Polynomial Rate Research Articles

Moment closure approximations for stochastic kinetic models with rational rate laws

Stochastic models are often used when modelling chemical species that have low numbers of molecules. However, as these models become large, it can become computationally expensive to simulate even a single realisation of the system since even efficient simulation techniques have a high computational cost. One possible technique to approximate the stochastic system is moment closure. The moment closure approximation is used to provide analytic approximations to non-linear stochastic models. Until now, this approximation has only been applied to models with polynomial rate laws. In this paper we extend the moment closure method to cover models with rational rate laws.

Arbitrary decays in linear viscoelasticity

It is by now well known that a necessary condition for the exponential (polynomial) decay of the energy of a problem arising in viscoelasticity is that the kernel (which appears in the memory term) itself be of exponential (polynomial) type. By a kernel of exponential (polynomial) type we mean that the product of this kernel with an exponential (polynomial) function is summable. Some researchers have started from this condition to seek other (sufficient) conditions ensuring exponential or polynomial decay of the energy. In this work we generalize these works to an arbitrary decay not necessarily of an exponential or polynomial rate.

Upper large deviations for Branching Processes in Random Environment with heavy tails

Branching Processes in Random Environment (BPREs) $(Z_n:n\geq0)$ are the generalization of Galton-Watson processes where 'in each generation' the reproduction law is picked randomly in an i.i.d. manner. The associated random walk of the environment has increments distributed like the logarithmic mean of the offspring distributions. This random walk plays a key role in the asymptotic behavior. In this paper, we study the upper large deviations of the BPRE $Z$ when the reproduction law may have heavy tails. More precisely, we obtain an expression for the limit of $-\log \mathbb{P}(Z_n\geq \exp(\theta n))/n$ when $n\rightarrow \infty$. It depends on the rate function of the associated random walk of the environment, the logarithmic cost of survival $\gamma:=-\lim_{n\rightarrow\infty} \log \mathbb{P}(Z_n \gt 0)/n$ and the polynomial rate of decay $\beta$ of the tail distribution of $Z_1$. This rate function can be interpreted as the optimal way to reach a given "large" value. We then compute the rate function when the reproduction law does not have heavy tails. Our results generalize the results of Böinghoff & Kersting (2009) and Bansaye & Berestycki (2008) for upper large deviations. Finally, we derive the upper large deviations for the Galton-Watson processes with heavy tails.

Finite element approximation of the Sobolev constant

Denoting by S the sharp constant in the Sobolev inequality in $${{\rm W}_0^{1,2}(B)}$$, being B the unit ball in $${\mathbb{R}^3}$$, and denoting by S h its approximation in a suitable finite element space, we show that S h converges to S as $${h\searrow0}$$with a polynomial rate of convergence. We provide both an upper and a lower bound on the rate of convergence, and present some numerical results.

On the rate of beta-mixing and convergence to a stationary distribution in continuous-time Erlang-type systems

We establish sufficient conditions for polynomial rate of convergence to a stationary distribution and of beta-mixing in continuous-time Erlang-type systems. Our results are a natural complement both to results of Erlang himself, dating back to the beginning of the 20th century, and to exponential estimates established later.

Lack of exponential decay of a coupled system of wave equations with memory

In this work, we consider a coupled system of wave equations with memory only acting in one of the equations of the system. We show that the solution of this system has a polynomial rate of decay as time tends to infinity, but does not have exponential decay.

Asymptotic Stability of the Relativistic Boltzmann Equation for the Soft Potentials

In this paper it is shown that unique solutions to the relativistic Boltzmann equation exist for all time and decay with any polynomial rate towards their steady state relativistic Maxwellian provided that the initial data starts out sufficiently close in \({L^\infty_\ell}\). If the initial data are continuous then so is the corresponding solution. We work in the case of a spatially periodic box. Conditions on the collision kernel are generic in the sense of Dudyński and Ekiel-Jeżewska (Commun Math Phys 115(4):607–629, 1985); this resolves the open question of global existence for the soft potentials.

A Kinetic Flocking Model with Diffusion

We study the stability of the equilibrium states and the rate of convergence of solutions towards them for the continuous kinetic version of the Cucker-Smale flocking in presence of diffusion whose strength depends on the density. This kinetic equation describes the collective behavior of an ensemble of organisms, animals or devices which are forced to adapt their velocities according to a certain rule implying a final configuration in which the ensemble flies at the mean velocity of the initial configuration. Our analysis takes advantage both from the fact that the global equilibrium is a Maxwellian distribution function, and, on the contrary to what happens in the Cucker-Smale model (IEEE Trans Autom Control 52:852–862, 2007), the interaction potential is an integrable function. Precise conditions which guarantee polynomial rates of convergence towards the global equilibrium are found.

Decay of Hankel singular values of analytic control systems

We show that control systems with an analytic semigroup and control and observation operators that are not too unbounded have a Hankel operator that belongs to the Schatten class S p for all positive p . This implies that the Hankel singular values converge to zero faster than any polynomial rate. This in turn implies fast convergence of balanced truncations. As a corollary, decay rates for the eigenvalues of the controllability and observability Gramians are also provided. Applications to the heat equation and a plate equation are given.

Rates of decay to weak thermoelastic Bresse system

We consider the Bresse system with temperature and we show that there exist exponential stability if and only if the wave propagation is equal. We show that, in general, the system is not exponentially stable but that there exists polynomial stability with rates that depend on the wave propagations and the regularity of the initial data. Moreover, we introduce a necessary condition to dissipative semi-group decay polynomially. This result allows us to show some optimality to the polynomial rate of decay.

Innovated higher criticism for detecting sparse signals in correlated noise

Higher criticism is a method for detecting signals that are both sparse and weak. Although first proposed in cases where the noise variables are independent, higher criticism also has reasonable performance in settings where those variables are correlated. In this paper we show that, by exploiting the nature of the correlation, performance can be improved by using a modified approach which exploits the potential advantages that correlation has to offer. Indeed, it turns out that the case of independent noise is the most difficult of all, from a statistical viewpoint, and that more accurate signal detection (for a given level of signal sparsity and strength) can be obtained when correlation is present. We characterize the advantages of correlation by showing how to incorporate them into the definition of an optimal detection boundary. The boundary has particularly attractive properties when correlation decays at a polynomial rate or the correlation matrix is Toeplitz.

Adaptive-resolution reinforcement learning with polynomial exploration in deterministic domains

We propose a model-based learning algorithm, the Adaptive-resolution Reinforcement Learning (ARL) algorithm, that aims to solve the online, continuous state space reinforcement learning problem in a deterministic domain. Our goal is to combine adaptive-resolution approximation schemes with efficient exploration in order to obtain polynomial learning rates. The proposed algorithm uses an adaptive approximation of the optimal value function using kernel-based averaging, going from coarse to fine kernel-based representation of the state space, which enables us to use finer resolution in the areas of the state space, and coarser resolution elsewhere. We consider an online learning approach, in which we discover these important areas online, using an uncertainty intervals exploration technique. In addition, we introduce an incremental variant of the ARL (IARL), which is a more practical version of the original algorithm with reduced computational complexity at each stage. Polynomial learning rates in terms of mistake bound (in a PAC framework) are established for these algorithms, under appropriate continuity assumptions.

On discrete time ergodic filters with wrong initial data, 2

For a class of non-uniformly ergodic Markov chains satisfying exponential or polynomial beta-mixing, under observations subject to an IID noise, it is shown that wrong initial data is forgotten in the mean total variation topology, with a certain exponential or polynomial rate. It is allowed that the density of the noise in the signal may vanish.

Limit theorems for monotone Markov processes

This article considers the convergence to steady states of Markov processes generated by the action of successive i.i.d. monotone maps on a subset S of an Eucledian space. Without requiring irreducibility or Harris recurrence, a “splitting” condition guarantees the existence of a unique invariant probability as well as an exponential rate of convergence to it in an appropriate metric. For a special class of Harris recurrent processes on [0,∞) of interest in economics, environmental studies and queuing theory, criteria are derived for polynomial and exponential rates of convergence to equilibrium in total variation distance. Central limit theorems follow as consequences.

A Rational Interpolation Scheme with Superpolynomial Rate of Convergence

The purpose of this study is to construct a high-order interpolation scheme for arbitrary scattered datasets. The resulting function approximation is an interpolation function when the dataset is exact, or a regression if measurement errors are present. We represent each datapoint with a Taylor series, and the approximation error as a combination of the derivatives of the target function. A weighted sum of the square of the coefficient of each derivative term in the approximation error is minimized to obtain the interpolation approximation. The resulting approximation function is a high-order rational function with no poles. When measurement errors are absent, the interpolation approximation converges to the target function faster than any polynomial rate of convergence.

Tail estimates for one-dimensional non nearest-neighbor random walk in random environment

x for 1 k g. For certain environment distributions, we determine the almost-sure asymptotic speed of the RWRE and show that the chance of the RWRE deviating below this speed has a polynomial rate of decay. This is the generalization of the results by Dembo, Peres and Zeitouni in 1996. In the proof we use a large deviation result for the product of random matrices and some tail estimates and moment estimates for the total population size in a multi-type branching process with random environment.

Model selection for quantum homodyne tomography

This paper deals with a non-parametric problem coming from physics, namely quantum tomography. That consists in determining the quantum state of a mode of light through a homodyne measurement. We apply several model selection procedures: penalized projection estimators, where we may use pattern functions or wavelets, and penalized maximum likelihood estimators. In all these cases, we get oracle inequalities. In the former we also have a polynomial rate of convergence for the non-parametric problem. We finish the paper with applications of similar ideas to the calibration of a photocounter, a measurement apparatus counting the number of photons in a beam. Here the mathematical problem reduces similarly to a non-parametric missing data problem. We again get oracle inequalities, and better speed if the photocounter is good.

Weighted [formula omitted]-estimates for dissipative wave equations with variable coefficients

We establish weighted L 2 -estimates for the wave equation with variable damping u t t − Δ u + a u t = 0 in R n , where a ( x ) ⩾ a 0 ( 1 + | x | ) − α with a 0 > 0 and α ∈ [ 0 , 1 ) . In particular, we show that the energy of solutions decays at a polynomial rate t − ( n − α ) / ( 2 − α ) − 1 if a ( x ) ∼ a 0 | x | − α for large | x | . We derive these results by strengthening significantly the multiplier method. This approach can be adapted to other hyperbolic equations with damping.

The Statistic Complexity Estimates and Its Application to Machine Learning

The concept and the estimates of statistic complex- ity are given in this paper.We show that for a function class with a polynomial rate of the VC dimension,one can find a set of linear empirical functionals of polynomial size in the dimension that are sufficient for constructing a learning algorithm which can approximate any function in the class.At the same time,a random procedure to select these functionals is given.

Non-exponential stability and decay rates in nonlinear stochastic difference equations with unbounded noise

We consider the stochastic difference equation where f and g are nonlinear, bounded functions, is a sequence of independent random variables, and h>0 is a nonrandom parameter. We establish results on asymptotic stability and instability of the trivial solution . We also show that, for some natural choices of f and g, the rate of decay of is approximately polynomial: there exists such that decays faster than but slower than , for any . It turns out that, if decays faster than as , the polynomial rate of decay can be established precisely: tends to a constant limit. On the other hand, if g does not decay quickly enough, the approximate decay rate is the best possible result.