Random Differential Equations Research Articles

Multilevel Monte Carlo estimators for elliptic PDEs with Lévy-type diffusion coefficient

General elliptic equations with spatially discontinuous diffusion coefficients may be used as a simplified model for subsurface flow in heterogeneous or fractured porous media. In such a model, data sparsity and measurement errors are often taken into account by a randomization of the diffusion coefficient of the elliptic equation which reveals the necessity of the construction of flexible, spatially discontinuous random fields. Subordinated Gaussian random fields are random functions on higher dimensional parameter domains with discontinuous sample paths and great distributional flexibility. In the present work, we consider a random elliptic partial differential equation (PDE) where the discontinuous subordinated Gaussian random fields occur in the diffusion coefficient. Problem specific multilevel Monte Carlo (MLMC) Finite Element methods are constructed to approximate the mean of the solution to the random elliptic PDE. We prove a-priori convergence of a standard MLMC estimator and a modified MLMC—control variate estimator and validate our results in various numerical examples.

Functional limit theorems for Volterra processes and applications to homogenization* *Johann Gehringer is supported by an EPSRC studentship, Xue-Mei Li by the EPRSC grants EP/V026100/1 and EP/S023925/1, and Julian Sieber by the EPSRC Centre for Doctoral Training in Mathematics of Random Systems: Analysis, Modelling and Simulation (EP/S023925/1).

We prove an enhanced limit theorem for additive functionals of a multi-dimensional Volterra process in the rough path topology. As an application, we establish weak convergence as ɛ → 0 of the solution of the random ordinary differential equation (ODE) and show that its limit solves a rough differential equation driven by a Gaussian field with a drift coming from the Lévy area correction of the limiting rough driver. Furthermore, we prove that the stochastic flows of the random ODE converge to those of the Kunita type Itô SDE dx t = G(x t , dt), where G(x, t) is a semi-martingale with spatial parameters.

Optimal control problem of various epidemic models with uncertainty based on deep reinforcement learning

AbstractWe investigate an optimal control problem of various epidemic models with uncertainty using stochastic differential equations, random differential equations, and agent‐based models. We discuss deep reinforcement learning (RL), which combines RL with deep neural networks, as one method to solve the optimal control problem. The deep Q‐network algorithm is introduced to approximate an action‐value function and consequently obtain the optimal policy. Numerical simulations show that in order to effectively prevent the spread of infectious diseases, it is essential to vaccinate at the highest rate for the first few days and then gradually reduce the rate.

On controllability for a class of multi-term time-fractional random differential equations with state-dependent delay

On controllability for a class of multi-term time-fractional random differential equations with state-dependent delay

Attractors of deterministic and random lattice difference equations

Lattice difference equations are essentially difference equations on a Hilbert space of bi-infinite sequences. They are motivated by the discretization of the spatial variable in integrodifference equations arising in theoretical ecology. It is shown here that under similar assumptions to those used for such integrodifference equations they have a global attractor, to which the global attractors of finite-dimensional approximations converge upper-semi-continuously. Corresponding results are also shown for the lattice difference equations when only a finite number of interconnection weights are nonzero and when the interconnection weights themselves vary and converge in an appropriate manner.

Uncertainty quantification for hybrid random logistic models with harvesting via density functions

The so-called logistic model with harvesting, p′(t)=rp(t)(1−p(t)K)−c(t)p(t), p(t0)=p0, is a classical ecological model that has been extensively studied and applied in the deterministic setting. It has also been studied, to some extent, in the stochastic framework using the Itô Calculus by formulating a Stochastic Differential Equation whose uncertainty is driven by the Gaussian white noise. In this paper, we present a new approach, based on the so-called theory of Random Differential Equations, that permits treating all model parameters as a random vector with an arbitrary join probability distribution (so, not just Gaussian). We take extensive advantage of the Random Variable Transformation method to probabilistically solve the full randomized version of the above logistic model with harvesting. It is done by exactly computing the first probability density function of the solution assuming that all model parameters are continuous random variables with an arbitrary join probability density function. The probabilistic solution is obtained in three relevant scenarios where the harvesting or influence function is mathematically described by discontinuous parametric stochastic processes having a biological meaning. The probabilistic analysis also includes the computation of the probability density function of the nontrivial equilibrium state, as well as the probability that stability is reached. All these results are new and extend their deterministic counterpart under very general assumptions. The theoretical findings are illustrated via two numerical examples. Finally, we show a detailed example where results are applied to describe the dynamics of stock of fishes over time using real data.

Uncertainty quantification by using Lie theory

In this short communication, the application of Lie theory in the context of forward uncertainty quantification for random ordinary differential equations is investigated. Two topics are treated. The first one concerns the automation of the Taylor series solution to an autonomous random ordinary differential equation through the Lie transformation, to derive its statistics by linearity. The second topic is about random Hamiltonian systems; when polynomial expansions of the solution are considered, the Lie formalism allows for constructing high order integrators for the Galerkin system of the expansion coefficients. These ideas are developed theoretically and assessed numerically.

Existence of solutions to a class of damped random impulsive differential equations under Dirichlet boundary value conditions

<abstract><p>In this paper, we study sufficient conditions for the existence of solutions to a class of damped random impulsive differential equations under Dirichlet boundary value conditions. By using variational method we first obtain the corresponding energy functional. Then the existence of critical points are obtained by using Mountain pass lemma and Minimax principle. Finally we assert the critical point of enery functional is the mild solution of damped random impulsive differential equations.</p></abstract>

Existence of solutions for first-order Hamiltonian random impulsive differential equations with Dirichlet boundary conditions

<p style='text-indent:20px;'>In this paper, we study the sufficient conditions for the existence of solutions of first-order Hamiltonian random impulsive differential equations under Dirichlet boundary value conditions. By using the variational method, we first obtain the corresponding energy functional. And by using Legendre transformation, we obtain the conjugation of the functional. Then the existence of critical point is obtained by mountain pass lemma. Finally, we assert that the critical point of the energy functional is the mild solution of the first order Hamiltonian random impulsive differential equation. Finally, an example is presented to illustrate the feasibility and effectiveness of our results.</p>

Nonlinear Random Differential Equations with n Sequential Fractional Derivatives

Abstract This article deals with a solvability for a problem of random fractional differential equations with n sequential derivatives and nonlocal conditions. The existence and uniqueness of solutions for the problem is obtained by using Banach contraction principle. New random data concepts for the considered problem are introduced and some related definitions are given. Also, some results related to the dependance on the introduced data are established for both random and deterministic cases.

Random neutral semilinear differential equations with delay

Random neutral semilinear differential equations with delay

A mathematical model of discrete attachment to a cellulolytic biofilm using random DEs.

We propose a new mathematical framework for the addition of stochastic attachment to biofilm models, via the use of random ordinary differential equations. We focus our approach on a spatially explicit model of cellulolytic biofilm growth and formation that comprises a PDE-ODE coupled system to describe the biomass and carbon respectively. The model equations are discretized in space using a standard finite volume method. We introduce discrete attachment events into the discretized model via an impulse function with a standard stochastic process as input. We solve our model with an implicit ODE solver. We provide basic simulations to investigate the qualitative features of our model. We then perform a grid refinement study to investigate the spatial convergence of our model. We investigate model behaviour while varying key attachment parameters. Lastly, we use our attachment model to provide evidence for a stable travelling wave solution to the original PDE-ODE coupled system.

Probabilistic analysis of a class of impulsive linear random differential equations forced by stochastic processes admitting Karhunen-Loève expansions

<p style='text-indent:20px;'>We study a full randomization of the complete linear differential equation subject to an infinite train of Dirac's delta functions applied at different time instants. The initial condition and coefficients of the differential equation are assumed to be absolutely continuous random variables, while the external or forcing term is a stochastic process. We first approximate the forcing term using the Karhunen-Loève expansion, and then we take advantage of the Random Variable Transformation method to construct a formal approximation of the first probability density function (1-p.d.f.) of the solution. By imposing mild conditions on the model parameters, we prove the convergence of the aforementioned approximation to the exact 1-p.d.f. of the solution. All the theoretical findings are illustrated by means of two examples, where different types of probability distributions are assumed to model parameters.</p>

Solving random fractional second-order linear equations via the mean square Laplace transform: Theory and statistical computing

This paper deals with random fractional differential equations of the form, CD0+αX(t)+AX˙(t)+BX(t)=0, t>0, with initial conditions, X(0)=C0 and X˙(0)=C1, where CD0+αX(t) stands for the Caputo fractional derivative of X(t). We consider the case that the fractional differentiation order is 1<α<2. For the sake of generality, we further assume that C0, C1, A and B are random variables satisfying certain mild hypotheses. Then, we first construct a solution stochastic process, via a generalized power series, which is mean square convergent for all t>0. Secondly, we provide explicit approximations of the expectation and variance functions of the solution. To complete the random analysis and from this latter key information, we take advantage of the Principle of Maximum Entropy to calculate approximations of the first probability density function of the solution. All the theoretical findings are illustrated via numerical experiments.

Some Characterization Results of Nonlocal Special Random Impulsive Evolution Differential Equations

In this paper, we present existence and uniqueness of special random impulsive differential evolution equations with nonlocal condition in Hilbert spaces. Moreover we study the stability results for the same evolution equations. Existence and uniqueness results are proved using Banach fixed point theorem where as stability results using fixed point approach and semi group theory. Finally we give some applications of the nonlocal impulsive differential equations as well as evolution equations, which shows the importance of our theoretical results.

The existence of solutions for Sturm\u2013Liouville differential equation with random impulses and boundary value problems

In this article, we consider the existence of solutions to the Sturm–Liouville differential equation with random impulses and boundary value problems. We first study the Green function of the Sturm–Liouville differential equation with random impulses. Then, we get the equivalent integral equation of the random impulsive differential equation. Based on this integral equation, we use Dhage’s fixed point theorem to prove the existence of solutions to the equation, and the theorem is extended to the general second order nonlinear random impulsive differential equations. Then we use the upper and lower solution method to give a monotonic iterative sequence of the generalized random impulsive Sturm–Liouville differential equations and prove that it is convergent. Finally, we give two concrete examples to verify the correctness of the results.

Asymptotics of Reinforcement Learning with Neural Networks

We prove that a single-layer neural network trained with the Q-learning algorithm converges in distribution to a random ordinary differential equation as the size of the model and the number of training steps become large. Analysis of the limit differential equation shows that it has a unique stationary solution that is the solution of the Bellman equation, thus giving the optimal control for the problem. In addition, we study the convergence of the limit differential equation to the stationary solution. As a by-product of our analysis, we obtain the limiting behavior of single-layer neural networks when trained on independent and identically distributed data with stochastic gradient descent under the widely used Xavier initialization.

Autonomous learning of nonlocal stochastic neuron dynamics.

Neuronal dynamics is driven by externally imposed or internally generated random excitations/noise, and is often described by systems of random or stochastic ordinary differential equations. Such systems admit a distribution of solutions, which is (partially) characterized by the single-time joint probability density function (PDF) of system states. It can be used to calculate such information-theoretic quantities as the mutual information between the stochastic stimulus and various internal states of the neuron (e.g., membrane potential), as well as various spiking statistics. When random excitations are modeled as Gaussian white noise, the joint PDF of neuron states satisfies exactly a Fokker-Planck equation. However, most biologically plausible noise sources are correlated (colored). In this case, the resulting PDF equations require a closure approximation. We propose two methods for closing such equations: a modified nonlocal large-eddy-diffusivity closure and a data-driven closure relying on sparse regression to learn relevant features. The closures are tested for the stochastic non-spiking leaky integrate-and-fire and FitzHugh-Nagumo (FHN) neurons driven by sine-Wiener noise. Mutual information and total correlation between the random stimulus and the internal states of the neuron are calculated for the FHN neuron.

Existence of periodic solutions of second-order nonlinear random impulsive differential equations via topological degree theory

In this paper, we investigate the existence of periodic solutions for a class of second order nonlinear random impulse differential equations. By extending the definitions of continuous function bound set, curvature bound set and Nagumo set in topological degree to PC space, and defining the appropriate operators and conditions, using the theory of topological degree and coincidence degree, we prove that the solution must be bounded if it exists and find the boundary value. Finally, we define appropriate Nagumo set and autonomous curvature bound set, and obtain the existence of periodic solutions of equations and simultaneous equations by using Mawhin’s continuity theorem.

Coupled fractional differential systems with random effects in Banach spaces

Abstract The purpose of this work is to investigate the existence of solutions for a system of random differential equations involving the Riemann–Liouville fractional derivative. The existence result is established by means of a random abstract formulation to Sadovskii’s fixed point theorem principle [A. Baliki, J. J. Nieto, A. Ouahab and M. L. Sinacer, Random semilinear system of differential equations with impulses, Fixed Point Theory Appl. 2017 2017, Paper No. 27] combined with a technique based on vector-valued metrics and convergent to zero matrices. An example is also provided to illustrate our result.