Fact-checked by Grok 2 weeks ago

Stochastic partial differential equation

A stochastic partial differential equation (SPDE) is a partial differential equation that includes a stochastic process, typically representing random noise or fluctuations, to model the evolution of systems influenced by uncertainty in both space and time.^[1] These equations generalize classical deterministic partial differential equations (PDEs) by incorporating terms such as white noise or Wiener processes, often formulated in infinite-dimensional spaces like Hilbert or Banach spaces.^[2] A canonical example is the stochastic heat equation \frac{\partial u}{\partial t} = \Delta u + f(u) + \dot{W}(t,x), where \Delta is the Laplacian, f is a nonlinear function, and \dot{W} denotes space-time white noise with covariance \mathbb{E}[\dot{W}(t,x)\dot{W}(s,y)] = \delta(t-s)\delta(x-y).^[3] The theory of SPDEs addresses significant mathematical challenges, including the definition of solutions in non-classical senses due to the irregularity introduced by noise, which often prevents pointwise evaluation and requires frameworks like mild or weak solutions via stochastic convolution and semigroup theory.^[4] Well-posedness, encompassing existence, uniqueness, and continuous dependence on initial data, typically holds in one spatial dimension for additive space-time white noise but necessitates regularization, such as colored noise or renormalization techniques like those in regularity structures, in higher dimensions.^[3] Key developments trace back to foundational works in the 1970s and 1980s, with rigorous treatments emerging through martingale measures and random field approaches, enabling analysis of linear and semilinear cases in infinite dimensions.^[1] SPDEs find broad applications across scientific disciplines, modeling phenomena where deterministic PDEs fall short in capturing environmental or internal randomness, such as fluctuating interfaces in materials science, turbulent fluid flows via stochastic Navier-Stokes equations, and population dynamics in biology influenced by random events.^[4] In physics, they describe quantum field theories and polymer chains under thermal noise, while in finance and economics, they underpin models of asset prices with spatial correlations or risk processes in networks.^[5] Numerical methods, including finite difference schemes and machine learning-based approximations, have advanced their simulation, particularly for complex systems in telecommunications and neuroscience.^[6]

Introduction

Definition and Scope

A stochastic partial differential equation (SPDE) is a partial differential equation in which one or more terms involve stochastic processes, typically representing random forcing or perturbations that introduce uncertainty into the system's evolution. In its general form, an SPDE can be expressed as

\frac{\partial u}{\partial t} = L u + \eta,

where u(t,x) is the solution depending on time t and spatial variables x, L is a deterministic spatial differential operator (such as the Laplacian), and \eta denotes a stochastic noise term modeled as a random field. This formulation extends classical partial differential equations by incorporating probabilistic elements, requiring tools from stochastic analysis like Itô calculus for well-posedness and solution theory. SPDEs arise as natural models for real-world phenomena exhibiting inherent randomness and variability that deterministic equations cannot capture adequately. For instance, in fluid dynamics, SPDEs describe turbulent flows where unpredictable fluctuations dominate the behavior, as seen in stochastic versions of the Navier-Stokes equations.^[7] Similarly, in mathematical finance, SPDEs model spatial aspects of asset pricing and portfolio optimization under uncertainty, such as the evolution of investment performance processes influenced by random market forces.^[8] These applications highlight the role of SPDEs in bridging deterministic dynamics with stochastic influences to better represent complex, noisy systems in physics, biology, and economics. The scope of SPDEs encompasses a wide range of equation types and settings, including linear SPDEs where the noise interacts additively or linearly with the solution, and nonlinear SPDEs that feature state-dependent noise or nonlinear operators, which pose greater challenges for existence and regularity of solutions. They are studied over finite domains, such as bounded regions with boundary conditions, or infinite domains like \mathbb{R}^d, affecting the choice of function spaces and solution methods.^[2] SPDEs also include evolution equations, which describe time-dependent processes like parabolic or hyperbolic systems, as well as elliptic SPDEs that model stationary or equilibrium states under random coefficients.^[9] A key distinction within SPDEs is between additive noise, where the stochastic term is independent of the solution (e.g., \eta as external forcing), and multiplicative noise, where the noise amplitude depends on the solution itself (e.g., \sigma(u) \eta), influencing the equation's stability and the need for advanced stochastic integration techniques. SPDEs can be viewed as infinite-dimensional extensions of stochastic ordinary differential equations, generalizing finite-dimensional stochastic dynamics to spatial continua.^[2]

Historical Development

The foundations of stochastic partial differential equations (SPDEs) trace back to the 1940s and 1950s, building on Norbert Wiener's pioneering work on Brownian motion from the 1920s, which established a rigorous mathematical description of random paths. During this period, physicists began extending these stochastic ideas to partial differential equations to model random phenomena such as diffusion and turbulence, laying the groundwork for incorporating noise into deterministic PDE frameworks. In the 1960s, Kiyosi Itô's development of stochastic calculus—particularly the Itô integral and Itô formula—provided essential tools for integrating stochastic processes into PDEs, enabling the formulation of SPDEs as evolutions driven by random noise. This era saw early applications, notably in nonlinear filtering problems, where Zakai's 1969 equation represented a seminal SPDE arising from signal processing contexts. Itô's framework, extended to infinite-dimensional settings, became crucial for deriving Itô-type formulas applicable to SPDEs. The 1970s witnessed the formal emergence of SPDE theory, with key contributions including Cabaña's 1970 study of linear wave equations perturbed by white noise and works by Bensoussan-Temam (1972) and Pardoux (1972) on existence results for evolution equations in Hilbert spaces. In the 1980s, advancements in regularity theory for nonlinear SPDEs addressed challenges in existence, uniqueness, and path properties, driven by the Russian school (e.g., Rozovskii's martingale methods) and broader stochastic analysis efforts, including Daniel Stroock's developments in infinite-dimensional stochastic calculus. A landmark event was the publication of John B. Walsh's 1986 lecture notes, which introduced a probabilistic approach to SPDEs, emphasizing mild solutions and multiplicative noise.^[10] By the 1990s, SPDE theory gained prominence in physics applications, such as the Kardar-Parisi-Zhang equation for interface growth, highlighting the need for robust analytical tools. Influential works included Giuseppe Da Prato and Jerzy Zabczyk's 1992 book, which systematized the semigroup approach for stochastic evolution equations in infinite dimensions, providing a cornerstone for mild solutions and stability analysis. This period solidified SPDEs as a bridge between probability and PDEs, with Itô's stochastic calculus remaining central to derivations and estimates. In the 2000s and 2010s, the field saw major breakthroughs in analyzing singular SPDEs driven by space-time white noise in higher dimensions, including the development of regularity structures by Martin Hairer in 2014 and paracontrolled calculus, enabling rigorous solutions to equations like the KPZ equation and stochastic quantization of Φ^4_3. These advances have expanded applications in quantum field theory and statistical mechanics.^[11] As of 2025, ongoing research incorporates machine learning for numerical solutions and fractional SPDEs for anomalous diffusion models.

Mathematical Foundations

Stochastic Processes

A stochastic process is a family of random variables \{X_t\}_{t \in T} defined on a probability space (\Omega, \mathcal{F}, P), where the index set T is typically time or a spatial domain, representing the evolution of a random system.^[12] These processes model phenomena where outcomes vary probabilistically over the index set, such as particle positions or financial asset prices. A prominent class is Markov processes, where the future state depends only on the current state, not the history, formalized by the Markov property: P(X_{t+s} \in A \mid \mathcal{F}_t) = P(X_{t+s} \in A \mid X_t) for events A and \sigma-algebras \mathcal{F}_t up to time t.^[13] The Wiener process, also known as Brownian motion, is a foundational continuous-time stochastic process W = \{W_t\}_{t \geq 0} with W_0 = 0 almost surely, independent increments such that W_t - W_s \sim \mathcal{N}(0, t-s) for t > s, and continuous sample paths. Its quadratic variation satisfies \langle W \rangle_t = t, reflecting the non-differentiability of paths, where formally dW_t dW_t = dt in stochastic differentials.^[14] This process captures diffusive behavior and serves as the canonical noise source in stochastic modeling. Itô stochastic calculus extends ordinary calculus to processes driven by Wiener motion, defining the Itô integral \int_0^t f_s \, dW_s for adapted square-integrable processes f as a limit of sums with left-endpoint evaluation, yielding a martingale with mean zero and variance \mathbb{E}\left[\left(\int_0^t f_s \, dW_s\right)^2\right] = \int_0^t \mathbb{E}[f_s^2] \, ds. Itô's lemma provides the chain rule for a twice-differentiable function f(t, X_t) of an Itô process dX_t = \mu_t dt + \sigma_t dW_t:

df(t, X_t) = \left( \frac{\partial f}{\partial t} + \mu_t \frac{\partial f}{\partial x} + \frac{1}{2} \sigma_t^2 \frac{\partial^2 f}{\partial x^2} \right) dt + \sigma_t \frac{\partial f}{\partial x} dW_t,

accounting for the quadratic variation term absent in deterministic calculus.^[15] This framework enables analysis of stochastic evolution in SPDEs. A filtration \{\mathcal{F}_t\}_{t \geq 0} is an increasing family of \sigma-algebras \mathcal{F}_s \subseteq \mathcal{F}_t for s < t, representing accumulating information over time.^[16] A process X = \{X_t\} is adapted to \{\mathcal{F}_t\} if X_t is \mathcal{F}_t-measurable for each t, ensuring values at time t depend only on information up to t, which is crucial for non-anticipating dynamics in stochastic settings.^[16] Martingales are adapted processes \{M_t\} with \mathbb{E}[M_t \mid \mathcal{F}_s] = M_s for t > s, modeling fair games where expected future values equal the present. Stopping times \tau satisfy \{\tau \leq t\} \in \mathcal{F}_t, allowing optional pausing based on observed information. Doob's optional stopping theorem ensures that for bounded stopping times, \mathbb{E}[M_\tau] = \mathbb{E}[M_0], facilitating convergence arguments in SPDE solution approximations by controlling stopped processes. White noise arises formally as the derivative of the Wiener process, \dot{W}_t, serving as idealized forcing in SPDE formulations.^[17]

White Noise and Wiener Processes

In the context of stochastic partial differential equations (SPDEs), white noise serves as a fundamental driving force, formally defined as the derivative of a Wiener process W(t), denoted \dot{W}(t). This generalized process exhibits zero mean and a covariance structure given by \mathbb{E}[\dot{W}(t) \dot{W}(s)] = \delta(t - s), where \delta is the Dirac delta function, capturing its uncorrelated nature at distinct times.^[18] For spatial domains, such as the interval [0,1], spatial white noise extends this concept through the cylindrical Wiener process in the Hilbert space H = L^2([0,1]). This process is constructed as W(t) = \sum_{k=1}^\infty \beta_k(t) e_k, where \{\beta_k(t)\}_{k=1}^\infty are independent standard Brownian motions and \{e_k\}_{k=1}^\infty forms an orthonormal basis of H. The cylindrical Wiener process models space-time white noise, representing uncorrelated fluctuations across both time and space, but it does not take values in H itself due to the infinite sum diverging in the H-norm.^[18] Colored noise provides smoother alternatives to white noise, particularly when spatial correlations are present, and is formalized via the Q-Wiener process in H. Here, Q is a positive semi-definite, trace-class covariance operator on H, ensuring the process is H-valued, with covariance \mathbb{E}[ \langle W(t), h \rangle_H \langle W(s), g \rangle_H ] = \min(t,s) \langle Q h, g \rangle_H for h, g \in H. The trace-class condition \operatorname{Tr}(Q) < \infty guarantees well-defined stochastic integrals, as it bounds the series expansion W(t) = \sum_{k=1}^\infty \sqrt{\lambda_k} \beta_k(t) e_k, where \lambda_k are the eigenvalues of Q and e_k its eigenbasis.^[18] A key challenge in employing white noise (corresponding to Q = I, the identity operator) in SPDEs arises because the cylindrical Wiener process is not function-valued in H, but rather a distribution in a larger space such as the Sobolev space H^{-\epsilon} for \epsilon > 0. Consequently, SPDEs driven by white noise must be interpreted in the sense of distributions or generalized functions to ensure meaningful solutions.^[18]

Formulation

General Equation Forms

Stochastic partial differential equations (SPDEs) are formulated in an abstract framework to capture the evolution of infinite-dimensional systems under random forcing, typically on a complete probability space (\Omega, \mathcal{F}, P). The state variable u(t) evolves in a separable Hilbert space H, such as L^2(D) for a spatial domain D \subseteq \mathbb{R}^d, or more regular Sobolev spaces like H^1(D) to ensure sufficient smoothness for operator applications. The driving noise is modeled by a cylindrical Q-Wiener process W(t) in H, which formalizes space-time white noise in a rigorous sense.^[2] The standard abstract form of a nonlinear SPDE is the Itô-type stochastic evolution equation

du(t) = \bigl[ A u(t) + F\bigl(u(t)\bigr) \bigr] \, dt + G\bigl(u(t)\bigr) \, dW(t),

with initial condition u(0) = u_0 \in H, where A: D(A) \subseteq H \to H is an unbounded linear operator (e.g., the Laplacian -\Delta with appropriate domain) that generates an analytic strongly continuous semigroup \{S(t)\}_{t \geq 0} on H, F: H \to H and G: H \to L_2^0(H, U) (with U another Hilbert space for the noise range and L_2^0 the space of Hilbert-Schmidt operators) are nonlinear maps satisfying suitable Lipschitz or growth conditions, and W(t) is the Q-Wiener process with trace-class covariance operator Q. This formulation accommodates a wide range of physical models by allowing A to represent deterministic diffusion or transport, F nonlinear interactions, and G \, dW multiplicative noise.^[2] Solutions to this equation are often sought in the mild sense, which avoids direct differentiation and leverages the semigroup structure:

u(t) = S(t) u_0 + \int_0^t S(t-s) F\bigl(u(s)\bigr) \, ds + \int_0^t S(t-s) G\bigl(u(s)\bigr) \, dW(s),

where the stochastic integral is an Itô integral in the Hilbert space setting. A mild solution is a H-valued stochastic process u(\cdot) that satisfies this integral equation almost surely for each t > 0. In contrast, a strong solution satisfies the original differential equation almost everywhere with respect to the product measure on [0,T] \times \Omega, requiring higher regularity such as u \in D(A) pathwise; under conditions like sectoriality of A and Lipschitz continuity of F and G, mild and strong solutions coincide. The mild formulation is preferred for existence and uniqueness theorems, as it handles the irregularity of the noise more tractably.^[2]^[2] For the linear case, where F \equiv 0 and G(u) = B is a constant operator B \in L(U, H), the SPDE simplifies to

du(t) = A u(t) \, dt + B \, dW(t),

and the mild solution admits an explicit expression via the variation of constants formula:

u(t) = S(t) u_0 + \int_0^t S(t-s) B \, dW(s).

This stochastic convolution term captures the noise propagation through the semigroup, and its regularity (e.g., in H^\alpha for \alpha < 1/2) depends on the eigenvalues of A and Q. Linear SPDEs serve as building blocks for perturbation analysis in nonlinear settings.^[2]^[2]

Initial and Boundary Conditions

In stochastic partial differential equations (SPDEs), initial and boundary conditions specify the solution's behavior at the start of the time interval and on the domain's boundary, ensuring the problem is well-posed within appropriate function spaces. These conditions integrate with the general SPDE formulation to constrain the evolution of the random field solution. The initial condition is typically given by u(0, x) = u_0(x) for x \in D, where D \subset \mathbb{R}^d is the spatial domain and u_0 may be deterministic or random. The initial data u_0 must belong to a suitable space, such as L^2(\Omega; H), where \Omega is the probability space and H is a Hilbert space like L^2(D) or a Sobolev space H^k(D) for k \geq 0, to guarantee measurability and integrability with respect to the underlying stochastic process.^[4]^[19]^[20] Boundary conditions on \partial D commonly take Dirichlet or Neumann forms. Dirichlet conditions prescribe u(t, y) = g(t, y) for y \in \partial D and t > 0, with homogeneous cases setting g \equiv 0 and inhomogeneous cases allowing time-dependent prescriptions. Neumann conditions specify the normal derivative \frac{\partial u}{\partial n}(t, y) = h(t, y) on \partial D, where n is the outward unit normal; these can incorporate noise effects through the choice of h, influencing the solution's regularity near the boundary.^[19]^[20] Stochastic boundary conditions, though less common, arise when noise perturbs the boundary directly, such as in white-noise Dirichlet conditions u(t, 0) = \dot{W}(t) or Neumann conditions \frac{\partial u}{\partial n}(t, 0) = \dot{W}(t), where \dot{W} denotes space-time white noise; these lead to solutions in weighted Sobolev spaces like H^\alpha with \alpha < 1/4 for Neumann cases. For existence of solutions, compatibility conditions require that the initial data u_0 and boundary functions g or h align at t=0, often verified via trace theorems in Sobolev spaces, ensuring the trace operator maps H^1(D) continuously to L^2(\partial D).^[21]^[22]

Classical Examples

Stochastic Heat Equation

The stochastic heat equation serves as a foundational linear stochastic partial differential equation (SPDE), capturing the evolution of a quantity undergoing diffusion perturbed by random forcing. It arises naturally in modeling physical systems where thermal diffusion is influenced by unpredictable environmental fluctuations, such as irregular heat sources or turbulent flows. Unlike deterministic diffusion equations, the inclusion of noise introduces probabilistic behavior, leading to solutions that exhibit spatial and temporal irregularity while preserving key smoothing properties of the Laplacian operator.^[23] The equation is formulated as

\frac{\partial u}{\partial t}(t,x) = \Delta u(t,x) + \dot{B}(t,x), \quad (t,x) \in [0,T] \times D,

where D is a bounded spatial domain (typically an interval or torus for simplicity), \Delta is the Laplacian, and \dot{B}(t,x) denotes space-time white noise, formally the derivative of a cylindrical Wiener process in the spatial variable. Appropriate initial condition u(0,x) = u_0(x) and boundary conditions (e.g., Dirichlet or periodic) are imposed to ensure well-posedness. This additive noise structure distinguishes it from more complex nonlinear variants, allowing explicit analysis in Hilbert spaces like L^2(D).^[23]^[2] Under suitable conditions, such as u_0 \in L^2(D) and homogeneous Dirichlet boundaries, the mild solution can be derived using the eigenfunction expansion of the Laplacian. Let \{-\lambda_k, \phi_k\}_{k=1}^\infty be the eigenvalues and orthonormal eigenfunctions of -\Delta on D with the given boundaries, satisfying \Delta \phi_k = -\lambda_k \phi_k. The solution decomposes into a deterministic component from the initial data, S(t)u_0(x) = \sum_k \langle u_0, \phi_k \rangle e^{-\lambda_k t} \phi_k(x), where S(t) is the heat semigroup, and a stochastic integral term:

u(t,x) = S(t)u_0(x) + \sum_{k=1}^\infty \phi_k(x) \int_0^t e^{-\lambda_k (t-s)} \, d\beta_k(s),

with \{\beta_k\}_{k=1}^\infty independent standard Brownian motions arising from the projection of the white noise onto the eigenbasis. Each stochastic coefficient \int_0^t e^{-\lambda_k (t-s)} \, d\beta_k(s) is a Gaussian Ornstein-Uhlenbeck process, ensuring the full solution is a centered Gaussian random field when u_0 = 0. The series converges in suitable spaces like L^2(\Omega; C([0,T]; H^{-\epsilon}(D))) for \epsilon > 0, reflecting the noise's roughening effect against the Laplacian's smoothing.^[23]^[2] Key properties of the additive solution include finite moments E[|u(t,x)|^p] for all p \geq 1, which grow at most exponentially in t and can be explicitly bounded using the series representation. For instance, the second moment is E[|u(t,x)|^2] = \sum_k \phi_k(x)^2 \frac{1 - e^{-2\lambda_k t}}{2\lambda_k}, highlighting spatial dependence and asymptotic stationarity as t \to \infty. The Gaussian nature implies higher moments follow from the variance via standard formulas, such as E[|u(t,x)|^p] = c_p [E[|u(t,x)|^2]]^{p/2} for even p, with no intermittency—defined as disproportionate growth of higher moments relative to lower ones—since the field remains homogeneous in its probabilistic structure. In contrast, the multiplicative variant \frac{\partial u}{\partial t} = \Delta u + \sigma(u) \dot{B}(t,x), where \sigma is Lipschitz (e.g., \sigma(u) = u), admits a unique solution but can exhibit intermittency effects, manifesting as localized "islands" of high intensity where higher moments E[|u(t,x)|^p] grow superexponentially in p, signaling non-Gaussian tails and potential blow-up for superlinear \sigma.^[23]^[2] This equation interprets heat propagation in media with stochastic forcing, such as fluctuating boundary temperatures or random heat injections, providing a benchmark for SPDE theory in physics and engineering applications.^[23]

Stochastic Burgers Equation

The stochastic Burgers equation serves as a prototypical example of a nonlinear hyperbolic stochastic partial differential equation (SPDE), incorporating both convective transport and diffusive regularization perturbed by random forcing. It models the evolution of a velocity field u(t,x) subject to nonlinear advection, viscous dissipation, and stochastic fluctuations, often defined on the real line \mathbb{R} or the one-dimensional torus \mathbb{T} to capture periodic boundary conditions. The standard viscous form is given by

\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial x} = \nu \frac{\partial^2 u}{\partial x^2} + \dot{B}(t,x),

where \nu > 0 is the viscosity coefficient, and \dot{B}(t,x) denotes space-time white noise, representing the formal derivative of a cylindrical Wiener process in the spatial variable.^[24] This equation extends the deterministic Burgers equation, which approximates one-dimensional compressible fluid flows, by introducing multiplicative or additive noise to account for environmental randomness.^[25] A key analytical tool for studying the stochastic Burgers equation is the Hopf-Cole transformation, which linearizes the nonlinear problem into a stochastic heat equation, thereby facilitating existence and uniqueness proofs for mild solutions in appropriate function spaces. Specifically, setting u = -2\nu \partial_x \log \phi transforms the equation into \partial_t \phi = \nu \partial_x^2 \phi + \phi \dot{B}(t,x), mirroring the connection in the deterministic case where Burgers relates to the Navier-Stokes equations via dimensionality reduction.^[24] This transform highlights the equation's ties to stochastic Navier-Stokes models in one dimension, though the noise term introduces additional challenges in higher dimensions. However, solving the stochastic Burgers equation remains nontrivial due to the nonlinearity fostering shock formation in the inviscid limit (\nu \to 0), where solutions develop discontinuities; the additive noise can either regularize these singularities by smoothing irregularities or, in singular regimes, amplify them through enhanced intermittency.^[26]^[27] In the periodic setting on \mathbb{T}, the stochastic Burgers equation admits a unique invariant measure, ensuring the existence of a stationary solution that describes the long-time statistical equilibrium under the random forcing. This measure is ergodic and supports moment bounds, reflecting the balance between dissipative viscosity and stochastic excitation.^[28] Such properties are crucial for understanding asymptotic behavior, as the invariant measure governs the probabilistic description of velocity fluctuations in the steady state. The equation finds applications in modeling fluctuating fluid velocity profiles, such as in turbulent flows or shock wave propagation with environmental noise, providing insights into stochastic homogenization in one-dimensional hydrodynamics.^[29]^[30]

Solution Methods

Analytical Approaches

Analytical approaches to solving stochastic partial differential equations (SPDEs) primarily focus on exact or semi-exact methods that leverage operator theory and transformation techniques, particularly for linear and mildly nonlinear cases. These methods often rely on the theory of strongly continuous semigroups generated by the deterministic differential operator to express solutions in integral form, allowing the incorporation of stochastic forcing terms. Such techniques provide a framework for establishing existence, uniqueness, and regularity of mild solutions in appropriate function spaces.^[31] For linear SPDEs of the form du = Au \, dt + dW, where A is the infinitesimal generator of a C_0-semigroup \{S(t)\}_{t \geq 0} on a Hilbert space H and W is a cylindrical Wiener process, the mild solution is given by

u(t) = S(t) u_0 + \int_0^t S(t-s) \, dW(s),

with the initial condition u(0) = u_0 \in H. The first term represents the deterministic evolution, while the stochastic convolution integral captures the noise effect, which is a Gaussian process whose properties depend on the semigroup's smoothing action. This representation facilitates analysis of the solution's regularity and moment estimates, assuming the semigroup is analytic or provides sufficient decay.^[31] Extending to semilinear SPDEs, such as du = Au \, dt + F(u) \, dt + dW, fixed-point theorems in Banach spaces are employed to establish the existence of mild solutions. The mild solution satisfies the integral equation

u(t) = S(t) u_0 + \int_0^t S(t-s) F(u(s)) \, ds + \int_0^t S(t-s) \, dW(s),

and Picard iteration—starting from an initial guess and iteratively applying the operator—converges under suitable Lipschitz or growth conditions on F. Contraction mapping principles in weighted Banach spaces, often with stopping times to control explosions, yield local existence and uniqueness; global results follow if additional dissipativity assumptions hold. This approach is pivotal for nonlinear problems where the nonlinearity is treated as a perturbation of the linear semigroup dynamics.^[31] In translation-invariant settings, such as SPDEs on \mathbb{R}^d with constant coefficients and space-time white noise, Fourier transform methods diagonalize the problem. Applying the Fourier transform converts the SPDE into a family of independent stochastic differential equations (SDEs) in frequency space, where each mode evolves as an Ornstein-Uhlenbeck process driven by complex Gaussian noise. For instance, the stochastic heat equation transforms to d\hat{u}(\xi, t) = -|\xi|^2 \hat{u}(\xi, t) \, dt + d\hat{W}(\xi, t), solvable explicitly via variation of constants. This spectral decomposition simplifies computations of correlation functions and invariant measures, particularly for periodic or unbounded domains.^[31] For hyperbolic SPDEs, such as stochastic transport or wave equations, the method of stochastic characteristics adapts the classical deterministic technique by integrating along random curves defined by the stochastic flow of the principal part. The solution is expressed as u(t, x) = u_0(\phi_{-t}(x)) + \int_0^t g(\phi_{s-t}(x), s) \, ds + \int_0^t \sigma(\phi_{s-t}(x), s) \, dB_s, where \phi is the stochastic flow solving the associated SDE, and B is Brownian motion. This approach resolves first-order hyperbolic systems by reducing them to stochastic ordinary differential equations along characteristics, preserving well-posedness under non-degeneracy of the flow.^[32] These analytical methods excel for linear SPDEs and those with mild nonlinearities, where the semigroup provides smoothing and the perturbations remain controllable. However, they encounter limitations with strong singularities, superlinear growth in nonlinearities, or multiplicative noise that destroys contractivity, often requiring renormalization or alternative frameworks for global existence.^[31]

Numerical Approximation Techniques

Numerical approximation of stochastic partial differential equations (SPDEs) involves discretizing both space and time to obtain computable finite-dimensional approximations, typically transforming the infinite-dimensional problem into a system of stochastic ordinary differential equations (SODEs) that can be simulated. These methods are essential due to the lack of closed-form solutions for most SPDEs, enabling practical simulations in applications like physics and finance. Key challenges include handling the irregularity introduced by the noise term and ensuring convergence in appropriate norms, such as mean-square or weak error senses.^[33] Spatial discretization reduces the SPDE to a semidiscrete SODE by approximating the solution on a finite grid or subspace. In finite difference methods, the spatial domain is partitioned into a grid with mesh size h, approximating the differential operator A by a matrix A_h and the noise by G_h dW, yielding the semidiscrete equation du_h = A_h u_h \, dt + G_h \, dW, where u_h is the vector of grid values and W is the Wiener process. Finite element methods similarly project onto a finite-dimensional subspace V_h \subset H, often using Galerkin projection, leading to an analogous form du_h = A_h u_h \, dt + G_h \, dW with A_h as the stiffness matrix and G_h incorporating the noise covariance. These approaches preserve the structure of the original SPDE while reducing dimensionality, with finite elements offering flexibility for irregular domains.^[33]^[34] Time discretization then approximates the semidiscrete SODE using stochastic integration schemes. The Euler-Maruyama method, a standard choice, advances the solution via u_h^{n+1} = u_h^n + \Delta t \, A_h u_h^n + G_h(u_h^n) \, \Delta W^n, where \Delta t is the time step and \Delta W^n \sim \mathcal{N}(0, \Delta t). This scheme achieves strong convergence of order \sqrt{\Delta t} in the mean-square sense for Lipschitz nonlinearities and additive noise, meaning \mathbb{E}[ \|u(T) - u_h^{N} \| ^2 ]^{1/2} \leq C (\Delta t^{1/2} + h^r) for some r > 0, while weak convergence (for expectations) is typically order \Delta t. Higher-order variants, like the Milstein scheme, incorporate additional Itô correction terms for improved accuracy but increase computational cost.^[33] To estimate quantities of interest, such as expectations \mathbb{E}[\phi(u(T))], Monte Carlo simulation generates an ensemble of independent realizations of the discretized SPDE and averages them, yielding an estimator with statistical error O(M^{-1/2}) where M is the number of samples. Variance reduction techniques enhance efficiency; for instance, antithetic variates pair noise realizations -W with W to exploit symmetry and halve the variance, while multilevel Monte Carlo (MLMC) combines simulations on successively refined grids to achieve near-optimal cost for weak errors. These methods are particularly effective when combined with spatial-temporal schemes, balancing bias and statistical errors.^[33]^[34] Spectral methods provide an alternative spatial discretization, especially suited for problems with smooth solutions and periodic boundaries, by projecting onto a basis of eigenfunctions via Galerkin methods. The solution is expanded as u_h = \sum_{k=1}^{N_h} u_k \phi_k in a subspace spanned by basis functions \{\phi_k\}, leading to a semidiscrete SODE where the matrices A_h and G_h are diagonal if the operator and noise covariance share the same eigenbasis, simplifying computations. This approach often yields higher-order spatial accuracy than finite differences for the same degrees of freedom.^[33]^[34] Error analysis for these schemes quantifies the total approximation error as the sum of spatial, temporal, and statistical components. Spatial errors typically converge at order h^r (e.g., r=1 for linear finite elements or r=2 for spectral methods), while temporal strong errors are order \sqrt{\Delta t} due to the roughness of the Wiener process; weak errors can reach order \Delta t under smoothness assumptions. Optimal implementation often sets \Delta t \sim h^2 to balance terms, with overall strong convergence O(h) for the full scheme. Rigorous bounds rely on regularity theory and Itô isometry, ensuring stability under conditions like \|A_h\| \Delta t < 1.^[33]^[34]

Applications

In Physics and Engineering

Stochastic partial differential equations (SPDEs) play a crucial role in modeling physical systems subject to random perturbations, particularly in wave propagation and noisy environments. One prominent example is the stochastic wave equation, given by

\frac{\partial^2 u}{\partial t^2} = \Delta u + \dot{B},

where \dot{B} represents space-time white noise, which models the vibrations of a string or membrane under random forcing. This equation captures the dynamics of elastic media subjected to stochastic excitations, such as thermal fluctuations or external noise, leading to phenomena like energy equipartition between kinetic and potential forms in the long-time limit.^[35]^[36] In interface growth processes, the Kardar-Parisi-Zhang (KPZ) equation,

\frac{\partial h}{\partial t} = \nu \Delta h + \frac{1}{2} |\nabla h|^2 + \dot{B},

describes the evolution of a growing surface height h, incorporating diffusion, nonlinear steepening, and random noise. Originally proposed to unify scaling behaviors in nonequilibrium growth, it applies to physical contexts like thin-film deposition and crystal surfaces, where noise induces roughness characterized by universal exponents.^[37]^[38] SPDEs also underpin constructions in quantum field theory, notably the Euclidean \Phi^4_3 model, which incorporates multiplicative noise to regularize the theory on \mathbb{R}^3. This stochastic quantization approach resolves ultraviolet divergences through parabolic SPDEs, enabling rigorous definitions of correlation functions for interacting fields in three dimensions.^[39]^[40] In engineering applications, SPDEs model and control flexible structures like beams under stochastic loads, as in the Timoshenko beam equation with boundary noise, where feedback controls stabilize vibrations against random disturbances. Similarly, in signal processing, ensemble Kalman filters for SPDEs estimate states from noisy spatio-temporal observations, enhancing robustness in systems like sensor networks affected by environmental noise.^[41]^[42] Recent advancements in the 2020s have integrated SPDEs into climate modeling for turbulent flows, using stochastic closures to parameterize subgrid-scale uncertainties in primitive equations, improving predictions of ocean-atmosphere interactions and wave-current mixing. Numerical simulations validate these models by comparing against large-eddy simulations of turbulent regimes.^[43]

In Biology and Finance

Stochastic partial differential equations (SPDEs) play a crucial role in modeling biological processes where spatial structure and random environmental fluctuations interact, such as in population dynamics. A prominent example is the stochastic Fisher-Kolmogorov-Petrovsky-Piscounov (Fisher-KPP) equation, which extends the deterministic reaction-diffusion model to incorporate multiplicative or additive noise representing environmental variability, like fluctuating resource availability or demographic stochasticity. The equation takes the form

\frac{\partial u}{\partial t} = D \Delta u + r u (1 - u) + \dot{B},

where u(x,t) denotes population density at position x and time t, D > 0 is the diffusion coefficient, r > 0 is the growth rate, and \dot{B} is space-time white noise capturing random perturbations. This model describes the invasion speed of a beneficial allele or species in a spatial habitat, where noise can slow wavefront propagation compared to the deterministic case, with the asymptotic speed influenced by noise intensity. For instance, in studies of expanding populations, stochastic effects lead to pulled fronts with logarithmic corrections to the speed, enhancing realism for ecological invasions.^[44]^[45] In neuroscience, SPDEs model neural fields to account for synaptic transmission amid fluctuating inputs, representing irregular presynaptic activity or synaptic noise. Neural field equations describe the evolution of cortical activity u(x,t) across a spatial domain x, incorporating nonlocal synaptic interactions and stochastic forcing to mimic variability in neural signaling. A typical stochastic neural field equation is

\tau \frac{\partial u}{\partial t} = -u + \int w(x-y) f(u(y,t)) dy + \xi(x,t),

where \tau > 0 is a time constant, w is the synaptic weight kernel, f is a nonlinear firing rate function, and \xi is Gaussian noise modeling fluctuating inputs. This framework captures phenomena like working memory persistence or pattern formation in the presence of noise, with rigorous existence results ensuring well-posedness under mild conditions on the nonlinearity and noise. Such models reveal how noise can stabilize or destabilize stationary bumps, relevant to sensory processing and decision-making in the brain.^[46]^[47] Turning to finance, SPDEs extend classical option pricing frameworks to handle spatial randomness in multi-asset markets, where asset prices exhibit correlated fluctuations across locations or factors. Backward SPDEs provide solutions for pricing derivatives under uncertainty, applicable to multi-asset options such as baskets, capturing inter-asset dependencies and enabling valuation under non-Markovian or jump-diffusion extensions. Adapted solutions to such backward SPDEs facilitate pricing of American options with early exercise.^[48] SPDEs also enhance risk assessment in credit markets by modeling spatial correlations in default probabilities across portfolios or regions. For mortgage-backed securities, an SPDE governs the pool's value evolution, incorporating stochastic intensity for prepayments and defaults with spatial dependence. This approach captures tranche pricing flexibly, with existence and uniqueness proven for mild solutions, outperforming Gaussian copula models in fitting observed spreads during crises. In credit risk, such spatial structure reflects contagion effects, improving portfolio hedging.^[49] Recent applications in the 2020s leverage SPDEs for epidemiology, particularly stochastic susceptible-infected-recovered (SIR) models of disease spread with spatial diffusion and noise from behavioral variability. The stochastic SIR SPDE tracks compartments S(x,t), I(x,t), R(x,t) as

dS = -\beta S I dt + D_S \Delta S dt + \sigma_S S dW_S, \quad dI = (\beta S I - \gamma I) dt + D_I \Delta I dt + \sigma_I I dW_I,

with recovery dR = \gamma I dt, where \beta is transmission rate, \gamma recovery rate, D diffusion coefficients, and W Wiener processes for demographic noise. These models, extended to infection-age dependence, predict outbreak thresholds and extinction probabilities more accurately than ODEs, aiding COVID-19 simulations by incorporating spatial heterogeneity in contact rates. Finite-time stability analyses confirm noise suppresses persistence under certain regimes.^[50]

Advanced Topics

Regularity and Existence Theory

The existence of solutions to stochastic partial differential equations (SPDEs) is typically established locally in time through fixed-point theorems applied to the mild formulation in suitable Banach spaces, particularly for semilinear equations with Lipschitz or locally Lipschitz nonlinearities.^[2] For global existence, monotonicity conditions on the drift term enable the use of Galerkin approximations and compactness arguments, yielding weak solutions that satisfy energy inequalities analogous to those in deterministic theory.^[51] A prominent example is the stochastic Navier-Stokes equations in three dimensions, where global Leray-Hopf weak solutions exist for additive or trace-class multiplicative noise, satisfying a stochastic energy equality and belonging to appropriate Sobolev spaces.^[2] Pathwise uniqueness of solutions follows from Lipschitz continuity of the drift coefficient F and diffusion coefficient G with respect to the state variable, ensuring that any two solutions starting from the same initial condition and driven by the same noise coincide almost surely.^[2] This result relies on Itô's formula and Gronwall-type inequalities to control the difference between solutions in mean-square norms.^[52] Under stronger conditions, such as subcritical growth, uniqueness extends to mild solutions in Hölder or Sobolev spaces.^[53] Regularity theory for SPDEs quantifies the smoothness of solutions, often revealing lower regularity compared to deterministic counterparts due to the irregularity of the noise. For the one-dimensional stochastic heat equation driven by space-time white noise,

\frac{\partial u}{\partial t} = \frac{\partial^2 u}{\partial x^2} + \dot{W}(t,x),

the mild solution exhibits Hölder continuity almost surely, with exponents \gamma < 1/4 in time and \beta < 1/2 in space, derived from Kolmogorov-type criteria applied to the stochastic convolution term.^[54] In higher dimensions or with colored noise, regularity improves, potentially reaching Sobolev spaces W^{\alpha,p} for suitable \alpha and p > 1.^[2] Boundary conditions, such as Dirichlet or Neumann, can reduce regularity near the domain edges but preserve interior Hölder estimates under compatibility assumptions.^[55] For singular SPDEs, where the nonlinearity interacts strongly with space-time white noise in dimensions greater than one, classical methods fail due to insufficient regularity. Modern frameworks like regularity structures, introduced by Hairer in 2014, provide a renormalization-based approach to define and prove well-posedness. These structures encode the solution and noise via a combinatorial algebraic object, allowing controlled expansions and fixed-point arguments in Hölder-like Besov spaces. Key applications include the KPZ equation and \Phi^4_3 model, yielding unique mild solutions with subcritical regularity. Paracontrolled distributions offer an alternative analytic tool for similar purposes. Recent extensions as of 2025 include noncommutative regularity structures for more general settings.^[4]^[56] Blow-up criteria for SPDEs with multiplicative noise characterize finite-time explosion, where the solution norm diverges if the noise coefficient \sigma(u) exhibits superlinear growth, such as in stochastic nonlinear Schrödinger equations.^[57] Specifically, for supercritical nonlinearities, blow-up occurs with positive probability if \|\sigma(u)\|_{L^\infty} \to \infty as the solution escapes compact sets, controlled by stochastic energy methods and comparison principles.^[58] Martingale solutions offer a weak existence framework by constructing probability measures on path spaces that satisfy the SPDE in an integral sense, often via tightness and Skorokhod embedding when the original probability space is inadequate for the noise.^[59] This approach, building on Krylov-Rozovskii theory, ensures convergence of Galerkin approximations to a martingale solution satisfying monotonicity and boundedness in probability.^[51]

Infinite-Dimensional Systems

Stochastic partial differential equations (SPDEs) in infinite-dimensional settings are often formulated as abstract stochastic evolution equations in Hilbert or Banach spaces, providing a general framework for analyzing systems like the stochastic heat or wave equations on unbounded domains. A canonical form is the mild solution to the equation

du(t) = Au(t)\,dt + dW(t), \quad u(0) = u_0 \in H,

where H is a separable Hilbert space, A generates a strongly continuous semigroup \{S(t)\}_{t \geq 0} on H, and W is a cylindrical Wiener process in H with respect to another Hilbert space K, typically taking values in H via a covariance operator Q. This setup allows SPDEs to be treated as infinite-dimensional stochastic differential equations (SDEs), where the partial differential operator is encoded in A, and the noise is modeled by the infinite-dimensional Wiener process.^[2] In the linear case, where the equation simplifies without nonlinear terms, the solution is the Ornstein-Uhlenbeck process, a Gaussian process whose stationary distribution exists under suitable dissipativity assumptions on A. The covariance operator Q of this stationary measure satisfies the Lyapunov equation AQ + QA^* + BB^* = 0, where B embeds the noise space into H, ensuring the process reaches a unique invariant measure in L^2(H). This framework extends finite-dimensional Ornstein-Uhlenbeck processes to infinite dimensions, facilitating the study of long-time behavior and ergodicity for linear SPDEs.^[2] For nonlinear extensions, such as du(t) = [Au(t) + F(u(t))]\,dt + dW(t), the Girsanov transform provides a powerful tool to change the underlying probability measure, equivalent to adding a drift term B(u(t)) to the noise. In infinite dimensions, this requires the Novikov condition on the exponential martingale, adapted via predictable processes in H, to ensure absolute continuity between measures and enable equivalence of solutions under different drifts. This technique is crucial for proving existence and uniqueness in nonlinear settings by reducing to the linear case.^[2] SPDEs can also be defined on non-Euclidean structures like manifolds or graphs, where the infinite-dimensional system arises as a scaling limit of discrete interacting particle models. On graphs, for instance, the equation takes the form of a stochastic evolution driven by the graph Laplacian, converging from finite particle systems to an infinite-dimensional limit as the graph scales, preserving spatial heterogeneity. Similarly, on manifolds, the framework adapts the semigroup generated by the Laplace-Beltrami operator, treating the SPDE as an infinite-dimensional SDE on the tangent bundle.^[60]^[61] A key challenge in these infinite-dimensional systems is ensuring the noise is well-defined, requiring the covariance operator Q to be nuclear (trace-class) when A is unbounded, as cylindrical Wiener processes in H must be regularized to avoid ill-posedness. This nuclearity condition guarantees the trace \operatorname{Tr}(Q) < \infty, allowing the Itô integral to exist in H, but imposes restrictions on the spatial correlation of the noise for unbounded operators like the Laplacian.^[2]

References

[1]
[PDF] An introduction to stochastic partial differential equations
definition of the stochastic integral may look unfamiliar at first, but we are merely following Ito's construction in a different setting. In the classical ...
[2]
Stochastic Equations in Infinite Dimensions
Giuseppe Da Prato, Scuola Normale Superiore, Pisa, Jerzy Zabczyk, Polish Academy of Sciences. Publisher: Cambridge University Press.
[3]
[PDF] A short introduction to Stochastic PDEs - HAL
Apr 4, 2014 · Abstract. The main two aims of these lecture notes are: a definition of the space-time white noise and the study of second-order stochastic ...Missing: scholarly | Show results with:scholarly
[4]
[PDF] An Introduction to Stochastic PDEs - of Martin Hairer
Mar 19, 2025 · It is an attempt to give a reasonably self-contained presentation of the basic theory of stochastic partial differential equations, taking for.Missing: scholarly | Show results with:scholarly
[5]
Stochastic Partial Differential Equation - ScienceDirect.com
Stochastic differential equations (SDEs) appear today as a modeling tool in several sciences as telecommunications, economics, finance, biology, and quantum ...
[6]
A Primer on Stochastic Partial Differential Equations with Spatially ...
Nov 11, 2024 · In this review, I seek to acquaint the reader with SPDEs from the point of view of numerically simulating their finite-difference approximations.
[7]
[PDF] Stochastic PDEs in Turbulence Theory - Math (Princeton)
Mar 11, 2000 · This paper reviews the recent progress on stochastic PDEs arising from different as- pects of the turbulence theory including the stochastic ...
[8]
[PDF] Stochastic partial differential equations and portfolio choice - UT Math
Abstract. We introduce a stochastic partial differential equation which describes the evolution of the investment performance process in portfolio choice.
[9]
Galerkin methods for linear and nonlinear elliptic stochastic partial ...
In Section 5 we describe stochastic Galerkin methods for general nonlinear elliptic boundary value problems, starting as before with the mathematical setting.
[10]
https://doi.org/10.1007/BFb0074920
[11]
[PDF] Lecture 5 : Stochastic Processes I - MIT OpenCourseWare
A stochastic process is a collection of random variables indexed by time. An alternate view is that it is a probability distribution over a space.
[12]
[PDF] Markov Processes and Queues - MIT OpenCourseWare
Stochastic processes. Markov processes. • A Markov process is a stochastic process in which the probability of finding X at some value at time.
[13]
[PDF] Lecture 17 : Stochastic Processes II - MIT OpenCourseWare
To formally define a stochastic process, there needs to be an underlying probability space (Ω,P). A stochastic process X is then a map from the universe Ω ...
[14]
[PDF] Lecture 18 : Itō Calculus - MIT OpenCourseWare
Theorem 1.1. ( Ito's lemma) Let f(t, x) be a smooth function of two vari- ables, and let Xt be a stochastic process satisfying dXt = µtdt + σtdBt for a.
[15]
[PDF] Lecture 9: Filteration and martingales - MIT OpenCourseWare
Oct 2, 2013 · A stochastic process {Xt} is said to be adapted to filtration {Ft} if Xt ∈ Ft for every t. Definition 2. A stochastic process {Xt} adapted to a ...
[16]
[PDF] A random wave process - Stanford Mathematics
white noise limit. We first describe the underlying noise process/~ which we will denote by W', the formal white noise process. Let 5:(R 2) denote the space ...
[17]
Stochastic Equations in Infinite Dimensions - Google Books
This book gives a systematic and self-contained presentation of basic results on stochastic evolution equations in infinite dimensional, typically Hilbert and ...
[18]
[PDF] Stochastic Partial Differential Equations Lecture Notes
Jul 10, 2019 · Introduction. Here are some basic examples of partial differential equations (PDEs). • Heat equation: ∂u. ∂t. (t, x) = c∆u(t, x), x ∈ D ...<|control11|><|separator|>
[19]
[PDF] A Primer on Stochastic Partial Differential Equations
These notes form a brief introductory tutorial to elements of Gaussian noise analysis and basic stochastic partial differential equations (SPDEs) in general,.<|control11|><|separator|>
[20]
Evolution equations with white-noise boundary conditions
The paper is devoted to nonlinear evolution equations with nonhomogenous boundary conditions of white noise type.Missing: initial | Show results with:initial
[21]
[PDF] Differential Problems with Stochastic Boundary Conditions - POLITesi
This thesis treats partial differential equations with white noise boundary conditions, where random perturbations come from the boundary of the domain.
[22]
An introduction to stochastic partial differential equations
Sep 16, 2006 · © 1986 Springer-Verlag. About this paper. Cite this paper. Walsh, J.B. (1986). An introduction to stochastic partial differential equations. In ...
[23]
The stochastic Burgers equation - Project Euclid
The stochastic Burgers equation. L. Bertini, N. Cancrini, G. Jona-Lasinio. DOWNLOAD PDF + SAVE TO MY LIBRARY. Comm. Math. Phys. 165(2): 211-232 (1994).
[24]
The stochastic Burgers Equation | Communications in Mathematical ...
We prove the existence of solutions by showing that the Cole-Hopf transformation is meaningful also in the stochastic case. The problem is thus reduced to ...
[25]
Stochastic Burgers PDEs with random coefficients and a ...
This transformation, which has been named the Cole–Hopf transformation, allows us to construct explicit analytic solutions for the Burgers equation. This ...
[26]
The Gaussian structure of the singular stochastic Burgers equation
We consider the stochastically forced Burgers equation with an emphasis on spatially rough driving noise. We show that the law of the process at a fixed ...
[27]
Invariant measures for Burgers equation with stochastic forcing - arXiv
May 1, 2000 · This paper studies the Burgers equation with stochastic forcing, proving the existence and uniqueness of an invariant measure using a 'one ...
[28]
A stochastic formulation of Burgers' equation - ScienceDirect.com
By means of the example of Burgers' equation a stochastic approach to computational fluid dynamics is proposed. The velocity field is regarded as the ...
[29]
On the Stochastic Burgers Equation and Some Applications to ...
In this article we will summarize a selection of results on the inviscid limit of the stochastic Burgers equation and outline some applications of these results ...
[30]
Stochastic Equations in Infinite Dimensions
The aim of this book is to give a systematic and self-contained presentation of basic results on stochastic evolution equations in infinite dimensional, ...
[31]
Stochastic perturbation of the Lighthill–Whitham–Richards model via ...
May 7, 2021 · In this paper we apply the method of stochastic characteristics to a Lighthill–Whitham–Richards model. The stochastic perturbation can be seen as errors in ...
[32]
[PDF] The Numerical Approximation of Stochastic Partial Differential ...
Sep 11, 2009 · Numerical approximation of SPDEs is similar to early SODEs, facing difficulties due to infinite noise. It uses stochastic Taylor expansions, ...
[33]
[PDF] Numerical methods for stochastic partial differential equations
This thesis consists of an introductory essay followed by 4 chapters based on 4 scien- tific articles. The purpose of the introduction is to provide an overview ...
[34]
The vibrating string forced by white noise - Semantic Scholar
Jun 1, 1970 · SummaryThe equation of the vibrating string forced by white noise is formally solved, using stochastic integrals with respect to a plane ...
[35]
Non-parametric estimation for the stochastic wave equation - arXiv
Apr 29, 2024 · The spatially dependent wave speed of a stochastic wave equation driven by space-time white noise is estimated using the local observation scheme.
[36]
Dynamic Scaling of Growing Interfaces | Phys. Rev. Lett.
Mar 3, 1986 · Rev. Lett. 57, 1809 (1986). Kardar, Parisi, and Zhang Respond. Mehran Kardar, Giorgio Parisi, and Yi-Cheng Zhang. Phys. Rev. Lett. 57, 1810 ( ...Missing: original | Show results with:original
[37]
Feedback control of surface roughness in a one-dimensional Kardar ...
Feb 3, 2020 · We study the control of surface roughness during a growth process described by the Kardar-Parisi-Zhang (KPZ) equation in ( 1 + 1 ) dimensions.
[38]
A PDE Construction of the Euclidean $$\Phi ^4_3$$ Quantum Field ...
Apr 15, 2021 · We present a new construction of the Euclidean $$\Phi ^4$$ quantum field theory on $${\mathbb {R}}^3$$ based on PDE arguments.
[39]
A PDE construction of the Euclidean $Φ^4_3$ quantum field theory
Oct 3, 2018 · Abstract:We present a new construction of the Euclidean \Phi^4 quantum field theory on \mathbb{R}^3 based on PDE arguments.
[40]
Stochastic boundary control design for Timoshenko beams with ...
Aug 18, 2017 · The exact nonlinear partial differential equations governing motion of the beams are derived and used in the control design.
[41]
Filtering of SPDEs: The Ensemble Kalman Filter and related methods
Jun 29, 2023 · The signal SPDE is allowed to be nonlinear and is posed in the standard abstract variational setting. Its coefficients are assumed to satisfy ...Missing: processing | Show results with:processing
[42]
Stochastic Closures for Wave–Current Interaction Dynamics
Jul 20, 2019 · The turbulent wave–current mixing by Langmuir circulation seen in the OSBL is important in climate modeling, because it controls the ...<|control11|><|separator|>
[43]
Asymptotic behaviors and stochastic traveling waves in stochastic ...
For the Fisher-KPP equation subjected to the environmental noise, the established asymptotic behavior of the front of the wavefront still holds even if the ...Missing: population | Show results with:population
[44]
https://www.aimsciences.org/article/doi/10.3934/dcdsb.2020323
[45]
Stochastic neural field equations: a rigorous footing - PMC - NIH
We here consider a stochastic version of the classical neural field equation that is currently actively studied in the mathematical neuroscience community.Missing: transmission | Show results with:transmission
[46]
Numerical solution of the stochastic neural field equation with ...
The main goal of the present work is to investigate the effect of noise in some neural fields, used to simulate working memory processes.
[47]
Adapted solution of a degenerate backward spde, with applications
These relations lead to an adapted version of stochastic Feynman-Kac formula, as well as a stochastic Black-Scholes formula in mathematical finance.
[48]
A stochastic partial differential equation model for the pricing of ...
We prove existence and uniqueness of solutions to this SPDE and show how our model is able to capture, in a flexible way, the prices of credit risky tranches ...
[49]
SPDE for stochastic SIR epidemic models with infection-age ... - arXiv
Oct 22, 2024 · We study the stochastic SIR epidemic model with infection-age dependent infectivity for which a measure-valued process is used to describe the ages of ...
[50]
[PDF] Stochastic evolution equations - Semantic Scholar
Jul 1, 1981 · Stochastic evolution equations · N. Krylov, B. Rozovskii · Published 1 July 1981 · Mathematics · Journal of Soviet Mathematics.
[51]
On the pathwise uniqueness of stochastic partial differential ...
Dec 10, 2010 · In this paper, we prove a pathwise uniqueness result of a class of stochastic partial differential equations driven by space-time white noise ...
[52]
Pathwise uniqueness for a SPDE with Hölder continuous coefficient ...
May 24, 2013 · Abstract:In this paper we study the pathwise uniqueness of solution to the following stochastic partial differential equation (SPDE) with ...
[53]
On the Hölder regularity of a linear stochastic partial-integro ... - arXiv
Nov 19, 2020 · When the 1d stochastic heat equation is driven by white noise, solutions are continuous with space and time regularity that is Hölder $(1/2 ...
[54]
[PDF] Hölder-continuity for the nonlinear stochastic heat equation with ...
Abstract: We study space-time regularity of the solution of the nonlinear stochas- tic heat equation in one spatial dimension driven by space-time white ...
[55]
Blow-up for the stochastic nonlinear Schrödinger equation with ...
We study the influence of a multiplicative Gaussian noise, white in time and correlated in space, on the blow-up phenomenon in the supercritical nonlinear ...Missing: criteria | Show results with:criteria
[56]
[math/0506595] Blow-up for the stochastic nonlinear Schrodinger ...
Jun 29, 2005 · We study the influence of a multiplicative Gaussian noise, white in time and correlated in space, on the blow-up phenomenon in the supercritical nonlinear ...Missing: criteria | Show results with:criteria
[57]
Martingale solutions and Markov selections for stochastic partial ...
We present a general framework for solving stochastic porous medium equations and stochastic Navier–Stokes equations in the sense of martingale solutions.
[58]
Stochastic PDEs on graphs as scaling limits of discrete interacting ...
Aug 5, 2017 · These give the first rigorous connection between SPDE on graphs and more discrete models, specifically, interacting particle systems and ...
[59]
[PDF] An Introduction to Stochastic PDEs - of Martin Hairer
Jul 3, 2023 · fact that measures in infinite-dimensional spaces have a strong tendency of being mutually singular. ... on more general manifolds or ...<|control11|><|separator|>