Hausdorff, Large Deviation and Legendre Multifractal Spectra of Lévy Multistable Processes
|
|
- Lee Eaton
- 5 years ago
- Views:
Transcription
1 Hausdorff, Large Deviation and Legendre Multifractal Spectra of Lévy Multistable Processes arxiv:4599v [mathpr] Dec 4 R Le Guével Université de Rennes - Haute Bretagne, Equipe de Statistique Irmar, UMR CNRS 665 Place du Recteur Henri Le Moal, CS 437, 3543 RENNES Cedex, France ronanleguevel@univ-rennesfr and J Lévy Véhel Regularity team, INRIA Saclay and MAS laboratory Ecole Centrale Paris, Grande Voie des Vignes, 99, Chatenay Malabry, France jacqueslevy-vehel@inriafr Abstract We compute the Hausdorff multifractal spectrum of two versions of multistable Lévy motions These processes extend classical Lévy motion by letting the stability exponent α evolve in time The spectra provide a decomposition of [, ] into an uncountable disjoint union of sets with Hausdorff dimension one We also compute the increments-based large deviations multifractal spectrum of the independent increments multistable Lévy motion This spectrum turns out to be concave and thus coincides with the Legendre multifractal spectrum, but it is different from the Hausdorff multifractal spectrum The independent increments multistable Lévy motion thus provides an example where the strong multifractal formalism does not hold Introduction and background Multifractal analysis gives a fairly complete description of the singularity structure of measures, functions or stochastic processes Various versions of multifractal analysis exist, which include the determinations of the so-called Hausdorff, large deviation, and Legendre multifractal spectra [] Multifractal analysis has been performed for various measures [, 7], functions [5], and stochastic processes [4, 5, 8, 9, 6] In the case of Lévy processes, substantially finer results have been obtained in [3] using -microlocal analysis This article deals with the multifractal analysis of extensions of Lévy stable motions known as multistable Lévy motions Generally speaking, multistable processes extend the well-known stable processes see, eg [3] by letting the stability index α evolve in time These processes have been introduced in [3] and have been studied for instance in [, 6, 4, 8, 9, ] They provide useful models in various applications where the data display jumps with varying intensity, such as financial records, EEG or natural terrains: indeed, multistability is one practical way to deal with increments- non-stationarities
2 observed in various real-world phenomena, since a multistable process X is tangent, at each time u, to a stable process Z u in the following sense [, ]: Xu+rt Xu lim = Z r r h u t for a suitable h the limit is taken either in finite dimensional distributions or, when X has a version with càdlàg paths, in distribution - one then speaks of strong localisability Without loss of generality, we shall consider our processes on [,] We will need the following ingredients: α : [,], is a C function Γ i i is a sequence of arrival times of a Poisson process with unit arrival time V i i is a sequence of iid random variables with uniform distribution on [,] γ i i is a sequence of iid random variables with distribution Pγ i = = Pγ i = = / The three sequencesγ i i,v i i, andγ i i are independent We denotec = inf αu, u [,] d = sup αu, u [,] We shall consider two versions of Lévy multistable processes: the first one has independent but non stationary increments It admits the following representation: Bt = + i= γ i C /αv i αv i Γ /αv i i Vi t, while the second one has correlated non stationary increments and reads: Dt = C /αt αt γ i Γ /αt i [,t] V i, 3 i= where C u = x u sinx dx Both processes are semi-martingales and are tangent, at each time t, to αt stable Lévy motion See [9] and the references therein for more details on these processes We shall also denote: Yt = i= γ i Γ /αt i [,t] V i Hausdorff multifractal spectra Let h Y t denote the pointwise Hölder exponent of Y at t The Hausdorff multifractal analysis of Y consists in measuring the Hausdorff dimension denoted dim H of the sets F h = {t [,] : h Y t = h} The Hausdorff multifractal spectrum is the function h f H h := dim H F h We will use the following notations: S = i {V i }, S = S N and R t = {r n n N S : r n t} If r n n R t, we put V φn = r n Finally, define the positive function δ for t / S, δt = inf liminf logφi V φn n R t i log V φi t
3 Main result The Hausdorff multifractal spectra of both B and D are described by the following theorem: Theorem With probability one, the common Hausdorff multifractal spectrum f H of B and D satisfies: for h < ; hd for h [, ]; d f H h = for h, ; 4 d c dim H {t [,] : αt = c} for h = c ; for h > c Theorem follows from a series of lemmas that are proven in the next section: Lemma Almost surely, t Yt is càdlàg Lemma 3 Almost surely, t [, ]\S, δt Lemma 4 Almost surely, t [,]\S, h Y t δt αt Lemma 5 Let g : [,] R be a càdlàg function, and f be the function defined on [,] by ft = t gxdx The pointwise Hölder exponent h f of f verifies: t,, h f t Lemma 6 Almost surely, t [,]\S, h Y t δt αt Lemma 7 Almost surely, h <,f H h = Lemma 8 Almost surely, f H = Lemma 9 Almost surely, h, d ],f Hh = hd Lemma Almost surely, h d, c,f Hh = Lemma Almost surely, f H c = dim H {t [,] : αt = c} Lemma Almost surely, h > c,f Hh = Proofs of the lemmas Proof of Lemma : Set Y N t = N γ i Γ /αt i [,t] V i Lemma 8 in [9] states that, amost surely, Y N i= converges to Yt uniformly on [, ] Fix ε > and choose N N such that, N N, sup Y N t Yt ε t [,] 3 5 3
4 st case : t S Let i N be such that t = V i, and N = maxi,n Then, for h R, Yt Yt+h = Yt Y N t+y N t Y N t+h+y N t+h Yt+h Since lim h +Y N t Y N t+h =, there exists h > such that h,h, Y N t Y N t+h ε 3 As a consequence, h,h, Yt Yt+h ε and thus lim +Yt Yt+h = lim h Y N t Y N t+h = lim h N i= h γ i Γ /αt i t+h,t] V i = γ i Γ /αv i, thus lim h Y N t Y N t+h γ i Γ /αv i i = Choose h < such that h h,, Thus, Yt Yt+h γ i Γ /αv i i ε lim h Yt Yt+h = γ i Γ /αv i i nd case : t / S Since lim h Y N t+h Y N t =, there exists h > such that h < h, Y N t Y N t+h ε 3 Using 5, one thus has, for h < h, Yt Yt+h ε, and thus lim h Yt+h Yt = Note We have shown precisely that Y is càdlàg with set of jump points exactly equal to S The jump at point V i is of size γ i Γ /αv i i i Proof of Lemma 3: For j, k =,, j, let E k,j denote the interval [ k E k,j = j, k + +j+ j j Let us show that liminf j j j+ j k= j+ +j+ j {V i E k,j } { t / S,δt } first, and then that i= j P {V i E k,j } = We denote a j = j k= i= j +j+ j Assume that there exists J N such that for all j J, and all k =,, j, we can fix ik,j [ j, j+ ] with V ik,j E k,j Let t / S For all j J, there exists kj [, j ] such that t E kj,j, because a j j 4
5 As a consequence, t V ikj,j j+ j ikj,j j Hence logikj,j log t V ikj,j This entails liminf logikj,j j j log t V ikj,j, and, since V ikj,j tends to t, δt Finally, distinguishing the cases k =, k =,, j and k = j, one estimates j P k= j+ {V i / E k,j } i= j j k= P {V i / E k,j } i= j j+ a j + j j a j j + P{V / E k,j } j k= a j + j +j a j j Since + j a j j < +, Borel-Cantelli lemma allows us to conclude j= Proof of Lemma 4: Recall Note Lemma of [5] entails that, for all sequences V φi R t, and all t / S, h Y t liminf i + log Γ αv φi φi log V φi t = liminf i + αv φi log Γ φi log V φi t Since α is continuous, φi tends to infinity, the sequences V φi i converges to t, and almost surely Γ i i i tends to when i tends to infinity, one obtains h Y t αt liminf i + log φi log V φi t This inequality holds for all sequences V φi R t, and thus, t / S, h Y t δt αt Proof of Lemma 5: Since g is càdlàg, h g is non negative for all t Integration increases pointwise regularity by at least one, and thus h f t for all t An alternative direct proof goes as follows: let t,, and h > One computes ft+h ft h gt + = h = 5 t t+h gx gt + dx gt+sh gt + ds
6 s,, lim h +gt+sh gt+ = Since g is càdlàg, it is bounded and thus: Likewise, for h <, ft+h ft lim = gt + h + h and This entails h f t ft+h ft h Proof of Lemma 6: Theorem 7 of [9] states that: where At = gt = gt+sh gt ds, ft+h ft lim = gt h h t + i= Dt = At+Bt, d C /α α Γ /α i γ i dt In addition, Lemma 8 of [9] entails that converges to + i= d γ i C /α α Γ /α i dt of Lemma shows that s + N d γ i i= s [,s V i ds s dt [,s V i C /α Γ /α α i N s [,s V i uniformly on [,] The same proof as the one d C /α Γ /α α i γ i dt i= s [,s V i is càdlàg Lemma 5 then entails that h A t Since c >, it is thus sufficient to show that h B t δt αt Write Bt = Wt+Zt where Wt = + i= γ i C /αv i αv i Γ /αv i i i /αv i Vi t 6 and Zt = + i= γ i C /αv i αv i i /αv i Vi t 7 SetI k,m = [ k m, k+ m,n m,j,k = Card{V i,i = j,, j+,v i I k,m },d k,m = max u I k,m αu and C = max t [,] C/αt αt Define It is easily seen that j+ M m,j,k t = γ i C /αvi αv i i /αvi [,t] Ik,m V i i= j 6
7 sup s,t I k,m For t,, let α n t = n j+ γ i C /αvi αv i i /αvi [s,t V i i= sup j ] n Vi t t [ n i= M m,j,k t t [,] 8 denote the empirical process, and w n a = sup α n t α n s denote the oscillation modulus of α n We apply Lemma 4 t s a of [5] with m 3,a = m,s = m /4 j,n = j,δ = This yields that there exists M N such that m M, jm with m jm m such that j jm : m /4 j j m and P sup α jt α js > m/4 j 56 m m e j 64 9 t s m m j+ We need to estimaten m,j,k and sup γ i C /αvi s,t I αv i i /αvi [s,t V i k,m i= fork =,,m j and j m Study of N m,j,k for m j jm : Set X i = k m V i and n = k+ j X i i is an iid sequence of Bernoulli random m variables with parameter p = For m j jm, one has mj j m = np, and m thus, for a >, using a classical bound on the sum of iid Bernoulli random variables, P N m,j,k > a mj j+ P X i a j m i= j j+ = P X i anp i= j As a consequence, P jm j=m m k= a n a j { Nm,j,k > a mj } 7 m j=m m k= m + j=m a a a j a j a m
8 Choosing a = 3, Borel Cantelli lemma entails that jm m { P lim sup Nm,j,k > 3 mj } = m + Study of N m,j,k for j jm: j=m P N m,j,k > j m = P j using 9 As a consequence, P + j=jm k= α j j P w j k + α m j m j m = P w j m j m m P w j m m/4 j m m k= Borel Cantelli lemma yields Study of sup s,t I k,m P 56 m m e j 64 lim sup m + { Nm,j,k > j m} + j=jm m k= + j=m 564 m { Nm,j,k j m} k + j m j m m m 56 m m e j 64 m e m 64 e m 64 = j+ γ i C /αvi αv i i /αvi [s,t V i for m j jm : i= j Almost surely, there exists m such that, for m m and for m j jm, k =,, m, N m,j,k 3 mj, and s,t I k,m, i = j,, j+, thus sup s,t I k,m γ i C /αv i αv i i /αvi [s,t V i C j d k,m j+ γ i C /αvi αv i i /αvi [s,t V i i= 3 mjc j j d k,m 8
9 j+ Study of sup γ i C /αvi s,t I αv i i /αvi [s,t V i for j jm: k,m i= j We consider the events and F m,j,k = sup s,t I k,m G m,j,k = E m,j,k = { N m,j,k j m}, j+ γ i C /αvi αv i i /αvi [s,t V i i= jc j m j j d k,m { sup M m,j,k t t [,] jc } j m j d k,m Relation 8 entails that PF m,j,k E m,j,k PG m,j,k E m,j,k In the following computation, l corresponds to the number of terms V i belonging to I k,m, andncorresponds to the number of thosev i among them that contribute to the supremum of G m,j,k G m,j,k E m,j,k j m l= l,,l l [ j, j+ ] Using independence of the V i, PG m,j,k E m,j,k j m l= l,,l l [ j, j+ ] G m,j,k i {l,l l } P G m,j,k V i I k,m i {l,l l } i/ {l,l l } V i I k,m P V i / I k,m i/ {l,l l } Now, P V i / I k,m = j l Let us fix an order on the V m i belonging i/ {l,l l } to I k,m : there are l! possibilities, all equiprobable, and thus P G m,j,k V i I k,m = l!p G m,j,k k < V m l < < V ll < k + m i {l,l l } V i / I k,m Let A l k,m denote the event { k m < V l < < V ll < k+ m } Then P l n G m,j,k A l k,m P A l k,m γ li C /αv l i αv li l /αv l i i jc j m n= i= j d k,m l n γ li C /αx i αx i l /αx i i jc n dx n= i= j dx l d k,m k m <x <<x l < k+ m P The probability inside the integral above may be estimated with the help of Lemma 5 in [7]: 9
10 One then computes PG m,j,k E m,j,k n P i= j m l= j m l= e j 8 C /αx i αx γ i li C l,,l l [ j, j+ ] j d k,m l /αx i i j n e j 8 l l! mj l n= j! l! j l! l l!le j 8 mj j m l= 4 j m e j 8 4 j m e j 8 j! l l! j l! l mj j m l= ml e j 8 ml l! j! l! j l! l mj ml dx dx l k m <x <<x l < k+ m As a consequence, P + j=jm m k= F m,j,k E m,j,k 4 + j=m + e j j=m j e j 8 for m Borel Cantelli lemma entails that P lim sup m + + j=jm m k= F m,j,k = Computation of the Hölder exponent: Let t,, t / S and let U be an open interval of, containing t Denote = max αt If δt =, then h Yt = and the formula holds Suppose now δt > t U Let ε > be such that δt > ε and > ε There exists i N such that i i, t V i Choose m large enough so that < min{ t V iδt ε m i,i =,,i } Let j = [mδt ε] Increasing m if necessary, we may and will assume that i j, and j j, j jε Let s, be such that t s < There exists m+ m+ k {,, m } such that t,s Ik,m Increasing again m if necessary, we may assume that I k,m U Then d k,m, and, for i j, [s,t V i = One computes:
11 + Zt Zs = γ i C /αv i αv i i /αvi [s,t V i i= + j+ = γ i C /αvi αv i i /αvi [s,t V i j=j i= j jm j+ γ i C /αvi αv i i /αvi [s,t V i j=j i= + j jm 3 mj + jc j m + j=j j d k,m j=jm j d k,m 3+ C + j m m 6+C m m 6+C m m K U,ε,C m m Since δt ε >, [ j j=j + jε+ j=j j d k,m j ε+ ε+ +δt εε+ ] + j=jm K U,ε,C log t s t s δt ε ε+ δt ε Zt Zs K U,ε,C log t s t s δt ε ε Let us now study W recall 6: there exists i such that, for all i i, One then computes: Γ /αv i i i /αv i i K c,d + d k,m K c,d i + j+ γ i C /αvi αv i i /αvi [s,t V i i= j
12 Wt Ws = + j=j jm j+ γ i C /αvi αv i Γ /αv i i i /αvi [s,t V i i= j mj + K c,d C j m j + + jε j=j j=j 3K c,dc K m + j=jm + K m + j + j=j j K log t s mδt εε ε +K t s mδt ε K log t s t s δt ε + ε +K t s +δt ε Gathering our results, we have shown that: Bt Bs K log t s t s δt ε ε In other words, h B t δt for any open interval U containing t Letting the diameter of U go to, one gets h B t δt αt Proof of Lemma 7: Lemma entails that Y is almost surely a càdlàg process Thus, for all t,, h Y t Proof of Lemma 8: We seek to compute the Hausdorff dimension of F = {t [,]\S : δt = } S j+ ] Let E γ = limsup [V i i γαv i,v i +i γαv i Since E γ {t : h Y t γ}, j + Now, i i= j ] [V i i γd,vi +i γd {t [,]\S : δt = } γ>e γ i is a covering of E γ, and thus dim H E γ γd As a consequence, dim H {t [,]\S : δt = } = Since dim H S =, we find that f H = Proof of Lemma 9: Following [4], set λ i = +Γ i For the system of points P = {V i,λ i } i and t [,], define the approximation rate of t by P as δ t P = sup{δ : t belongs to an infinite number of balls BV i,λ δ i } Let us show that δ t P = δt Γ In that view, note first that, since lim i i + i = almost surely,
13 δt = inf liminf log +Γ φi V φn n R t i log V φi t Let δ be such that t belongs to an infinite number of balls BV i,λ δ i There exists φ : N N such that V φn n R t and t V φi λ δ φi, ie log +Γ φi log V φi As a t δ consequence δt Taking the supremum over all admissible δ, one gets δ δ tp δt For the reverse inequality, consider two cases: δt = : since δ t P, one gets δ t P δt ; δt < : choose ε > such that δt +ε There exists φ : N N such that for all large enough i N, log +Γ φi log V φi δt+ε, ie t V t φi By definition, this entails δ t P δt+ε, and finally δ tp δt +Γ φi δt+ε by letting ε go to We now apply Theorem of [4]: since hd, one has hαt for all t, The function t is continuous and thus hαt ie f H h = hd dim H {t, : δ t P = } = sup{hαt,t,} = hd hαt Proof of Lemma : Since hc < < hd and α is C, there exist t,t, such that hαt = and hαt for all t I := t,t or t,t Define g : t min, Theorem in [4] yields: hαt dim H {t I : δ t P = gt} = sup{,t I} gt = sup{hαt,t I} = hαt = One gets that f H h dim H {t I : δt = hαt} =, ie f H h = Proof of Lemma : By definition and Lemma 3, F /c = {t [,] : δt = αt } = {t [,] : αt = c} {t [,] : δt = } c Set E = {t [,] : αt = c}, E = {t [,] : δt < } and E = {t [,] : δt = } Then [,] = E E, F /c = E E and thus f H c dim HE Now, if dim H E =, the lemma holds true since F /c is not empty and thus f H /c Suppose then that dim H E > Choose s < dim H E This implies that H s E = +, and Theorem 4 in [] entails that there exist a compact set E c E such that < H s E c < + 3
14 Set µ s = H s E c This is a finite and positive Borel measure on [,] Theorem 37 in [8], along with Lemmas 4 and 6, entail that for all t,, Pt E = and thus Pt E c E = t Ec One computes: [ E[µ s E ] = E ] t E µ s dt = E[ t E ]µ s dt = Pt E µ s dt E c = Thus, µ s E is a positive random variable with vanishing expectation: almost surely, µ s E = Since µ s E c = µ s E c E +µ s E c E, one obtains that, almost surely, µ s E c = µ s E c E Now, H s E E H s E c E = H s E c > Thus dim H E E s, s dim H E and dim H E E dim H E Proof of Lemma : For all t [,], αt c and almost surely, for all t [,], δt, thus, almost surely, for allt [,], h Y t c As a consequence, forh > c, F h = andf H h = 3 Large deviation and Legendre multifractal spectra We compute in this section the large deviation and Legendre multifractal spectra of the process B on an interval Recall that we consider the process on [,], and that the large deviation multifractal spectrum of a process X on [,] is the random function f g defined on R by logn f g β = lim ε +liminf n εβ, where, for a positive integer n and ε >, N ε n log Xj+ n β = #{j {,,n } : β ε Xj n β +ε} Other large deviation multifractal spectra can be defined by replacing the increments X j+ n Xj by other measures of the variation of X, such as its oscillations, but we n will not consider these in this work We do not recall the definition of the Legendre multifractal spectrum, and refer the reader to [, ] instead We shall denote Pn j = P β ε log Yj+ n Yj n β +ε + = P Yj nβ+ε n Yj n n β ε 4
15 Set also X j = { n β+ε Bj+ n B j n n β ε} which follows a Bernoulli law with parameter P j n Clearly, N ε nβ = For U an open interval of,, we write n X j j= J n U = {j : j n U} There exists a constant K U > such that, for n large enough, #J n U K U n Finally, we will make use of the characteristic function of B, which reads [9]: m E exp i θ j B = exp m αs θ j [,tj ]s ds j= where m N,θ,,θ m R m,t,,t m R m 3 Main result The large deviation and Legendre multifractal spectra of B are described by the following theorem: Theorem 3 With probability one, the large deviation and Legendre multifractal spectra of B satisfy: for β < ; βd for β [, ]; d f g β = f l β = for β, ]; d c j= + c β for β,+ ]; c c for β > c The fact that f l = f g stems from the general result that f l is always the concave hull of f g when the set {β : f g β } is bounded The part concerning f g in Theorem 3 follows from a series of lemmas that are proven in the next sections We note in passing that, comparing with Theorem, we see that the weak multifractal formalism holds for B, but the strong one does not, that is, f H f g and f H f g The decreasing part with slope - for large exponents present inf g but not inf H is a common phenomenon when variations are measured with increments In order to prove Theorem 3, we will first show in each case of that the equality holds true for any given β with probability one Permuting for all β and almost surely will then often be achieved thanks to the two following general simple but useful lemmas on the large deviation spectrum, which are of independent interest 5
16 Lemma 4 The large deviation spectrum of any real function is an upper semicontinuous function Proof Let f g be the large deviation spectrum of a real function Consider β R, x j j a sequence such that lim x j = β, and set ε j = sup β x k j + k j Note first that lim liminf lognnβ ε = lim ε liminf j + logn ε j n β For all j and all l j, N ε j n β N ε j n x j N ε l n x j As a consequence, Letting l tend to infinity, one gets liminf one finally obtains logn ε j n β liminf logn ε j n β f g β limsupf g x j j + logn ε l n x j f g x j and letting j tend to infinity Lemma 5 Assume that there exist four functions h,h,g,g with lim gu = lim gu = u u such that, for all β in some interval I and all sufficiently small ε >, almost surely hβ+gε liminf logn ε nβ hβ+gε Then, almost surely, for all β in I, hβ f g β hβ Proof Define M n β,β = #{j {,,n } : β log Xj+ n Xj n Then, for β < β < β, and The set N β β β β n β M n β,β N β β β β n β f g β = inf liminf β,β Q,β <β<β A = β,β Q Mβ,β logm n β,β β } is countable, and, thus f g is obtained by a countable infimum for all β I This yields the result 6
17 3 Preliminary lemmas 3 Statements Define Ht = H λ,p t = λt log p+pe t Lemma 6 If < p < λ <, then Lemma 7 If < λ < p <, then supht = λlog λ t> p + λlog λ p supht = λlog λ t< p + λlog λ p Lemma 8 If p = pn = Kn b and λ = λn = n a, where K > and > a > b, then there exists n N such that, for all n n, sup t> Ht a b n a Lemma 9 If p = pn = K n b and λ = λn = K n a, where K >, K > and > b > a, then there exists n N such that, for all n n, supht K t< 4 nb Lemma If p = pn = Kn b and λ = λn = n a, where K > and > a > b, then there exists n N such that, for all n n, inf t> e λtn p+pe t n e a b n +a Lemma If p = pn = K n b and λ = λn = K n a, where K >, K > and > b > a, then there exists n N such that, for all n n, inf t< e λtn p+pe t KUn e K K U n +b 4 Lemma Assume there exist b, and K > such that, for all n n and for all j,n, P j n Knb Then, almost surely, logn ε nβ +b Lemma 3 Assume there exist an open interval U, a real b, and K > such that, for all n n and for all j J n U, P j n Knb Then, almost surely, logn ε nβ +b 7
18 3 Proofs Proof of Lemma 6 supht = Ht where t = log t> Proof of Lemma 7 supht = Ht where t = log t< λ p p λ λ p p λ Proof of Lemma 8 With t = log λ p, one has, for n large enough, p λ n Ht = n a a b n log + n a a log K Kn b n = n a a b log + n a log + Knb n a K Kn b n n a a b Kn log + n a b n a K Kn [ b ] = a bn a +n a n a Knb a logk Kn b Since lim na Knb a Kn b logk = logk, there exists n N such that, for all n n, n a Ht a b+ [ ] n a Knb a logk Kn b a b Proof of Lemma 9 With t = log λ p, one has, for n large enough, p λ Ht = K n a K n a b log + K n a log + K n b K n a K K n b K n a K n a b log + K K n a K n b K n a K n b = n [K b n a b K n a b log + K n a K ] K n a b K n b K which yields the result since lim K n a b log K n a b K + K n a K K n a b K = K n b Proof of Lemma One has Lemma 8 then implies that, for n n, inf t> e λtn p+pe t n = inf t> e nht = e nsup t>ht inf t> e λtn p+pe t n e a b n +a 8
19 Proof of Lemma Write inf t< e λtn p+pe t K Un = e K UnsupHt t< where Ht = λ K U t log p+pe t Lemma 9 ensures that, for n n, inf t< e λtn p+pe t n e K K U n 4 +b Proof of Lemma Fix a b, Then, for all t >, P N ε n β n+a = P e t n j= e ntλ E X j e tn +a [ n where λ = n a The X j are independent and E [ e tx j] = P j n +P j n et, thus j= e tx j P n Nn ε β n+a e ntλ Pn j +Pj n et, t > For t >, the function p p+pe t is increasing and so, by assumption on P j n, j= P N ε nβ n +a e ntλ p+pe t n, where p = Kn b Minimizing over t > and using Lemma, one gets and thus n n,p N ε n β n+a e a b n +a P Nn ε β n+a < + n N The Borel-Cantelli lemma then ensures that, almost surely, or n N, n n,nn ε β n+a logn ε n β +a Since this inequality holds true for any a b, one has indeed that, almost surely, ] logn ε n β +b 9
20 Proof of Lemma 3 Fix a < b For all t <, P Nn ε β n+a P = P e tn+a j J nu e t j JnU X j n +a j J nu X j e tn +a P j n +P j ne t When t <, the function p p + pe t is decreasing As a consequence, by assumption on P j n and with p = Knb, λ = n a, one has, for n large enough, P N ε n β n+a e ntλ p+pe t #JnU Minimizing over t < and using Lemma, one gets and thus e ntλ p+pe t K Un n n,p N ε nβ n +a e K K U 4 n +b P Nn ε β n+a < + n N As in the proof of Lemma, this leads to and finally, almost surely, a < b, almost surely, logn ε nβ logn ε n β +b +a 33 Estimates of P j n For U an open interval, denote c U = inf t U αt and = sup t U αt Set also = j n 33 Lemmas Lemma 4 Assume β < d Then, K >, n N such that n n, j,n, P j n Kn αβ+α ε Lemma 5 Assume β < Then, K >, n N such that n n, j J n U, P j n Kn αβ α ε Lemma 6 Assume β > c and ε,β c Then, K >, n N such that n n, j,n, P j n Kn α +ε β Lemma 7 Assume β > c U and ε,β c U Then, K >, n N such that n n, j J n U, Pn j Kn α +ε β
21 33 Proofs Proof of Lemma 4 Set µ j = α β + α ε Using the truncation inequality [, Section 3, p 9], one computes Pn j = P n Y+ Yt β+ε j P Y+ Y n β+ε 7 n β+ε = n β+ε e + e + + ξ αx dx dξ n β ε n β+εαx v αx dx dv n β+εαx v αx dxdv n β+εαx dx = 7n µ j + n +β+εαx αtj dx Since α is C, there exists a constant K such that, for all x,+, As a consequence, for a constant K β +ε αx α K x K n P j n 7n µ j + = 7n µ j n K n K n µ j n + K n dx n β ε Proof of Lemma 5 Set µ = + n β+ε n and σ = β ε Choose a function ϕ that satisfies the following properties: suppϕ [,] n β+ε
22 ϕ is even 3 ϕ is C 4 4 x R, ϕx 5 ϕ is not identically These properties imply in particular that ˆϕ is real and even In addition, for all x R: ϕ x+µ x µ +ϕ σ σ [ n β+ε, n ]x+ β ε [n β ε,n ]x β+ε Since the Fourier transform of ϕ x µ σ is σexp iµξˆϕσξ, Parseval formula yields P j n R π σ ϕ x+µ σ +ϕx µ σ PY+ Y dx cosµξˆϕσξe + t ξ αx dx j dξ = π R + cos µ + σ ηˆϕηe t η j σ αx dx dη Now, cos µ σ ηˆϕηdη = ϕµ σ = since µ > σ One may thus write: R P j n π + cos µ + σ ηˆϕηe t η j σ αx dx dη =: π nαβ α ε I j n, where I j n = I j n, +I j n, +I j n,3 with I j n, = + ˆϕη cos µ σ η cosη n αβ+εα e + η σ αx dx dη, I j n, = + I j n,3 = ˆϕηcosηn αβ+εα + tj+ ˆϕηcosηn αβ+εα e + η σ αx dx + η σ αx dxdη tj+ η σ αx dx dη,
23 Let us show that there exists K U > such that, for all j J n U, tj+ n αβ+εα η σ αx dx K U η c +η d Since σ = nβ+ε n ε and, there exits K such that, for all x,+, αx α K n, one has σ α αx K and tj+ n αβ+εα As a consequence, I j n, η σ αx dx = n αtjβ+εαtj tj+ n η αx σ αtj αx dx σ α + α n εα n ε α ηc +η d n Kη c +η d ˆϕη µ σ n αβ+εα K µ + σ ˆϕη η c +η d dη tj+ tj+ η σ αx dxdη σ α αx dx One finally obtains that sup I j n, K and lim sup I j n ε n, = j J nu j J nu Let us now deal with In, j I j n, + K U ˆϕη n αβ+εα tj+ η σ αx dx dη + K U n αβ ε α K U n β ε and thus lim sup I j n, = j J nu Fubini s theorem implies that I j n,3 = = ˆϕη n αβ+εα η c +η d n αβ εα dη tj+ n αtjβ+εαtj σ αx tj+ n αtjβ+εαtj σ αx + Fαxdx 3 ˆϕηcosηη αx dη dx
24 + where Fδ = ˆϕηcosηη δ dη The appendix contains a proof that min Fδ > As a consequence, δ [c,d] I j n,3 min Fδn αtjβ+εαtj δ [c,d] σ α ] αtj = min Fδ δ [c,d] K U [ n ε n ε [ n ε n ε ] [ c n tj+ tj+ n σ αtj αx dx n tj+ e α αxlogσ dx e α αxlogσ dx+ ] Now lim consequence, n ε n ε = and lim sup j J nu n + e αtj αxlogσ dx = As a and thus K U >, n N : n n, inf j J nu Ij n,3 K U > inf j J n αβ+εα Pn j > nu Proofs of Lemma 6 and Lemma 7 Parseval formula yields Pn j = sin ξ sin ξ n β ε n β+ε e + t ξ αx dx j dξ π ξ Set µ j = one gets Define α R +ε β with ε,β c U Using the change of variable ξ = n /α v, P j n = π R sinn µ j v sinn µ j ε v v I j n = R + tj e αx v αx n + αx v αx αt n j dx t e j dv α dx dv Since α is Lipschitz and c αx d, one deduces that Now, KU >, K U >, n N, KU inf j J Ij n sup In j K U nu j J nu P j n = n µ j I j n + I j n = n µ j I j n +Lj n R sinnµ j v sinn µ j ε v n µ j v 4 + tj e αx v αx αt n j dx dv
25 One computes: L j n [ sinnµ j v KU R t j+ KU R 6 nµ j v + n εe tj 6K U n +ε β c U ] n µ j v n µ j v + sinnµj ε v R + v tj e αx v αx n α dx dv + tj e αx v αx αt n j dx dv + K U KU There exist K 3 U R and n N such that, for all n n, t j+ sup v tj e j J nu R As a consequence, lim sup L j n = j J nu αx v αx n α dx dv K 3 U < + αx v αx n n ε α dx dv 34 Estimates for the number of increments and determination of the spectrum 34 Lemmas Lemma 8 Almost surely, β <, f g β = Lemma 9 Almost surely, β, d, f gβ = βd Lemma 3 Almost surely, f g = Lemma 3 Almost surely, β [ d, c ], f gβ = Lemma 3 Almost surely, β c,+ c, f gβ = + c β Lemma 33 Almost surely, f g + c = Lemma 34 Almost surely, β > + c, f gβ = 34 Proofs Proof of Lemma 8 Fix β < Denote E β =, β If ε E β, then PNn ε β = PNε n β = n = Pn j j= 5
26 Lemma 4 implies that, for n large enough, PN ε nβ Kn dβ+dε n e 3 Kndβ+dε 3 Kndβ+dε, and thus lim PNε nβ = Since Nnβ ε tends to in probability when n tends to infinity, there exists a subsequence σn such that Nσn ε β tends to almost surely logn This implies that, almost surely, liminf nβ ε = We have proved that: logn β <, ε E β, almost surely, liminf nβ ε Let Ω β,ε = {ω : holds}, Ω β = Ω β,ε and Ω = ε E β Q β, Q probability, and consider ω Ω Choose β < and j N Set β j = [βj] j enough, N /j n β N /j n β j In addition liminf n logn /j n β n logn /j n β j log n = and, almost surely, for all β <, f g β = = Ω β Note that Ω has For j large = As a consequence, Proof of Lemma 9 Fix β, Denote E d β = {ε,minβ, β such that d d εβ ε,} Choose ε E β By Lemma 4, there exists K > and n N such that, for all n n and all j,n, Lemma then implies that, almost surely, P j n Kn dβ+dε logn ε n β dβ +dε 3 There exists an open interval U such that, for all t U, αt d ε Using Lemma 5, there exist K > and n N such that, for all n n and all j J n U, P j n Kn d εβ ε, and Lemma 3 then implies that, almost surely, logn ε n β d εβ ε We thus have proved that, for all β, d and all ε E β, almost surely, d εβ ε liminf logn ε n β dβ +dε 4 Then Lemma 5 ensures that almost surely, for all β, d, f gβ = dβ 6
27 Proof of Lemma 3 We obtain Inequality 3 for β = by applying Lemma 4 and Lemma This yields that, almost surely, logn ε n dε As a consequence, f g Then apply Lemma 4 and Lemma 9 to obtain f g limsupf g j + j d = limsup j + j = Proof of Lemma 3 Let β, and ε, d c Choose an open interval U such that β > c U and β < +ε Lemma 7 then ensures that there exist K > and n N such that, for all n n and all j J n U, P j n Kn α +ε β Kn +ε β Kn ε By Lemma 3, for all β,, almost surely, d c logn ε n β We conclude by applying Lemma 5 For β =, we apply Lemma 4 to obtain c ε f g c limsup f g [βj] j + j = limsup j + For β =, Lemma 4 and Lemma 9 lead to d f g d limsup f g [βj] j + j = limsupd [βj] j + j = 7
28 Proof of Lemma 3 Let β, + Denote E c c β =,minβ, + β Fix c c ε E β By Lemma 6, there exist K > and n N such that, for all n n and all j,n, Pn j Kn c +ε β Lemma then implies that, almost surely, logn ε nβ + +ε β 5 c Choose an open interval U such that c ε β ε By Lemma 7, there exist K > and n N such that, for alln n and all j J n U, Lemma 3 then implies that, almost surely, logn ε n β P j n Kn +ε β + +ε β + c ε β We have proved that, for all β c,+ c and all ε E β, almost surely, + c ε β liminf logn ε nβ + c +ε β 6 The result then follows from Lemma 5 Proof of Lemma 33 We obtain Inequality 5 for β = + c Lemma This yields that, almost surely, by applying Lemma 6 and logn ε n + c dε As a consequence, f g + c Then, Lemma 4 and Lemma 3 lead to f g + c limsup j + f g [+ j] c j = limsup+ j + c [+ j] c j = Proof of Lemma 34 Let β > + and denote E c β =,β For ε E c β, PNnβ ε = PNnβ ε = n = Pn j j= 8
29 Lemma 6 ensures that, for n large enough, PN ε nβ Kn c +ε β n 3 Kn+ c +ε β Thus, lim PNε nβ =, which implies that Nnβ ε tends to in probability when n tends to infinity There exists a subsequence σn such that Nσn ε β tends to almost surely As a consequence, for all β > + and all ε E c β, almost surely, logn ε n β We conclude as in the last part of the proof of Lemma 8 Appendix The following result is due to R Schelling [4]: Lemma 35 For all β >, Fβ = Proof The Lévy Khintchine formula yields η β = cosyην β dy with By Fubini s theorem, y = 7 η β cosη ϕηdη > 8 ν β dy = c βdy y +β η β cosη ϕηdη = cosη cosyηcosη ϕηdην β dy y = cosη cos+yη cos yη ϕηdην β dy y = c ϕ ϕ+y ϕ y ν β dy, y where c is a positive constant By definition, ϕ is smooth and supported on[, ], thus ϕ = As a consequence, we find that η β cosη ϕηdη = c ϕ+y+ϕ yν β dy < This is inequality 8 y It is easy to see that the function β Fβ is continuous min δ [c,d] Fδ > As a consequence, 9
30 References [] Arbeiter M and Patzschke N 996 Random self-similar multifractals Math Nachr 8, p 5-4 [] Ayache, A 3 Sharp estimates on the tail behavior of a multistable distribution, Statistics and Probability Letters, 83, 3, p [3] Balança, P 4 Fine regularity of Lévy processes and linear multifractional stable motion Electron J Probab 9, Article, p -37 [4] Barral, J, Fournier, N, Jaffard, S and Seuret, S A pure jump Markov process with a random singularity spectrum Annals of Probability 38 5, p [5] Barral, J and Lévy Véhel, J 4 Multifractal analysis of a class of additive processes with correlated non-stationary increments Electron J Probab 9, Article 6, p [6] Biermé, H and Lacaux, C 3 Linear multifractional multistable motion: LePage series representation and modulus of continuity, Ann Univ Bucharest Math Series 4 LXII, p [7] Brown G, Michon G and Peyrière, J 99 On the multifractal analysis of measures, J Statist Phys , p [8] Durand, A 9 Singularity sets of Lévy processes Probab Theory Related Fields , p [9] Durand, A and Jaffard, S Multifractal analysis of Lévy fields Probab Theory Related Fields 53, p [] Falconer, K 99 Fractal Geometry: Mathematical Foundations and Applications John Wiley, New York [] Falconer, K Tangent fields and the local structure of random fields J Theoret Probab 5, p [] Falconer, K 3 The local structure of random processes J London Math Soc 67, p [3] Falconer, K and Lévy Véhel, J 9 Multifractional, multistable, and other processes with prescribed local form J Theoret Probab, p [4] Falconer, K and Liu, L Multistable Processes and Localisability Stochastic Models, 8 p [5] Jaffard, S 997 Old friends revisited The multifractal nature of some classical functions J Fourier Analysis App 3, p - [6] Jaffard, S 999 The multifractal nature of Lévy processes Probab Theory Related Fields 4, p7-7 3
31 [7] Ledoux, M and Talagrand, M 996 Probability in Banach spaces Springer-Verlag [8] Le Guével R and Lévy Véhel J 3 Incremental moments and Hölder exponents of multifractional multistable processes ESAIM PS DOI: [9] Le Guével, R, Lévy-Véhel, J and Lining, L On two multistable extensions of stable Lévy motion and their semimartingale representation J Theoret Probab, to appear, doi: 7/s [] Lévy Véhel, J and Vojak, R 998 Multifractal Analysis of Choquet Capacities: Preliminary Results Adv Appl Maths, p -43 [] Loeve, M 977 Probability Theory I 4th edn Springer, New York [] Molchanov, I and Ralchenko, K 5 Multifractional Poisson process, multistable subordinator and related limit theorems Stat Probab Lett 96, p 95- [3] Samorodnitsky G and Taqqu, M 994 Stable Non-Gaussian Random Process Chapman and Hall [4] Schelling, R Private communication [5] Stute, W 98 The oscillation behavior of empirical processes Annals of Probability, p
Statistical test for some multistable processes
Statistical test for some multistable processes Ronan Le Guével Joint work in progress with A. Philippe Journées MAS 2014 1 Multistable processes First definition : Ferguson-Klass-LePage series Properties
More informationFunctional Limit theorems for the quadratic variation of a continuous time random walk and for certain stochastic integrals
Functional Limit theorems for the quadratic variation of a continuous time random walk and for certain stochastic integrals Noèlia Viles Cuadros BCAM- Basque Center of Applied Mathematics with Prof. Enrico
More informationA Tapestry of Complex Dimensions
A Tapestry of Complex Dimensions John A. Rock May 7th, 2009 1 f (α) dim M(supp( β)) (a) (b) α Figure 1: (a) Construction of the binomial measure β. spectrum f(α) of the measure β. (b) The multifractal
More informationIntroduction 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 informationUniformly and strongly consistent estimation for the Hurst function of a Linear Multifractional Stable Motion
Uniformly and strongly consistent estimation for the Hurst function of a Linear Multifractional Stable Motion Antoine Ayache and Julien Hamonier Université Lille 1 - Laboratoire Paul Painlevé and Université
More informationTYPICAL BOREL MEASURES ON [0, 1] d SATISFY A MULTIFRACTAL FORMALISM
TYPICAL BOREL MEASURES ON [0, 1] d SATISFY A MULTIFRACTAL FORMALISM Abstract. In this article, we prove that in the Baire category sense, measures supported by the unit cube of R d typically satisfy a
More informationMAJORIZING MEASURES WITHOUT MEASURES. By Michel Talagrand URA 754 AU CNRS
The Annals of Probability 2001, Vol. 29, No. 1, 411 417 MAJORIZING MEASURES WITHOUT MEASURES By Michel Talagrand URA 754 AU CNRS We give a reformulation of majorizing measures that does not involve measures,
More informationOptimal series representations of continuous Gaussian random fields
Optimal series representations of continuous Gaussian random fields Antoine AYACHE Université Lille 1 - Laboratoire Paul Painlevé A. Ayache (Lille 1) Optimality of continuous Gaussian series 04/25/2012
More informationConvergence at first and second order of some approximations of stochastic integrals
Convergence at first and second order of some approximations of stochastic integrals Bérard Bergery Blandine, Vallois Pierre IECN, Nancy-Université, CNRS, INRIA, Boulevard des Aiguillettes B.P. 239 F-5456
More informationBrownian Motion. 1 Definition Brownian Motion Wiener measure... 3
Brownian Motion Contents 1 Definition 2 1.1 Brownian Motion................................. 2 1.2 Wiener measure.................................. 3 2 Construction 4 2.1 Gaussian process.................................
More informationHölder regularity for operator scaling stable random fields
Stochastic Processes and their Applications 119 2009 2222 2248 www.elsevier.com/locate/spa Hölder regularity for operator scaling stable random fields Hermine Biermé a, Céline Lacaux b, a MAP5 Université
More informationEULER MARUYAMA APPROXIMATION FOR SDES WITH JUMPS AND NON-LIPSCHITZ COEFFICIENTS
Qiao, H. Osaka J. Math. 51 (14), 47 66 EULER MARUYAMA APPROXIMATION FOR SDES WITH JUMPS AND NON-LIPSCHITZ COEFFICIENTS HUIJIE QIAO (Received May 6, 11, revised May 1, 1) Abstract In this paper we show
More informationBrownian motion. Samy Tindel. Purdue University. Probability Theory 2 - MA 539
Brownian motion Samy Tindel Purdue University Probability Theory 2 - MA 539 Mostly taken from Brownian Motion and Stochastic Calculus by I. Karatzas and S. Shreve Samy T. Brownian motion Probability Theory
More informationFinite-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 informationSEPARABILITY AND COMPLETENESS FOR THE WASSERSTEIN DISTANCE
SEPARABILITY AND COMPLETENESS FOR THE WASSERSTEIN DISTANCE FRANÇOIS BOLLEY Abstract. In this note we prove in an elementary way that the Wasserstein distances, which play a basic role in optimal transportation
More informationFractal Strings and Multifractal Zeta Functions
Fractal Strings and Multifractal Zeta Functions Michel L. Lapidus, Jacques Lévy-Véhel and John A. Rock August 4, 2006 Abstract. We define a one-parameter family of geometric zeta functions for a Borel
More informationErgodic Theorems. Samy Tindel. Purdue University. Probability Theory 2 - MA 539. Taken from Probability: Theory and examples by R.
Ergodic Theorems Samy Tindel Purdue University Probability Theory 2 - MA 539 Taken from Probability: Theory and examples by R. Durrett Samy T. Ergodic theorems Probability Theory 1 / 92 Outline 1 Definitions
More informationPACKING-DIMENSION PROFILES AND FRACTIONAL BROWNIAN MOTION
PACKING-DIMENSION PROFILES AND FRACTIONAL BROWNIAN MOTION DAVAR KHOSHNEVISAN AND YIMIN XIAO Abstract. In order to compute the packing dimension of orthogonal projections Falconer and Howroyd 997) introduced
More informationA Uniform Dimension Result for Two-Dimensional Fractional Multiplicative Processes
To appear in Ann. Inst. Henri Poincaré Probab. Stat. A Uniform Dimension esult for Two-Dimensional Fractional Multiplicative Processes Xiong Jin First version: 8 October 213 esearch eport No. 8, 213, Probability
More informationWiener Measure and Brownian Motion
Chapter 16 Wiener Measure and Brownian Motion Diffusion of particles is a product of their apparently random motion. The density u(t, x) of diffusing particles satisfies the diffusion equation (16.1) u
More informationReal Analysis Problems
Real Analysis Problems Cristian E. Gutiérrez September 14, 29 1 1 CONTINUITY 1 Continuity Problem 1.1 Let r n be the sequence of rational numbers and Prove that f(x) = 1. f is continuous on the irrationals.
More informationBrownian Motion. Chapter Stochastic Process
Chapter 1 Brownian Motion 1.1 Stochastic Process A stochastic process can be thought of in one of many equivalent ways. We can begin with an underlying probability space (Ω, Σ,P and a real valued stochastic
More informationZdzis law Brzeźniak and Tomasz Zastawniak
Basic Stochastic Processes by Zdzis law Brzeźniak and Tomasz Zastawniak Springer-Verlag, London 1999 Corrections in the 2nd printing Version: 21 May 2005 Page and line numbers refer to the 2nd printing
More informationMultiple points of the Brownian sheet in critical dimensions
Multiple points of the Brownian sheet in critical dimensions Robert C. Dalang Ecole Polytechnique Fédérale de Lausanne Based on joint work with: Carl Mueller Multiple points of the Brownian sheet in critical
More informationThe Codimension of the Zeros of a Stable Process in Random Scenery
The Codimension of the Zeros of a Stable Process in Random Scenery Davar Khoshnevisan The University of Utah, Department of Mathematics Salt Lake City, UT 84105 0090, U.S.A. davar@math.utah.edu http://www.math.utah.edu/~davar
More informationIndependence of some multiple Poisson stochastic integrals with variable-sign kernels
Independence of some multiple Poisson stochastic integrals with variable-sign kernels Nicolas Privault Division of Mathematical Sciences School of Physical and Mathematical Sciences Nanyang Technological
More informationPacking-Dimension Profiles and Fractional Brownian Motion
Under consideration for publication in Math. Proc. Camb. Phil. Soc. 1 Packing-Dimension Profiles and Fractional Brownian Motion By DAVAR KHOSHNEVISAN Department of Mathematics, 155 S. 1400 E., JWB 233,
More informationMulti-operator scaling random fields
Available online at www.sciencedirect.com Stochastic Processes and their Applications 121 (2011) 2642 2677 www.elsevier.com/locate/spa Multi-operator scaling random fields Hermine Biermé a, Céline Lacaux
More informationStochastic integration. P.J.C. Spreij
Stochastic integration P.J.C. Spreij this version: April 22, 29 Contents 1 Stochastic processes 1 1.1 General theory............................... 1 1.2 Stopping times...............................
More informationBLOWUP THEORY FOR THE CRITICAL NONLINEAR SCHRÖDINGER EQUATIONS REVISITED
BLOWUP THEORY FOR THE CRITICAL NONLINEAR SCHRÖDINGER EQUATIONS REVISITED TAOUFIK HMIDI AND SAHBI KERAANI Abstract. In this note we prove a refined version of compactness lemma adapted to the blowup analysis
More informationA NOTE ON THE COMPLETE MOMENT CONVERGENCE FOR ARRAYS OF B-VALUED RANDOM VARIABLES
Bull. Korean Math. Soc. 52 (205), No. 3, pp. 825 836 http://dx.doi.org/0.434/bkms.205.52.3.825 A NOTE ON THE COMPLETE MOMENT CONVERGENCE FOR ARRAYS OF B-VALUED RANDOM VARIABLES Yongfeng Wu and Mingzhu
More informationMandelbrot s cascade in a Random Environment
Mandelbrot s cascade in a Random Environment A joint work with Chunmao Huang (Ecole Polytechnique) and Xingang Liang (Beijing Business and Technology Univ) Université de Bretagne-Sud (Univ South Brittany)
More informationRandom measures, intersections, and applications
Random measures, intersections, and applications Ville Suomala joint work with Pablo Shmerkin University of Oulu, Finland Workshop on fractals, The Hebrew University of Jerusalem June 12th 2014 Motivation
More informationMultifractal analysis of Bernoulli convolutions associated with Salem numbers
Multifractal analysis of Bernoulli convolutions associated with Salem numbers De-Jun Feng The Chinese University of Hong Kong Fractals and Related Fields II, Porquerolles - France, June 13th-17th 2011
More informationOn the convergence of sequences of random variables: A primer
BCAM May 2012 1 On the convergence of sequences of random variables: A primer Armand M. Makowski ECE & ISR/HyNet University of Maryland at College Park armand@isr.umd.edu BCAM May 2012 2 A sequence a :
More informationThe exact asymptotics for hitting probability of a remote orthant by a multivariate Lévy process: the Cramér case
The exact asymptotics for hitting probability of a remote orthant by a multivariate Lévy process: the Cramér case Konstantin Borovkov and Zbigniew Palmowski Abstract For a multivariate Lévy process satisfying
More informationAdditive 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 informationPoisson random measure: motivation
: motivation The Lévy measure provides the expected number of jumps by time unit, i.e. in a time interval of the form: [t, t + 1], and of a certain size Example: ν([1, )) is the expected number of jumps
More informationStatistics and Probability Letters
Statistics and Probability Letters 83 (23) 83 93 Contents lists available at SciVerse ScienceDirect Statistics and Probability Letters journal homepage: www.elsevier.com/locate/stapro Fractal dimension
More informationFiltrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition
Filtrations, Markov Processes and Martingales Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition David pplebaum Probability and Statistics Department,
More informationRichard F. Bass Krzysztof Burdzy University of Washington
ON DOMAIN MONOTONICITY OF THE NEUMANN HEAT KERNEL Richard F. Bass Krzysztof Burdzy University of Washington Abstract. Some examples are given of convex domains for which domain monotonicity of the Neumann
More informationStochastic 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 informationA TWO PARAMETERS AMBROSETTI PRODI PROBLEM*
PORTUGALIAE MATHEMATICA Vol. 53 Fasc. 3 1996 A TWO PARAMETERS AMBROSETTI PRODI PROBLEM* C. De Coster** and P. Habets 1 Introduction The study of the Ambrosetti Prodi problem has started with the paper
More informationAsymptotic stability of an evolutionary nonlinear Boltzmann-type equation
Acta Polytechnica Hungarica Vol. 14, No. 5, 217 Asymptotic stability of an evolutionary nonlinear Boltzmann-type equation Roksana Brodnicka, Henryk Gacki Institute of Mathematics, University of Silesia
More informationMetric 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 informationAlmost sure limit theorems for U-statistics
Almost sure limit theorems for U-statistics Hajo Holzmann, Susanne Koch and Alesey Min 3 Institut für Mathematische Stochasti Georg-August-Universität Göttingen Maschmühlenweg 8 0 37073 Göttingen Germany
More informationA generic property of families of Lagrangian systems
Annals of Mathematics, 167 (2008), 1099 1108 A generic property of families of Lagrangian systems By Patrick Bernard and Gonzalo Contreras * Abstract We prove that a generic Lagrangian has finitely many
More informationDiffusion Approximation of Branching Processes
UDC 519.218.23 Diffusion Approximation of Branching Processes N. Limnios Sorbonne University, Université de technologie de Compiègne LMAC Laboratory of Applied Mathematics of Compiègne - CS 60319-60203
More informationStable Lévy motion with values in the Skorokhod space: construction and approximation
Stable Lévy motion with values in the Skorokhod space: construction and approximation arxiv:1809.02103v1 [math.pr] 6 Sep 2018 Raluca M. Balan Becem Saidani September 5, 2018 Abstract In this article, we
More informationProbability and Measure
Part II Year 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2018 84 Paper 4, Section II 26J Let (X, A) be a measurable space. Let T : X X be a measurable map, and µ a probability
More informationLARGE 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 informationTHE SKOROKHOD OBLIQUE REFLECTION PROBLEM IN A CONVEX POLYHEDRON
GEORGIAN MATHEMATICAL JOURNAL: Vol. 3, No. 2, 1996, 153-176 THE SKOROKHOD OBLIQUE REFLECTION PROBLEM IN A CONVEX POLYHEDRON M. SHASHIASHVILI Abstract. The Skorokhod oblique reflection problem is studied
More informationExtremal Solutions of Differential Inclusions via Baire Category: a Dual Approach
Extremal Solutions of Differential Inclusions via Baire Category: a Dual Approach Alberto Bressan Department of Mathematics, Penn State University University Park, Pa 1682, USA e-mail: bressan@mathpsuedu
More informationEstimation of density level sets with a given probability content
Estimation of density level sets with a given probability content Benoît CADRE a, Bruno PELLETIER b, and Pierre PUDLO c a IRMAR, ENS Cachan Bretagne, CNRS, UEB Campus de Ker Lann Avenue Robert Schuman,
More informationA generalization of Cramér large deviations for martingales
A generalization of Cramér large deviations for martingales Xiequan Fan, Ion Grama, Quansheng Liu To cite this version: Xiequan Fan, Ion Grama, Quansheng Liu. A generalization of Cramér large deviations
More informationDirichlet uniformly well-approximated numbers
Dirichlet uniformly well-approximated numbers Lingmin LIAO (joint work with Dong Han Kim) Université Paris-Est Créteil (University Paris 12) Hyperbolicity and Dimension CIRM, Luminy December 3rd 2013 Lingmin
More informationPointwise convergence rates and central limit theorems for kernel density estimators in linear processes
Pointwise convergence rates and central limit theorems for kernel density estimators in linear processes Anton Schick Binghamton University Wolfgang Wefelmeyer Universität zu Köln Abstract Convergence
More informationSome 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 informationMATHS 730 FC Lecture Notes March 5, Introduction
1 INTRODUCTION MATHS 730 FC Lecture Notes March 5, 2014 1 Introduction Definition. If A, B are sets and there exists a bijection A B, they have the same cardinality, which we write as A, #A. If there exists
More informationEstimation of a quadratic regression functional using the sinc kernel
Estimation of a quadratic regression functional using the sinc kernel Nicolai Bissantz Hajo Holzmann Institute for Mathematical Stochastics, Georg-August-University Göttingen, Maschmühlenweg 8 10, D-37073
More informationPROBABILITY: LIMIT THEOREMS II, SPRING HOMEWORK PROBLEMS
PROBABILITY: LIMIT THEOREMS II, SPRING 218. HOMEWORK PROBLEMS PROF. YURI BAKHTIN Instructions. You are allowed to work on solutions in groups, but you are required to write up solutions on your own. Please
More informationNotation. General. Notation Description See. Sets, Functions, and Spaces. a b & a b The minimum and the maximum of a and b
Notation General Notation Description See a b & a b The minimum and the maximum of a and b a + & a f S u The non-negative part, a 0, and non-positive part, (a 0) of a R The restriction of the function
More informationM. Ledoux Université de Toulouse, France
ON MANIFOLDS WITH NON-NEGATIVE RICCI CURVATURE AND SOBOLEV INEQUALITIES M. Ledoux Université de Toulouse, France Abstract. Let M be a complete n-dimensional Riemanian manifold with non-negative Ricci curvature
More informationDoléans measures. Appendix C. C.1 Introduction
Appendix C Doléans measures C.1 Introduction Once again all random processes will live on a fixed probability space (Ω, F, P equipped with a filtration {F t : 0 t 1}. We should probably assume the filtration
More informationDirichlet uniformly well-approximated numbers
Dirichlet uniformly well-approximated numbers Lingmin LIAO (joint work with Dong Han Kim) Université Paris-Est Créteil (University Paris 12) Approximation and numeration Université Paris Diderot-Paris
More informationMATH 220 solution to homework 4
MATH 22 solution to homework 4 Problem. Define v(t, x) = u(t, x + bt), then v t (t, x) a(x + u bt) 2 (t, x) =, t >, x R, x2 v(, x) = f(x). It suffices to show that v(t, x) F = max y R f(y). We consider
More informationConvergence of Feller Processes
Chapter 15 Convergence of Feller Processes This chapter looks at the convergence of sequences of Feller processes to a iting process. Section 15.1 lays some ground work concerning weak convergence of processes
More informationGeometric ρ-mixing property of the interarrival times of a stationary Markovian Arrival Process
Author manuscript, published in "Journal of Applied Probability 50, 2 (2013) 598-601" Geometric ρ-mixing property of the interarrival times of a stationary Markovian Arrival Process L. Hervé and J. Ledoux
More informationSTOCHASTIC ANALYSIS FOR JUMP PROCESSES
STOCHASTIC ANALYSIS FOR JUMP PROCESSES ANTONIS PAPAPANTOLEON Abstract. Lecture notes from courses at TU Berlin in WS 29/1, WS 211/12 and WS 212/13. Contents 1. Introduction 2 2. Definition of Lévy processes
More informationHardy-Stein identity and Square functions
Hardy-Stein identity and Square functions Daesung Kim (joint work with Rodrigo Bañuelos) Department of Mathematics Purdue University March 28, 217 Daesung Kim (Purdue) Hardy-Stein identity UIUC 217 1 /
More informationStability of optimization problems with stochastic dominance constraints
Stability of optimization problems with stochastic dominance constraints D. Dentcheva and W. Römisch Stevens Institute of Technology, Hoboken Humboldt-University Berlin www.math.hu-berlin.de/~romisch SIAM
More informationGärtner-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 informationSTAT 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 informationCalculus of Variations. Final Examination
Université Paris-Saclay M AMS and Optimization January 18th, 018 Calculus of Variations Final Examination Duration : 3h ; all kind of paper documents (notes, books...) are authorized. The total score of
More informationAn essay on the general theory of stochastic processes
Probability Surveys Vol. 3 (26) 345 412 ISSN: 1549-5787 DOI: 1.1214/1549578614 An essay on the general theory of stochastic processes Ashkan Nikeghbali ETHZ Departement Mathematik, Rämistrasse 11, HG G16
More informationarxiv: v1 [math.pr] 22 Dec 2018
arxiv:1812.09618v1 [math.pr] 22 Dec 2018 Operator norm upper bound for sub-gaussian tailed random matrices Eric Benhamou Jamal Atif Rida Laraki December 27, 2018 Abstract This paper investigates an upper
More informationWEAK CONVERGENCE TO THE TANGENT PROCESS OF THE LINEAR MULTIFRACTIONAL STABLE MOTION. Stilian Stoev, Murad S. Taqqu 1
Pliska Stud. Math. Bulgar. 17 (2005), 271 294 STUDIA MATHEMATICA BULGARICA WEAK CONVERGENCE TO THE TANGENT PROCESS OF THE LINEAR MULTIFRACTIONAL STABLE MOTION Stilian Stoev, Murad S. Taqqu 1 V pamet na
More informationFRACTAL BEHAVIOR OF MULTIVARIATE OPERATOR-SELF-SIMILAR STABLE RANDOM FIELDS
Communications on Stochastic Analysis Vol. 11, No. 2 2017) 233-244 Serials Publications www.serialspublications.com FRACTAL BEHAVIOR OF MULTIVARIATE OPERATOR-SELF-SIMILAR STABLE RANDOM FIELDS ERCAN SÖNMEZ*
More informationLecture 9. d N(0, 1). Now we fix n and think of a SRW on [0,1]. We take the k th step at time k n. and our increments are ± 1
Random Walks and Brownian Motion Tel Aviv University Spring 011 Lecture date: May 0, 011 Lecture 9 Instructor: Ron Peled Scribe: Jonathan Hermon In today s lecture we present the Brownian motion (BM).
More informationRegular Variation and Extreme Events for Stochastic Processes
1 Regular Variation and Extreme Events for Stochastic Processes FILIP LINDSKOG Royal Institute of Technology, Stockholm 2005 based on joint work with Henrik Hult www.math.kth.se/ lindskog 2 Extremes for
More informationStochastic Differential Equations.
Chapter 3 Stochastic Differential Equations. 3.1 Existence and Uniqueness. One of the ways of constructing a Diffusion process is to solve the stochastic differential equation dx(t) = σ(t, x(t)) dβ(t)
More informationEstimates for probabilities of independent events and infinite series
Estimates for probabilities of independent events and infinite series Jürgen Grahl and Shahar evo September 9, 06 arxiv:609.0894v [math.pr] 8 Sep 06 Abstract This paper deals with finite or infinite sequences
More informationCHAPTER V DUAL SPACES
CHAPTER V DUAL SPACES DEFINITION Let (X, T ) be a (real) locally convex topological vector space. By the dual space X, or (X, T ), of X we mean the set of all continuous linear functionals on X. By the
More informationLectures 16-17: Poisson Approximation. Using Lemma (2.4.3) with θ = 1 and then Lemma (2.4.4), which is valid when max m p n,m 1/2, we have
Lectures 16-17: Poisson Approximation 1. The Law of Rare Events Theorem 2.6.1: For each n 1, let X n,m, 1 m n be a collection of independent random variables with PX n,m = 1 = p n,m and PX n,m = = 1 p
More informationStochastic Proceses Poisson Process
Stochastic Proceses 2014 1. Poisson Process Giovanni Pistone Revised June 9, 2014 1 / 31 Plan 1. Poisson Process (Formal construction) 2. Wiener Process (Formal construction) 3. Infinitely divisible distributions
More informationStochastic Processes. Winter Term Paolo Di Tella Technische Universität Dresden Institut für Stochastik
Stochastic Processes Winter Term 2016-2017 Paolo Di Tella Technische Universität Dresden Institut für Stochastik Contents 1 Preliminaries 5 1.1 Uniform integrability.............................. 5 1.2
More information2 (Bonus). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?
MA 645-4A (Real Analysis), Dr. Chernov Homework assignment 1 (Due 9/5). Prove that every countable set A is measurable and µ(a) = 0. 2 (Bonus). Let A consist of points (x, y) such that either x or y is
More informationThe Lévy-Itô decomposition and the Lévy-Khintchine formula in31 themarch dual of 2014 a nuclear 1 space. / 20
The Lévy-Itô decomposition and the Lévy-Khintchine formula in the dual of a nuclear space. Christian Fonseca-Mora School of Mathematics and Statistics, University of Sheffield, UK Talk at "Stochastic Processes
More informationUniformly Uniformly-ergodic Markov chains and BSDEs
Uniformly Uniformly-ergodic Markov chains and BSDEs Samuel N. Cohen Mathematical Institute, University of Oxford (Based on joint work with Ying Hu, Robert Elliott, Lukas Szpruch) Centre Henri Lebesgue,
More informationPROBABILITY: LIMIT THEOREMS II, SPRING HOMEWORK PROBLEMS
PROBABILITY: LIMIT THEOREMS II, SPRING 15. HOMEWORK PROBLEMS PROF. YURI BAKHTIN Instructions. You are allowed to work on solutions in groups, but you are required to write up solutions on your own. Please
More informationSTOCHASTIC GEOMETRY BIOIMAGING
CENTRE FOR STOCHASTIC GEOMETRY AND ADVANCED BIOIMAGING 2018 www.csgb.dk RESEARCH REPORT Anders Rønn-Nielsen and Eva B. Vedel Jensen Central limit theorem for mean and variogram estimators in Lévy based
More informationSome basic elements of Probability Theory
Chapter I Some basic elements of Probability Theory 1 Terminology (and elementary observations Probability theory and the material covered in a basic Real Variables course have much in common. However
More informationON THE COMPLETE CONVERGENCE FOR WEIGHTED SUMS OF DEPENDENT RANDOM VARIABLES UNDER CONDITION OF WEIGHTED INTEGRABILITY
J. Korean Math. Soc. 45 (2008), No. 4, pp. 1101 1111 ON THE COMPLETE CONVERGENCE FOR WEIGHTED SUMS OF DEPENDENT RANDOM VARIABLES UNDER CONDITION OF WEIGHTED INTEGRABILITY Jong-Il Baek, Mi-Hwa Ko, and Tae-Sung
More informationLecture 4: Introduction to stochastic processes and stochastic calculus
Lecture 4: Introduction to stochastic processes and stochastic calculus Cédric Archambeau Centre for Computational Statistics and Machine Learning Department of Computer Science University College London
More informationThe Equivalence of Ergodicity and Weak Mixing for Infinitely Divisible Processes1
Journal of Theoretical Probability. Vol. 10, No. 1, 1997 The Equivalence of Ergodicity and Weak Mixing for Infinitely Divisible Processes1 Jan Rosinski2 and Tomasz Zak Received June 20, 1995: revised September
More informationProblem 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 informationµ X (A) = P ( X 1 (A) )
1 STOCHASTIC PROCESSES This appendix provides a very basic introduction to the language of probability theory and stochastic processes. We assume the reader is familiar with the general measure and integration
More informationRecurrence of Simple Random Walk on Z 2 is Dynamically Sensitive
arxiv:math/5365v [math.pr] 3 Mar 25 Recurrence of Simple Random Walk on Z 2 is Dynamically Sensitive Christopher Hoffman August 27, 28 Abstract Benjamini, Häggström, Peres and Steif [2] introduced the
More informationMOMENT CONVERGENCE RATES OF LIL FOR NEGATIVELY ASSOCIATED SEQUENCES
J. Korean Math. Soc. 47 1, No., pp. 63 75 DOI 1.4134/JKMS.1.47..63 MOMENT CONVERGENCE RATES OF LIL FOR NEGATIVELY ASSOCIATED SEQUENCES Ke-Ang Fu Li-Hua Hu Abstract. Let X n ; n 1 be a strictly stationary
More informationRegularity of the density for the stochastic heat equation
Regularity of the density for the stochastic heat equation Carl Mueller 1 Department of Mathematics University of Rochester Rochester, NY 15627 USA email: cmlr@math.rochester.edu David Nualart 2 Department
More information