-inverses. In particular, using the pseudo-inverse

More information

The Erwin Schrodinger International Pasteurgasse 6/7. Institute for Mathematical Physics A-1090 Wien, Austria

The Erwin Schrodinger International Pasteurgasse 6/7. Institute for Mathematical Physics A-1090 Wien, Austria ESI The Erwin Schrodinger International Pasteurgasse 6/7 Institute for Mathematical Physics A-1090 Wien, Austria On the Point Spectrum of Dirence Schrodinger Operators Vladimir Buslaev Alexander Fedotov

More information

arxiv:math/ v1 [math.fa] 26 Oct 1993

arxiv:math/ v1 [math.fa] 26 Oct 1993 arxiv:math/9310217v1 [math.fa] 26 Oct 1993 ON COMPLEMENTED SUBSPACES OF SUMS AND PRODUCTS OF BANACH SPACES M.I.Ostrovskii Abstract. It is proved that there exist complemented subspaces of countable topological

More information

Limit Theorems for Exchangeable Random Variables via Martingales

Limit Theorems for Exchangeable Random Variables via Martingales Limit Theorems for Exchangeable Random Variables via Martingales Neville Weber, University of Sydney. May 15, 2006 Probabilistic Symmetries and Their Applications A sequence of random variables {X 1, X

More information

LARGE DEVIATION PROBABILITIES FOR SUMS OF HEAVY-TAILED DEPENDENT RANDOM VECTORS*

LARGE DEVIATION PROBABILITIES FOR SUMS OF HEAVY-TAILED DEPENDENT RANDOM VECTORS* LARGE EVIATION PROBABILITIES FOR SUMS OF HEAVY-TAILE EPENENT RANOM VECTORS* Adam Jakubowski Alexander V. Nagaev Alexander Zaigraev Nicholas Copernicus University Faculty of Mathematics and Computer Science

More information

Balance properties of multi-dimensional words

Balance properties of multi-dimensional words Theoretical Computer Science 273 (2002) 197 224 www.elsevier.com/locate/tcs Balance properties of multi-dimensional words Valerie Berthe a;, Robert Tijdeman b a Institut de Mathematiques de Luminy, CNRS-UPR

More information

Additive functionals of infinite-variance moving averages. Wei Biao Wu The University of Chicago TECHNICAL REPORT NO. 535

Additive functionals of infinite-variance moving averages. Wei Biao Wu The University of Chicago TECHNICAL REPORT NO. 535 Additive functionals of infinite-variance moving averages Wei Biao Wu The University of Chicago TECHNICAL REPORT NO. 535 Departments of Statistics The University of Chicago Chicago, Illinois 60637 June

More information

Max-Planck-Institut fur Mathematik in den Naturwissenschaften Leipzig Uniformly distributed measures in Euclidean spaces by Bernd Kirchheim and David Preiss Preprint-Nr.: 37 1998 Uniformly Distributed

More information

JOININGS, FACTORS, AND BAIRE CATEGORY

JOININGS, FACTORS, AND BAIRE CATEGORY JOININGS, FACTORS, AND BAIRE CATEGORY Abstract. We discuss the Burton-Rothstein approach to Ornstein theory. 1. Weak convergence Let (X, B) be a metric space and B be the Borel sigma-algebra generated

More information

16 1 Basic Facts from Functional Analysis and Banach Lattices

16 1 Basic Facts from Functional Analysis and Banach Lattices 16 1 Basic Facts from Functional Analysis and Banach Lattices 1.2.3 Banach Steinhaus Theorem Another fundamental theorem of functional analysis is the Banach Steinhaus theorem, or the Uniform Boundedness

More information

Gärtner-Ellis Theorem and applications.

Gärtner-Ellis Theorem and applications. Gärtner-Ellis Theorem and applications. Elena Kosygina July 25, 208 In this lecture we turn to the non-i.i.d. case and discuss Gärtner-Ellis theorem. As an application, we study Curie-Weiss model with

More information

On the Set of Limit Points of Normed Sums of Geometrically Weighted I.I.D. Bounded Random Variables

On the Set of Limit Points of Normed Sums of Geometrically Weighted I.I.D. Bounded Random Variables On the Set of Limit Points of Normed Sums of Geometrically Weighted I.I.D. Bounded Random Variables Deli Li 1, Yongcheng Qi, and Andrew Rosalsky 3 1 Department of Mathematical Sciences, Lakehead University,

More information

2 BAISHENG YAN When L =,it is easily seen that the set K = coincides with the set of conformal matrices, that is, K = = fr j 0 R 2 SO(n)g: Weakly L-qu

2 BAISHENG YAN When L =,it is easily seen that the set K = coincides with the set of conformal matrices, that is, K = = fr j 0 R 2 SO(n)g: Weakly L-qu RELAXATION AND ATTAINMENT RESULTS FOR AN INTEGRAL FUNCTIONAL WITH UNBOUNDED ENERGY-WELL BAISHENG YAN Abstract. Consider functional I(u) = R jjdujn ; L det Duj dx whose energy-well consists of matrices

More information

Boundedly complete weak-cauchy basic sequences in Banach spaces with the PCP

Boundedly complete weak-cauchy basic sequences in Banach spaces with the PCP Journal of Functional Analysis 253 (2007) 772 781 www.elsevier.com/locate/jfa Note Boundedly complete weak-cauchy basic sequences in Banach spaces with the PCP Haskell Rosenthal Department of Mathematics,

More information

The Degree of the Splitting Field of a Random Polynomial over a Finite Field

The Degree of the Splitting Field of a Random Polynomial over a Finite Field The Degree of the Splitting Field of a Random Polynomial over a Finite Field John D. Dixon and Daniel Panario School of Mathematics and Statistics Carleton University, Ottawa, Canada fjdixon,danielg@math.carleton.ca

More information

LOGARITHMIC MULTIFRACTAL SPECTRUM OF STABLE. Department of Mathematics, National Taiwan University. Taipei, TAIWAN. and. S.

LOGARITHMIC MULTIFRACTAL SPECTRUM OF STABLE. Department of Mathematics, National Taiwan University. Taipei, TAIWAN. and. S. LOGARITHMIC MULTIFRACTAL SPECTRUM OF STABLE OCCUPATION MEASURE Narn{Rueih SHIEH Department of Mathematics, National Taiwan University Taipei, TAIWAN and S. James TAYLOR 2 School of Mathematics, University

More information

A regeneration proof of the central limit theorem for uniformly ergodic Markov chains

A regeneration proof of the central limit theorem for uniformly ergodic Markov chains A regeneration proof of the central limit theorem for uniformly ergodic Markov chains By AJAY JASRA Department of Mathematics, Imperial College London, SW7 2AZ, London, UK and CHAO YANG Department of Mathematics,

More information

Garrett: `Bernstein's analytic continuation of complex powers' 2 Let f be a polynomial in x 1 ; : : : ; x n with real coecients. For complex s, let f

Garrett: `Bernstein's analytic continuation of complex powers' 2 Let f be a polynomial in x 1 ; : : : ; x n with real coecients. For complex s, let f 1 Bernstein's analytic continuation of complex powers c1995, Paul Garrett, garrettmath.umn.edu version January 27, 1998 Analytic continuation of distributions Statement of the theorems on analytic continuation

More information

MEASURE AND CATEGORY { Department of Mathematics, University of Tennessee, Chattanooga,

MEASURE AND CATEGORY { Department of Mathematics, University of Tennessee, Chattanooga, Mathematica Pannonica 7/1 (1996), 69 { 78 MEASURE AND CATEGORY { SOME NON-ANALOGUES Harry I. Miller Department of Mathematics, University of Tennessee, Chattanooga, TN, 371403, USA Franz J. Schnitzer Institut

More information

Fuzzy Statistical Limits

Fuzzy Statistical Limits Fuzzy Statistical Limits Mark Burgin a and Oktay Duman b a Department of Mathematics, University of California, Los Angeles, California 90095-1555, USA b TOBB Economics and Technology University, Faculty

More information

SOME MEASURABILITY AND CONTINUITY PROPERTIES OF ARBITRARY REAL FUNCTIONS

SOME MEASURABILITY AND CONTINUITY PROPERTIES OF ARBITRARY REAL FUNCTIONS LE MATEMATICHE Vol. LVII (2002) Fasc. I, pp. 6382 SOME MEASURABILITY AND CONTINUITY PROPERTIES OF ARBITRARY REAL FUNCTIONS VITTORINO PATA - ALFONSO VILLANI Given an arbitrary real function f, the set D

More information

Supplementary Notes for W. Rudin: Principles of Mathematical Analysis

Supplementary Notes for W. Rudin: Principles of Mathematical Analysis Supplementary Notes for W. Rudin: Principles of Mathematical Analysis SIGURDUR HELGASON In 8.00B it is customary to cover Chapters 7 in Rudin s book. Experience shows that this requires careful planning

More information

A Martingale Central Limit Theorem

A Martingale Central Limit Theorem A Martingale Central Limit Theorem Sunder Sethuraman We present a proof of a martingale central limit theorem (Theorem 2) due to McLeish (974). Then, an application to Markov chains is given. Lemma. For

More information

An almost sure invariance principle for additive functionals of Markov chains

An almost sure invariance principle for additive functionals of Markov chains Statistics and Probability Letters 78 2008 854 860 www.elsevier.com/locate/stapro An almost sure invariance principle for additive functionals of Markov chains F. Rassoul-Agha a, T. Seppäläinen b, a Department

More information

We suppose that for each "small market" there exists a probability measure Q n on F n that is equivalent to the original measure P n, suchthats n is a

We suppose that for each small market there exists a probability measure Q n on F n that is equivalent to the original measure P n, suchthats n is a Asymptotic Arbitrage in Non-Complete Large Financial Markets Irene Klein Walter Schachermayer Institut fur Statistik, Universitat Wien Abstract. Kabanov and Kramkov introduced the notion of "large nancial

More information

4th Preparation Sheet - Solutions

4th Preparation Sheet - Solutions Prof. Dr. Rainer Dahlhaus Probability Theory Summer term 017 4th Preparation Sheet - Solutions Remark: Throughout the exercise sheet we use the two equivalent definitions of separability of a metric space

More information

Nets Hawk Katz Theorem. There existsaconstant C>so that for any number >, whenever E [ ] [ ] is a set which does not contain the vertices of any axis

Nets Hawk Katz Theorem. There existsaconstant C>so that for any number >, whenever E [ ] [ ] is a set which does not contain the vertices of any axis New York Journal of Mathematics New York J. Math. 5 999) {3. On the Self Crossing Six Sided Figure Problem Nets Hawk Katz Abstract. It was shown by Carbery, Christ, and Wright that any measurable set E

More information

REPRESENTABLE BANACH SPACES AND UNIFORMLY GÂTEAUX-SMOOTH NORMS. Julien Frontisi

REPRESENTABLE BANACH SPACES AND UNIFORMLY GÂTEAUX-SMOOTH NORMS. Julien Frontisi Serdica Math. J. 22 (1996), 33-38 REPRESENTABLE BANACH SPACES AND UNIFORMLY GÂTEAUX-SMOOTH NORMS Julien Frontisi Communicated by G. Godefroy Abstract. It is proved that a representable non-separable Banach

More information

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f

290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f Numer. Math. 67: 289{301 (1994) Numerische Mathematik c Springer-Verlag 1994 Electronic Edition Least supported bases and local linear independence J.M. Carnicer, J.M. Pe~na? Departamento de Matematica

More information

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi Real Analysis Math 3AH Rudin, Chapter # Dominique Abdi.. If r is rational (r 0) and x is irrational, prove that r + x and rx are irrational. Solution. Assume the contrary, that r+x and rx are rational.

More information

ON COMPLEMENTED SUBSPACES OF SUMS AND PRODUCTS OF BANACH SPACES

ON COMPLEMENTED SUBSPACES OF SUMS AND PRODUCTS OF BANACH SPACES PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 124, Number 7, July 1996 ON COMPLEMENTED SUBSPACES OF SUMS AND PRODUCTS OF BANACH SPACES M. I. OSTROVSKII (Communicated by Dale Alspach) Abstract.

More information

Asymptotics of generating the symmetric and alternating groups

Asymptotics of generating the symmetric and alternating groups Asymptotics of generating the symmetric and alternating groups John D. Dixon School of Mathematics and Statistics Carleton University, Ottawa, Ontario K2G 0E2 Canada jdixon@math.carleton.ca October 20,

More information

THEOREMS, ETC., FOR MATH 515

THEOREMS, ETC., FOR MATH 515 THEOREMS, ETC., FOR MATH 515 Proposition 1 (=comment on page 17). If A is an algebra, then any finite union or finite intersection of sets in A is also in A. Proposition 2 (=Proposition 1.1). For every

More information

Tsung-Lin Cheng and Yuan-Shih Chow. Taipei115,Taiwan R.O.C.

Tsung-Lin Cheng and Yuan-Shih Chow. Taipei115,Taiwan R.O.C. A Generalization and Alication of McLeish's Central Limit Theorem by Tsung-Lin Cheng and Yuan-Shih Chow Institute of Statistical Science Academia Sinica Taiei5,Taiwan R.O.C. hcho3@stat.sinica.edu.tw Abstract.

More information

Semi-strongly asymptotically non-expansive mappings and their applications on xed point theory

Semi-strongly asymptotically non-expansive mappings and their applications on xed point theory Hacettepe Journal of Mathematics and Statistics Volume 46 (4) (2017), 613 620 Semi-strongly asymptotically non-expansive mappings and their applications on xed point theory Chris Lennard and Veysel Nezir

More information

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product Chapter 4 Hilbert Spaces 4.1 Inner Product Spaces Inner Product Space. A complex vector space E is called an inner product space (or a pre-hilbert space, or a unitary space) if there is a mapping (, )

More information

ON DENSITY TOPOLOGIES WITH RESPECT

ON DENSITY TOPOLOGIES WITH RESPECT Journal of Applied Analysis Vol. 8, No. 2 (2002), pp. 201 219 ON DENSITY TOPOLOGIES WITH RESPECT TO INVARIANT σ-ideals J. HEJDUK Received June 13, 2001 and, in revised form, December 17, 2001 Abstract.

More information

CONVERGENCE OF POLYNOMIAL ERGODIC AVERAGES

CONVERGENCE OF POLYNOMIAL ERGODIC AVERAGES CONVERGENCE OF POLYNOMIAL ERGODIC AVERAGES BERNARD HOST AND BRYNA KRA Abstract. We prove the L 2 convergence for an ergodic average of a product of functions evaluated along polynomial times in a totally

More information

Stochastic Processes

Stochastic Processes Stochastic Processes A very simple introduction Péter Medvegyev 2009, January Medvegyev (CEU) Stochastic Processes 2009, January 1 / 54 Summary from measure theory De nition (X, A) is a measurable space

More information

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria

The Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A-1090 Wien, Austria ESI The Erwin Schrodinger International Boltzmanngasse 9 Institute for Mathematical Physics A-1090 Wien, Austria Noncommutative Contact Algebras Hideki Omori Yoshiaki Maeda Naoya Miyazaki Akira Yoshioka

More information

STAT 7032 Probability Spring Wlodek Bryc

STAT 7032 Probability Spring Wlodek Bryc STAT 7032 Probability Spring 2018 Wlodek Bryc Created: Friday, Jan 2, 2014 Revised for Spring 2018 Printed: January 9, 2018 File: Grad-Prob-2018.TEX Department of Mathematical Sciences, University of Cincinnati,

More information

On some properties of the dierence. spectrum. D. L. Salinger and J. D. Stegeman

On some properties of the dierence. spectrum. D. L. Salinger and J. D. Stegeman On some properties of the dierence spectrum D. L. Salinger and J. D. Stegeman Abstract. We consider problems arising from trying to lift the non-synthesis spectrum of a closed set in a quotient of a locally

More information

DILWORTH and GIRARDI 2 2. XAMPLS: L 1 (X) vs. P 1 (X) Let X be a Banach space with dual X and let (; ;) be the usual Lebesgue measure space on [0; 1].

DILWORTH and GIRARDI 2 2. XAMPLS: L 1 (X) vs. P 1 (X) Let X be a Banach space with dual X and let (; ;) be the usual Lebesgue measure space on [0; 1]. Contemporary Mathematics Volume 00, 0000 Bochner vs. Pettis norm: examples and results S.J. DILWORTH AND MARIA GIRARDI Contemp. Math. 144 (1993) 69{80 Banach Spaces, Bor-Luh Lin and William B. Johnson,

More information

COMPACTNESS IN L 1, DUNFORD-PETTIS OPERATORS, GEOMETRY OF BANACH SPACES Maria Girardi University of Illinois at Urbana-Champaign Proc. Amer. Math. Soc

COMPACTNESS IN L 1, DUNFORD-PETTIS OPERATORS, GEOMETRY OF BANACH SPACES Maria Girardi University of Illinois at Urbana-Champaign Proc. Amer. Math. Soc COMPCTNESS IN L 1, DUNFOD-PETTIS OPETOS, GEOMETY OF NCH SPCES Maria Girardi University of Illinois at Urbana-Champaign Proc. mer. Math. Soc. 111 (1991) 767{777 bstract. type of oscillation modeled on MO

More information

2 W. LAWTON, S. L. LEE AND ZUOWEI SHEN is called the fundamental condition, and a sequence which satises the fundamental condition will be called a fu

2 W. LAWTON, S. L. LEE AND ZUOWEI SHEN is called the fundamental condition, and a sequence which satises the fundamental condition will be called a fu CONVERGENCE OF MULTIDIMENSIONAL CASCADE ALGORITHM W. LAWTON, S. L. LEE AND ZUOWEI SHEN Abstract. Necessary and sucient conditions on the spectrum of the restricted transition operators are given for the

More information

P i [B k ] = lim. n=1 p(n) ii <. n=1. V i :=

P i [B k ] = lim. n=1 p(n) ii <. n=1. V i := 2.7. Recurrence and transience Consider a Markov chain {X n : n N 0 } on state space E with transition matrix P. Definition 2.7.1. A state i E is called recurrent if P i [X n = i for infinitely many n]

More information

A dyadic endomorphism which is Bernoulli but not standard

A dyadic endomorphism which is Bernoulli but not standard A dyadic endomorphism which is Bernoulli but not standard Christopher Hoffman Daniel Rudolph November 4, 2005 Abstract Any measure preserving endomorphism generates both a decreasing sequence of σ-algebras

More information

1. Introduction Let fx(t) t 1g be a real-valued mean-zero Gaussian process, and let \kk" be a semi-norm in the space of real functions on [ 1]. In the

1. Introduction Let fx(t) t 1g be a real-valued mean-zero Gaussian process, and let \kk be a semi-norm in the space of real functions on [ 1]. In the Small ball estimates for Brownian motion under a weighted -norm by hilippe BERTHET (1) and Zhan SHI () Universite de Rennes I & Universite aris VI Summary. Let fw (t) t 1g be a real-valued Wiener process,

More information

THE ALTERNATIVE DUNFORD-PETTIS PROPERTY FOR SUBSPACES OF THE COMPACT OPERATORS

THE ALTERNATIVE DUNFORD-PETTIS PROPERTY FOR SUBSPACES OF THE COMPACT OPERATORS THE ALTERNATIVE DUNFORD-PETTIS PROPERTY FOR SUBSPACES OF THE COMPACT OPERATORS MARÍA D. ACOSTA AND ANTONIO M. PERALTA Abstract. A Banach space X has the alternative Dunford-Pettis property if for every

More information

Krzysztof Burdzy University of Washington. = X(Y (t)), t 0}

Krzysztof Burdzy University of Washington. = X(Y (t)), t 0} VARIATION OF ITERATED BROWNIAN MOTION Krzysztof Burdzy University of Washington 1. Introduction and main results. Suppose that X 1, X 2 and Y are independent standard Brownian motions starting from 0 and

More information

THE CYCLIC DOUGLAS RACHFORD METHOD FOR INCONSISTENT FEASIBILITY PROBLEMS

THE CYCLIC DOUGLAS RACHFORD METHOD FOR INCONSISTENT FEASIBILITY PROBLEMS THE CYCLIC DOUGLAS RACHFORD METHOD FOR INCONSISTENT FEASIBILITY PROBLEMS JONATHAN M. BORWEIN AND MATTHEW K. TAM Abstract. We analyse the behaviour of the newly introduced cyclic Douglas Rachford algorithm

More information

A NOTE ON FUNCTION SPACES GENERATED BY RADEMACHER SERIES

A NOTE ON FUNCTION SPACES GENERATED BY RADEMACHER SERIES Proceedings of the Edinburgh Mathematical Society (1997) 40, 119-126 A NOTE ON FUNCTION SPACES GENERATED BY RADEMACHER SERIES by GUILLERMO P. CURBERA* (Received 29th March 1995) Let X be a rearrangement

More information

25.1 Ergodicity and Metric Transitivity

25.1 Ergodicity and Metric Transitivity Chapter 25 Ergodicity This lecture explains what it means for a process to be ergodic or metrically transitive, gives a few characterizes of these properties (especially for AMS processes), and deduces

More information

AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES

AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES Lithuanian Mathematical Journal, Vol. 4, No. 3, 00 AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES V. Bentkus Vilnius Institute of Mathematics and Informatics, Akademijos 4,

More information

Bounds on expectation of order statistics from a nite population

Bounds on expectation of order statistics from a nite population Journal of Statistical Planning and Inference 113 (2003) 569 588 www.elsevier.com/locate/jspi Bounds on expectation of order statistics from a nite population N. Balakrishnan a;, C. Charalambides b, N.

More information

HAIYUN ZHOU, RAVI P. AGARWAL, YEOL JE CHO, AND YONG SOO KIM

HAIYUN ZHOU, RAVI P. AGARWAL, YEOL JE CHO, AND YONG SOO KIM Georgian Mathematical Journal Volume 9 (2002), Number 3, 591 600 NONEXPANSIVE MAPPINGS AND ITERATIVE METHODS IN UNIFORMLY CONVEX BANACH SPACES HAIYUN ZHOU, RAVI P. AGARWAL, YEOL JE CHO, AND YONG SOO KIM

More information

Entropy and Ergodic Theory Lecture 15: A first look at concentration

Entropy and Ergodic Theory Lecture 15: A first look at concentration Entropy and Ergodic Theory Lecture 15: A first look at concentration 1 Introduction to concentration Let X 1, X 2,... be i.i.d. R-valued RVs with common distribution µ, and suppose for simplicity that

More information

Mi-Hwa Ko. t=1 Z t is true. j=0

Mi-Hwa Ko. t=1 Z t is true. j=0 Commun. Korean Math. Soc. 21 (2006), No. 4, pp. 779 786 FUNCTIONAL CENTRAL LIMIT THEOREMS FOR MULTIVARIATE LINEAR PROCESSES GENERATED BY DEPENDENT RANDOM VECTORS Mi-Hwa Ko Abstract. Let X t be an m-dimensional

More information

A NOTE ON MATRIX REFINEMENT EQUATIONS. Abstract. Renement equations involving matrix masks are receiving a lot of attention these days.

A NOTE ON MATRIX REFINEMENT EQUATIONS. Abstract. Renement equations involving matrix masks are receiving a lot of attention these days. A NOTE ON MATRI REFINEMENT EQUATIONS THOMAS A. HOGAN y Abstract. Renement equations involving matrix masks are receiving a lot of attention these days. They can play a central role in the study of renable

More information

Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek

Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek J. Korean Math. Soc. 41 (2004), No. 5, pp. 883 894 CONVERGENCE OF WEIGHTED SUMS FOR DEPENDENT RANDOM VARIABLES Han-Ying Liang, Dong-Xia Zhang, and Jong-Il Baek Abstract. We discuss in this paper the strong

More information

MORE ON CONTINUOUS FUNCTIONS AND SETS

MORE ON CONTINUOUS FUNCTIONS AND SETS Chapter 6 MORE ON CONTINUOUS FUNCTIONS AND SETS This chapter can be considered enrichment material containing also several more advanced topics and may be skipped in its entirety. You can proceed directly

More information

Baire measures on uncountable product spaces 1. Abstract. We show that assuming the continuum hypothesis there exists a

Baire measures on uncountable product spaces 1. Abstract. We show that assuming the continuum hypothesis there exists a Baire measures on uncountable product spaces 1 by Arnold W. Miller 2 Abstract We show that assuming the continuum hypothesis there exists a nontrivial countably additive measure on the Baire subsets of

More information

ALMOST EVERYWHERE CONVERGENCE OF FEJÉR MEANS OF SOME SUBSEQUENCES OF FOURIER SERIES FOR INTEGRABLE FUNCTIONS WITH RESPECT TO THE KACZMARZ SYSTEM

ALMOST EVERYWHERE CONVERGENCE OF FEJÉR MEANS OF SOME SUBSEQUENCES OF FOURIER SERIES FOR INTEGRABLE FUNCTIONS WITH RESPECT TO THE KACZMARZ SYSTEM ADV MATH SCI JOURAL Advances in Mathematics: Scientic Journal 4 (205), no., 6577 ISS 857-8365 UDC: 57.58.5 ALMOST EVERWHERE COVERGECE OF FEJÉR MEAS OF SOME SUBSEQUECES OF FOURIER SERIES FOR ITEGRABLE FUCTIOS

More information

LEBESGUE INTEGRATION. Introduction

LEBESGUE INTEGRATION. Introduction LEBESGUE INTEGATION EYE SJAMAA Supplementary notes Math 414, Spring 25 Introduction The following heuristic argument is at the basis of the denition of the Lebesgue integral. This argument will be imprecise,

More information

Metric Spaces and Topology

Metric Spaces and Topology Chapter 2 Metric Spaces and Topology From an engineering perspective, the most important way to construct a topology on a set is to define the topology in terms of a metric on the set. This approach underlies

More information

THE LARGEST INTERSECTION LATTICE OF A CHRISTOS A. ATHANASIADIS. Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin.

THE LARGEST INTERSECTION LATTICE OF A CHRISTOS A. ATHANASIADIS. Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin. THE LARGEST INTERSECTION LATTICE OF A DISCRIMINANTAL ARRANGEMENT CHRISTOS A. ATHANASIADIS Abstract. We prove a conjecture of Bayer and Brandt [J. Alg. Combin. 6 (1997), 229{246] about the \largest" intersection

More information

Problem set 1, Real Analysis I, Spring, 2015.

Problem set 1, Real Analysis I, Spring, 2015. Problem set 1, Real Analysis I, Spring, 015. (1) Let f n : D R be a sequence of functions with domain D R n. Recall that f n f uniformly if and only if for all ɛ > 0, there is an N = N(ɛ) so that if n

More information

The small ball property in Banach spaces (quantitative results)

The small ball property in Banach spaces (quantitative results) The small ball property in Banach spaces (quantitative results) Ehrhard Behrends Abstract A metric space (M, d) is said to have the small ball property (sbp) if for every ε 0 > 0 there exists a sequence

More information

ANSWER TO A QUESTION BY BURR AND ERDŐS ON RESTRICTED ADDITION, AND RELATED RESULTS Mathematics Subject Classification: 11B05, 11B13, 11P99

ANSWER TO A QUESTION BY BURR AND ERDŐS ON RESTRICTED ADDITION, AND RELATED RESULTS Mathematics Subject Classification: 11B05, 11B13, 11P99 ANSWER TO A QUESTION BY BURR AND ERDŐS ON RESTRICTED ADDITION, AND RELATED RESULTS N. HEGYVÁRI, F. HENNECART AND A. PLAGNE Abstract. We study the gaps in the sequence of sums of h pairwise distinct elements

More information

arxiv: v1 [math.ds] 24 Jul 2011

arxiv: v1 [math.ds] 24 Jul 2011 Simple Spectrum for Tensor Products of Mixing Map Powers V.V. Ryzhikov arxiv:1107.4745v1 [math.ds] 24 Jul 2011 1 Introduction June 28, 2018 In this note we consider measure-preserving transformations of

More information

PACIFIC JOURNAL OF MATHEMATICS Vol. 190, No. 2, 1999 ENTROPY OF CUNTZ'S CANONICAL ENDOMORPHISM Marie Choda Let fs i g n i=1 be generators of the Cuntz

PACIFIC JOURNAL OF MATHEMATICS Vol. 190, No. 2, 1999 ENTROPY OF CUNTZ'S CANONICAL ENDOMORPHISM Marie Choda Let fs i g n i=1 be generators of the Cuntz PACIFIC JOURNAL OF MATHEMATICS Vol. 190, No. 2, 1999 ENTROPY OF CUNTZ'S CANONICAL ENDOMORPHISM Marie Choda Let fs i g n i=1 be generators of the Cuntz algebra OP n and let n be the *-endomorphism of O

More information

Spurious Chaotic Solutions of Dierential. Equations. Sigitas Keras. September Department of Applied Mathematics and Theoretical Physics

Spurious Chaotic Solutions of Dierential. Equations. Sigitas Keras. September Department of Applied Mathematics and Theoretical Physics UNIVERSITY OF CAMBRIDGE Numerical Analysis Reports Spurious Chaotic Solutions of Dierential Equations Sigitas Keras DAMTP 994/NA6 September 994 Department of Applied Mathematics and Theoretical Physics

More information

arxiv: v1 [math.pr] 17 May 2009

arxiv: v1 [math.pr] 17 May 2009 A strong law of large nubers for artingale arrays Yves F. Atchadé arxiv:0905.2761v1 [ath.pr] 17 May 2009 March 2009 Abstract: We prove a artingale triangular array generalization of the Chow-Birnbau- Marshall

More information

Brownian Motion and Conditional Probability

Brownian Motion and Conditional Probability Math 561: Theory of Probability (Spring 2018) Week 10 Brownian Motion and Conditional Probability 10.1 Standard Brownian Motion (SBM) Brownian motion is a stochastic process with both practical and theoretical

More information