SearcharxivSearch

arXiv subjects

Yuri Kifer

Publications and source records attributed to Yuri Kifer.

At least 19 recordsLinked to original sources

Large Deviations for Iterated Sums and Integrals

We describe large deviations for normalized multiple iterated sums and integrals of the form $\bbS_N^{(\nu)}(t)=N^{-\nu}\sum_{0\leq k_1<...<k_\nu\leq Nt}\xi(k_1)\otimes\cdots\otimes\xi(k_\nu)$, $t\in[0,T]$ and $\bbS_N^{(\nu)}(t)=N^{-\nu}\int_{0\leq s_1\leq...\leq s_\nu\leq Nt}\xi(s_1)\otimes\cdots\otimes\xi(s_\nu)ds_1\cdots ds_\nu$, where $\{\xi(k)\}_{-\infty<k<\infty}$ and $\{\xi(s)\}_{-\infty<s<\infty}$ are centered bounded stationary vector processes whose sums or integrals satisfy a trajectorial large deviations principle.

math.PR

Iterated Ergodic Theorems and Erd\" os--R\' enyi law of large numbers

We obtain ergodic theorems for multiple iterated sums and integrals of the form $\Sigma^{(\nu)}(t)=\sum_{0\leq k_1<...<k_\nu\leq t}\xi(k_1)\otimes\cdots\otimes\xi(k_\nu)$, $t\in[0,T]$ and $\Sigma^{(\nu)}(t)=\int_{0\leq s_1\leq...\leq s_\nu\leq t}\xi(s_1)\otimes\cdots\otimes\xi(s_\nu)ds_1\cdots ds_\nu$ where $\{\xi(k)\}_{-\infty<k<\infty}$ and $\{\xi(s)\}_{-\infty<s<\infty}$ are vector processes for which standard ergodic theorems, i.e. when $\nu=1$, hold true. At the end we prove also a version of the Erd\" os--R\" enyi law of large numbers for iterated sums and integrals.

math.PR

Some strong limit theorems in averaging

The paper deals with the fast-slow motions setups in the discrete time $X^ε((n+1)ε)=X^ε(nε)+εB(X^ε(nε),ξ(n))$, $n=0,1,...,[T/ε]$ and the continuous time $\frac {dX^ε(t)}{dt}=B(X^ε(t),ξ(t/ε)).\, t\in [0,T]$ where $B$ is a smooth vector function and $ξ$ is a sufficiently fast mixing stationary stochastic process. It is known since 1966 (Khasminskii) that if $\bar X$ is the averaged motion then $G^ε=ε^{-1/2}(X^ε-\bar X)$ weakly converges to a Gaussian process $G$. We will show that for each $ε$ the processes $ξ$ and $G$ can be redefined on a sufficiently rich probability space without changing their distributions so that $E\sup_{0\leq t\leq T}|G^ε(t)-G(t)|^{2M} =O(ε^δ)$, $δ>0$ which gives also $O(ε^{δ/3})$ Prokhorov distance estimate between the distributions of $G^ε$ and $G$. In the product case $B(x,ξ)=Σ(x)ξ$ we obtain almost sure convergence estimates of the form $\sup_{0\leq t\leq T}|G^ε(t)-G(t)|=O(ε^δ)$ a.s., as well as the functional form of the law of iterated logarithm for $G^ε$. We note that our mixing assumptions are adapted to fast motions generated by important classes of dynamical systems.

math.PR

Almost sure diffusion approximation in averaging via rough paths theory

The paper deals with the fast-slow motions setups in the continuous time $\frac {dX^\ve(t)}{dt}=\frac 1\ve\sig(X^\ve(t))ξ(t/\ve^2)+b(X^\ve(t)),\, t\in [0,T]$ and the discrete time $X_N((n+1)/N)=X_N(n/N)+N^{-1/2}\sig(X_N(n/N))ξ(n))+N^{-1}b(X_N(n/N))ξ(n)$, $n=0,1,...,[TN]$ where $\sig$ and $b$ are smooth matrix and vector functions, respectively, $ξ$ is a centered stationary vector stochastic process and $\ve, 1/N$ are small parameters. We derive, first, estimates in the strong invariance principles for sums $S_{N}(t)=N^{-1/2}\sum_{0\leq k< [Nt]}ξ(k)$ and iterated sums $\bbS^{ij}_{N}(t)=N^{-1}\sum_{0\leq k<l<[Nt]}ξ_i(k)ξ_j(l)$ together with the corresponding results for integrals in the continuous time case which, in fact, yields almost sure invariance principles for iterated sums and integrals of any order and, moreover, implies laws of iterated logarithm for these objects. Then, relying on the rough paths theory, we obtain strong almost sure approximations of processes $X^\ve$ and $X_N$ by corresponding diffusion processes $Ξ^\ve$ and $Ξ_N$, respectively. Previous results for the above setup dealt either with weak or moment diffusion approximations and not with almost sure approximation which is the new and natural generalization of well known works on strong invariance principles for sums of weakly dependent random variables.

math.PR

Almost Sure Diffusion Approximation in Averaging: Direct Proofs with Rough Paths Flavors

We consider again the fast-slow motions setups in the continuous time $\frac {dX_N(t)}{dt}=N^{1/2} \sig(X_N(t))(\xi(tN))+b(X_N(t)),\, t\in [0,T]$ and the discrete time $X_N((n+1)/N)=X_N(n/N)+N^{-1/2}\sig(X_N(n/N))\xi(n)+N^{-1}b(X_N(n/N)),\, n=0,1,...,[TN]$ where $\sig$ and $b$ are smooth matrix and vector functions, respectively, $\xi$ is a centered vector stationary stochastic process with weak dependence in time and $N$ is a big parameter. We obtain estimates for the almost sure approximations of the process $X_N$ by certain diffusion process $\Sig$. In \cite{FK} and in other recent papers concerning similar setups the results were obtained relying fully on the rough paths theory. Here we derive our probabilistic results as corollaries of quite general deterministic estimates which are obtained with all details provided following somewhat ideology of the rough paths theory but not relying on this theory per se which should allow a more general readership to follow complete arguments.

math.PR

Almost sure approximations and laws of iterated logarithm for signatures

We obtain strong invariance principles for normalized multiple iterated sums and integrals of the form $\bbS_N^{(\nu)}(t)=N^{-\nu/2}\sum_{0\leq k_1<...<k_\nu\leq Nt}\xi(k_1)\otimes\cdots\otimes\xi(k_\nu)$, $t\in[0,T]$ and $\bbS_N^{(\nu)}(t)=N^{-\nu/2}\int_{0\leq s_1\leq...\leq s_\nu\leq Nt}\xi(s_1)\otimes\cdots\otimes\xi(s_\nu)ds_1\cdots ds_\nu$, where $\{\xi(k)\}_{-\infty<k<\infty}$ and $\{\xi(s)\}_{-\infty<s<\infty}$ are centered stationary vector processes with some weak dependence properties. These imply also laws of iterated logarithm and an almost sure central limit theorem for such objects. In the continuous time we work both under direct weak dependence assumptions and also within the suspension setup which is more appropriate for applications in dynamical systems. Similar results under substantially more restricted conditions were obtained in \cite{FK} relying heavily on rough paths theory and notations while here we obtain these results in a more direct way which makes them accessible to a wider readership. This is a companion paper of our paper "Limit theorems for signatures" and we consider a similar setup and rely on many result from there.

math.PR

Limit theorems for signatures

We obtain strong moment invariance principles for normalized multiple iterated sums and integrals of the form $\mathbb{S}^{(\nu)}(t)=N^{-\nu/2}\sum_{0\leq k_1<... 0$ in the Prokhorov and the Wasserstein metrics to the distribution of certain stochastic processes $\mathbb{W}_N^{(\nu)}$ constructed recursively starting from $W_N=\mathbb{W}_N^{(1)}$ which is a Brownian motion with covariances. This is done by constructing a coupling between $\mathbb{S}_N^{(1)}$ and $\mathbb{W}_N^{(1)}$, estimating directly the moment variational norm of $\mathbb{S}_N^{(\nu)}-\mathbb{W}_N^{(\nu)}$ for $\nu=1,2$ and extending these estimates to $\nu>2$ relying partially on arguments borrowed from the rough paths theory. In the continuous time we work both under direct weak dependence assumptions and also within the suspension setup which is more appropriate for applications in dynamical systems.

math.PR

Strong diffusion approximation in averaging and value computation in Dynkin's games

It is known that the slow motion $X^\varepsilon$ in the time-scaled multidimensional averaging setup $\frac {dX^\varepsilon(t)}{dt}=\frac 1\varepsilon B(X^\varepsilon(t),\,ξ(t/\varepsilon^2))+b(X^\varepsilon(t),\,ξ(t/\ve^2)),\, t\in [0,T]$ converges weakly as $\varepsilon\to 0$ to a diffusion process provided $EB(x,ξ(s))\equiv 0$ where $ξ$ is a sufficiently fast mixing stochastic process. In this paper we show that both $X^\varepsilon$ and a family of diffusions $Ξ^\varepsilon$ can be redefined on a common sufficiently rich probability space so that $E\sup_{0\leq t\leq T}|X^\varepsilon(t)-Ξ^\varepsilon(t)|^{2M}\leq C(M)\varepsilon^\del$ for some $C(M),δ>0$ and all $M\ge 1,\,\varepsilon>0$, where all $Ξ^\varepsilon,\, \varepsilon>0$ have the same diffusion coefficients but underlying Brownian motions may change with $\varepsilon$. This is the first strong approximation result both in the above setup and at all when the limit is a nontrivial multidimensional diffusion. We obtain also a similar result for the corresponding discrete time averaging setup which was not considered before at all. As an application we consider Dynkin's games with path dependent payoffs involving a diffusion and obtain error estimates for computation of values of such games by means of such discrete time approximations which provides a more effective computational tool than the standard discretization of the diffusion itself.

math.PR

Strong diffusion approximation in averaging with dynamical systems fast motion

The paper deals with the fast-slow motions setups in the continuous time $\frac {dX^(t)}{dt}=\frac 1\varepsilon B(X^\varepsilon(t),ξ(t/\varepsilon^2))+b(X^\varepsilon(t),\,ξ(t/\varepsilon^2)),\, t\in [0,T]$ and the discrete time $X^\varepsilon((n+1)\varepsilon^2)=X^\varepsilon(n\varepsilon^2)+\varepsilon B(X^\varepsilon(n\varepsilon^2),ξ(n)) +\varepsilon^2 b(X^\varepsilon(n\varepsilon^2),ξ(n))$, $n=0,1,...,[T/\varepsilon^2]$ where $Σ$ and $b$ are smooth vector functions and $ξ$ is a stationary vector stochastic process such that $Eξ(0)=0$ for all $x\in\mathbb{R}^d$. Unlike \cite{Ki20} the assumptions imposed on the process $ξ$ allow applications to a wide class of observables $g$ in the dynamical systems setup so that $ξ$ can be taken in the form $ξ(t)=g(F^tξ(0))$ or $ξ(n)=g(F^nξ(0))$ where $F$ is either a flow or a diffeomorphism with some hyperbolicity and $g$ is a vector function. In this paper we show that both $X^\varepsilon$ and a family of diffusions $Ξ^\varepsilon$ can be redefined on a common sufficiently rich probability space so that $E\sup_{0\leq t\leq T}|X^\varepsilon(t)-Ξ^\varepsilon(t)|^{p}\leq C\varepsilon^δ,\, p\geq 1$ for some $C,δ>0$ and all $\varepsilon>0$, where all $Ξ^\varepsilon,\, \varepsilon>0$ have the same diffusion coefficients but underlying Brownian motions may change with $\varepsilon$.

math.PR

Error estimates for discrete approximations of game options with multivariate diffusion asset prices

We obtain error estimates for strong approximations of a diffusion with a diffusion matrix $σ$ and a drift b by the discrete time process defined recursively X_N((n+1)/N) = X_N(n/N)+N^{1/2}σ(X_N(n/N))ξ(n+1)+N^{-1}b(XN(n/N)); where ξ(n); n\geq 1 are i.i.d. random vectors, and apply this in order to approximate the fair price of a game option with a diffusion asset price evolution by values of Dynkin's games with payoffs based on the above discrete time processes. This provides an effective tool for computations of fair prices of game options with path dependent payoffs in a multi asset market with diffusion evolution.

math.PR

The strong Borel--Cantelli property in conventional and nonconventional setups

We study the strong Borel-Cantelli property both for events and for shifts on sequence spaces considering both a conventional and a nonconventional setups. Namely, under certain conditions on events $Γ_1,Γ_2,...$ we show that with probability one \[ \left(\sum_{n=1}^N\prod_{i=1}^\ell P(Γ_{q_i(n)})\right)^{-1}\sum_{n=1}^N\prod_{i=1}^\ell\mathbb{I}_{Γ_{q_i(n)}}\to 1\,\,\mbox{as}\,\, N\to\infty \] where $q_i(n),\, i=1,...,\ell$ are integer valued functions satisfying certain assumptions and $\mathbb{I}_Γ$ denotes the indicator of $Γ$. When $\ell=1$ (called the conventional setup) this convergence can be established under $ϕ$-mixing conditions while when $\ell>1$ (called a nonconventional setup) the stronger $ψ$-mixing condition is required. These results are extended to shifts $T$ of sequence spaces where $Γ_{q_i(n)}$ is replaced by $T^{-q_i(n)}C_n^{(i)}$ where $C_n^{(i)},\, i=1,...,\ell,\, n\geq 1$ is a sequence of cylinder sets. As an application we study the asymptotical behavior of maximums of certain logarithmic distance functions and of (multiple) hitting times of shrinking cylinders.

math.PR

Limit theorems for numbers of returns in arrays under $ϕ$-mixing

We consider a $ϕ$-mixing shift $T$ on a sequence space $\Om$ and study the number $\cN_N$ of returns $\{ T^{q_N(n)}\om\in A^a_n\}$ at times $q_N(n)$ to a cylinder $A^a_n$ constructed by a sequence $a\in\Om$ where $n$ runs either until a fixed integer $N$ or until a time $τ_N$ of the first return $\{ T^{q_N(n)}\om\in A^b_m\}$ to another cylinder $A^b_m$ constructed by $b\in\Om$. Here $q_N(n)$ are certain functions of $n$ taking on nonnegative integer values when $n$ runs from 0 to $N$ and the dependence on $N$ is the main generalization here which requires certain conditions under which we obtain Poisson distributions limits of $\cN_N$ when counting is until $N$ as $N\to\infty$ and geometric distributions limits when counting is until $τ_N$ as $N\to\infty$.

math.DS

Limit theorems for numbers of multiple returns in nonconventional arrays

For a $ψ$-mixing process $ξ_0,ξ_1,ξ_2,...$ we consider the number $\mathcal{N}_N$ of multiple returns $\{ξ_{q_{i,N}(n)}\inΓ_N,\, i=1,...,\ell\}$ to a set $Γ_N$ for $n$ until either a fixed number $N$ or until the moment $τ_N$ when another multiple return $\{ξ_{q_{i,N}(n)}\inΔ_N,\, i=1,...,\ell\}$ takes place for the first time where $Γ_N\capΔ_N=\emptyset$ and $q_{i,N},\, i=1,...,\ell$ are certain functions of $n$ taking on nonnegative integer values when $n$ runs from 0 to $N$. The dependence of $q_{i,N}(n)$'s on both $n$ and $N$ is the main novelty of the paper. Under some restrictions on the functions $q_{i,N}$ we obtain Poisson distributions limits of $\mathcal{N}_N$ when counting is until $N$ as $N\to\infty$ and geometric distributions limits when counting is until $τ_N$ as $N\to\infty$. We obtain also similar results in the dynamical systems setup considering a $ψ$-mixing shift $T$ on a sequence space $Ω$ and studying the number of multiple returns $\{ T^{q_{i,N}(n)}ω\in A^a_n,\, i=1,...,\ell\}$ until the first occurrence of another multiple return $\{ T^{q_{i,N}(n)}ω\in A^b_m,\, i=1,...,\ell\}$ where $A^a_n,\, A_m^b$ are cylinder sets of length $n$ and $m$ constructed by sequences $a,b\inΩ$, respectively, and chosen so that their probabilities have the same order.

math.DS

Geometric law for numbers of returns until a hazard under $ϕ$-mixing

We consider a $ϕ$-mixing shift $T$ on a sequence space $Ω$ and study the number of returns $\{ T^kω\in U\}$ to a union $U$ of cylinders of length $n$ until the first return $\{ T^kω\in V\}$ to another union $V$ of cylinder sets of length $m$. It turns out that if probabilities of the sets $U$ and $V$ are small and of the same order then the above number of returns has approximately geometric distribution. Under appropriate conditions, we extend this result for some dynamical systems to geometric balls and Young towers with integrable tails. This work is motivated by a number of papers on asymptotical behavior of numbers of returns to shrinking sets, as well as by the papers on open systems studying their behavior until an exit through a "hole".

math.DS

Nonconventional Random Matrix Products

Let $ξ_1,ξ_2,...$ be independent identically distributed random variables and $F:\bbR^\ell\to SL_d(\bbR)$ be a Borel measurable matrix-valued function. Set $X_n=F(ξ_{q_1(n)},ξ_{q_2(n)},...,ξ_{q_\ell(n)})$ where $0\leq q_1<q_2<...<q_\ell$ are increasing functions taking on integer values on integers. We study the asymptotic behavior as $N\to\infty$ of the singular values of the random matrix product $Π_N=X_N\cdots X_2X_1$ and show, in particular, that (under certain conditions) $\frac 1N\log\|Π_N\|$ converges with probability one as $N\to\infty$. We also obtain similar results for such products when $ξ_i$ form a Markov chain. The essential difference from the usual setting appears since the sequence $(X_n)$ is long-range dependent and nonstationary.

math.PR

Geometric Law for Multiple Returns until a Hazard

For a $ψ$-mixing stationary process $ξ_0,ξ_1,ξ_2,...$ we consider the number $\mathcal N_N$ of multiple recurrencies $\{ξ_{q_i(n)}\inΓ_N,\, i=1,...,\ell\}$ to a set $Γ_N$ for $n$ until the moment $τ_N$ (which we call a hazard) when another multiple recurrence $\{ξ_{q_i(n)}\inΔ_N,\, i=1,...,\ell\}$ takes place for the first time where $Γ_N\capΔ_N= \emptyset$ and $q_i(n)<q_{i+1}(n),\, i=1,...,\ell$ are nonnegative increasing functions taking on integer values on integers. It turns out that if $P\{ξ_0\inΓ_N\}$ and $P\{ξ_0\inΔ_N\}$ decay in $N$ with the same speed then $\mathcal N_N$ converges weakly to a geometrically distributed random variable. We obtain also a similar result in the dynamical systems setup considering a $ψ$-mixing shift $T$ on a sequence space $Ω$ and study the number of multiple recurrencies $\{ T^{q_i(n)}ω\in A_n^b,\, i=1,...,\ell\}$ until the first occurence of another multiple recurrence $\{ T^{q_i(n)}ω\in A_m^a,\, i=1,...,\ell\}$ where $A_m^a,\, A_n^b$ are cylinder sets of length $m$ and $n$ constructed by sequences $a,b\inΩ$, respectively, and chosen so that their probabilities have the same order. This work is motivated by a number of papers on asymptotics of numbers of single and multiple returns to shrinking sets, as well as by the papers on open systems studying their behavior until an exit through a "hole".

math.PR