Fact-checked by Grok 2 weeks ago

Delay differential equation

A delay differential equation (DDE) is a type of functional differential equation in which the time derivative of an unknown function at a given time depends not only on the function's value at that time but also on its values at one or more previous times, referred to as delays. Unlike ordinary differential equations (ODEs), which require only initial conditions at a single point, DDEs necessitate an initial history function defined over an interval equal to the longest delay to ensure existence and uniqueness of solutions, assuming the right-hand side is continuous and locally Lipschitz. DDEs model systems where time lags are inherent, such as maturation periods in or signal transmission in , leading to behaviors like oscillations or switches not captured by ODEs. Their solutions often exhibit discontinuities in derivatives at the initial time, which propagate through higher-order derivatives but smooth out over time, complicating numerical methods like Runge-Kutta solvers adapted for . Historically, DDEs trace back to early 20th-century work by on hereditary effects and Vito Volterra's 1931 integrodifferential models for predator-prey interactions, with systematic study emerging in the 1950s through contributions from A.D. Myshkis and N.N. Krasovskii. Applications span (e.g., models with delays), (e.g., feedback control systems), (e.g., delayed responses), and (e.g., neural firing with synaptic delays), highlighting their role in capturing realistic temporal dynamics. Analysis techniques include characteristic equations for and for nonlinear cases, often revealing Hopf bifurcations as delays increase.

Fundamentals

Definition and General Form

A delay differential equation (DDE) is a functional differential equation in which the derivative of the unknown function at time t depends not only on the function's value at t, but also on its values at previous times, thereby incorporating a time delay into the dynamics. This contrasts with ordinary differential equations (ODEs), where the derivative depends solely on the current state and time. The general form of a first-order with a single constant delay \tau > 0 is \dot{x}(t) = f(t, x(t), x(t - \tau)), where x(t) \in \mathbb{R}^n is the , t \geq 0, and f: \mathbb{R} \times \mathbb{R}^n \times \mathbb{R}^n \to \mathbb{R}^n is a specifying the right-hand side. A more general formulation, accommodating variable delays or distributed delays, is \dot{x}(t) = f(t, x(t), x_t), where x_t denotes the history function defined by x_t(\theta) = x(t + \theta) for \theta \in [-\tau, 0], and f: \mathbb{R} \times \mathbb{R}^n \times C([-\tau, 0], \mathbb{R}^n) \to \mathbb{R}^n. Here, C([-\tau, 0], \mathbb{R}^n) is the space of continuous functions on the delay interval, emphasizing the functional dependence on past states. To initiate a unique solution, DDEs require an initial condition specified not by a single point value as in ODEs, but by a continuous initial function \phi \in C([-\tau, 0], \mathbb{R}^n) on the entire delay interval, such that x(\theta) = \phi(\theta) for \theta \in [-\tau, 0]. This history function captures the system's state over the lag period, rendering the solution framework infinite-dimensional, as the state at any time is determined by an uncountable set of past values rather than a finite-dimensional vector. Consequently, DDEs evolve in a Banach space of functions, leading to richer dynamical behaviors compared to finite-dimensional ODEs. Such equations model real-world phenomena involving inherent time lags, such as signal propagation delays in transmission lines or periods in biological populations.

Historical Development

The origins of delay differential equations (DDEs) can be traced to the early . Émile Picard, in 1908 at an international of mathematicians, emphasized the significance of accounting for hereditary effects in models, advocating for functional equations over classical equations to incorporate past states. Foundational connections emerged from Vito 's work on integral equations that modeled hereditary effects in biological populations and physical phenomena like . In his 1931 monograph Leçons sur la théorie mathématique de la lutte pour la vie, introduced delay-like structures in predator-prey models, highlighting the role of past states in dynamic systems. A significant early milestone in the linear theory occurred in 1950, when N.D. Hayes analyzed the roots of transcendental equations arising in difference- systems, providing essential tools for understanding solution behavior in linear DDEs with constant delays. The 1940s and 1950s marked growing interest in DDEs, driven by applications in engineering and , where delays represented signal transmission lags, including systematic studies by A.D. Myshkis in 1951 and N.N. Krasovskii in 1959. Edmund Pinney's 1958 book Ordinary Difference-Differential Equations advanced the field by detailing analytical methods, including the introduction of characteristic equations for determining and oscillatory solutions in linear cases. This period culminated in the 1960s with expanded theoretical frameworks, notably Richard Bellman and Kenneth L. Cooke's 1963 monograph Differential-Difference Equations, which offered a systematic treatment of both linear and nonlinear DDEs, emphasizing asymptotic , perturbation methods, and connections to integral equations. The late 20th century saw the maturation of DDE theory, with Jack K. Hale's 1977 text Theory of Functional Differential Equations establishing a rigorous abstract framework that encompassed retarded, neutral, and advanced types, focusing on existence, uniqueness, and qualitative properties like invariance and attractors. From the 1980s through the 2000s, research proliferated on numerical methods for practical computation, including Runge-Kutta collocation schemes and event-driven integrators, alongside deeper explorations of nonlinear DDEs in chaotic dynamics and bifurcation analysis, as detailed in monographs like Bellen and Zennaro's 2003 Numerical Methods for Delay Differential Equations. Since the , and accelerating into the , the field has emphasized DDEs to incorporate random perturbations in models from to . Computational advancements have included specialized software like DDE23 in , while post-2020 innovations leverage for parameter estimation and solution discovery, exemplified by data-driven frameworks such as DDE-Find for inferring delays and dynamics from time-series data.

Classifications and Examples

Types of Delay Differential Equations

Delay differential equations (DDEs) are classified based on the nature of the delays involved, the linearity of the functional terms, and specific structural features that distinguish particular subclasses. The primary categorization by delay structure includes equations with constant delays, variable delays, and distributed delays. Constant delay equations feature a fixed time lag \tau > 0, as in the prototypical form \dot{x}(t) = f(t, x(t), x(t - \tau)). Variable delay equations allow the delay to depend on time, \tau(t), or even on the state, introducing additional complexity in the dependence on past values. Distributed delay equations incorporate a continuum of delays through an integral over past states, such as \dot{x}(t) = \int_{-\infty}^t k(t - s) f(x(s)) \, ds, where k is a kernel function representing the distribution of delays. Regarding linearity, DDEs are divided into linear and nonlinear types. Linear DDEs take the form \dot{x}(t) = A(t) x(t) + B(t) x(t - \tau) for vector-valued x, where A and B are matrices, enabling techniques like characteristic equations for analysis. Nonlinear DDEs generalize this by allowing f to be a nonlinear of x(t) and x(t - \tau), which complicates and properties but arises in many realistic models. A further distinction within linear and nonlinear DDEs is between retarded and types: retarded DDEs, the most common, have delays only in the state variables, as in \dot{x}(t) = f(t, x(t), x(t - \tau)); DDEs include delays in the derivatives, exemplified by \dot{x}(t) - \dot{x}(t - \tau) = f(t, x(t), x(t - \tau)), affecting the equation's and requiring adjusted state spaces. Special types of DDEs include pantograph equations, which feature proportional delays scaling with time, such as \dot{x}(t) = a x(t) + b x(q t) for constants a, b and $0 < q < 1. These arise in applications like electric locomotive dynamics and exhibit singular behavior at t=0 due to the contracting delay. Advanced DDEs, where delays point to future states (e.g., t + \tau), are less common and often ill-posed for initial value problems but appear in control contexts. DDEs occupy a position intermediate between ordinary differential equations (ODEs), which have finite-dimensional dynamics, and partial differential equations (PDEs), which involve infinite-dimensional spatial variations; the time delays in DDEs induce an infinite-dimensional state space via history functions, bridging these categories without explicit spatial dependence.

Illustrative Examples

A fundamental example of a linear delay differential equation with constant delay is the scalar equation \dot{x}(t) = -a x(t) + b x(t - \tau), where a > 0, b > 0, and \tau > 0 is the delay. This model arises in contexts such as feedback control systems where the rate of change depends on both current and past states. For appropriate parameter values, such as when b > a and \tau exceeds a critical , solutions exhibit oscillatory around the x = 0, with the and of oscillations increasing as \tau grows, leading to sustained or growing not present in the corresponding \dot{x}(t) = -a x(t) + b x(t). An illustrative nonlinear delay differential equation with constant delay is the Mackey-Glass equation, modeling oscillations in production: \dot{x}(t) = -\gamma x(t) + \beta \frac{x(t - \tau)}{1 + x(t - \tau)^n}, where \gamma > 0 is the decay rate, \beta > 0 the production scaling, \tau > 0 the maturation delay, and n > 0 a nonlinearity (often n = 10). This equation captures physiological where cell production is inhibited by recent high concentrations, producing complex dynamics including periodic oscillations and for \tau > 17 under standard parameters like \beta = 0.2, \gamma = 0.1. For distributed delays, where the past history is weighted by a memory , consider the logistic population growth model \dot{x}(t) = r x(t) \left(1 - \frac{1}{K} \int_{-\infty}^t e^{-\mu (t-s)} x(s) \, ds \right), with intrinsic growth rate r > 0, carrying capacity K > 0, and decay parameter \mu > 0 in the exponential kernel representing fading of past population sizes. This equation extends the classic logistic model to account for averaged delays in maturation or in ecological systems, such as predator-prey interactions or . These examples highlight how delays in differential equations can induce qualitative behaviors absent in their delay-free counterparts, such as damped or undamped oscillations and potential , due to the incorporation of historical dependencies that amplify phase shifts in loops. The scalar linear case represents constant-delay linear DDEs, the Mackey-Glass nonlinear constant-delay DDEs, and the integral form distributed-delay DDEs.

Solution Methods

Analytical Approaches

Analytical approaches to solving delay differential equations (DDEs) are primarily exact methods applicable to specific classes, such as linear equations with constant coefficients or simple nonlinear forms with constant delays, but they do not yield general closed-form solutions for arbitrary DDEs. These techniques transform the infinite-dimensional nature of DDEs—due to dependence on past states—into solvable forms, often by leveraging (ODE) theory or integral transforms. One fundamental analytical method is the method of steps, which applies to DDEs with constant delay \tau > 0 and initial condition given by a continuous function \phi on [-\tau, 0]. For an equation like \dot{x}(t) = f(t, x(t), x(t - \tau)), the solution is constructed piecewise on successive intervals [n\tau, (n+1)\tau] for n = 0, 1, 2, \dots. On each interval, the delayed term x(t - \tau) is known from the previous solution segment, reducing the DDE to a standard ODE solvable by conventional methods. For example, consider the simple linear DDE \dot{x}(t) = x(t - 1) with \phi(\theta) = 1 for \theta \in [-1, 0]; on [0, 1], it becomes \dot{x}(t) = 1, yielding x(t) = t + 1; on [1, 2], \dot{x}(t) = t, so x(t) = \frac{t^2}{2} + c, with continuity determining c = \frac{3}{2}, and this process continues explicitly for each interval. This method provides the unique solution in analytic form for retarded DDEs but requires infinite steps for the full trajectory, making it impractical for long times without patterns. Another approach involves reducing DDEs to systems through , either exactly into infinite-dimensional spaces or approximately into finite-dimensional s. In the functional-analytic framework, a retarded \dot{x}(t) = f(x_t), where x_t(\theta) = x(t + \theta) for \theta \in [-\tau, 0], is embedded into an abstract evolution equation in the C([-\tau, 0], \mathbb{R}^n) of continuous functions, treating the history x_t as the and deriving \frac{d}{dt} x_t = A x_t via the linear operator A \phi(\theta) = \phi'( \theta) for \theta \in [-\tau, 0) and A \phi(0) = f(\phi). For finite-dimensional approximations, one can embed via state augmentation: for a scalar \dot{x}(t) = f(x(t), x(t - \tau)), introduce y_1(t) = x(t) and y_2(t) = x(t - \tau), leading to the system \dot{y}_1(t) = f(y_1(t), y_2(t)) and \dot{y}_2(t) = y_1(t) - y_2(t)/\tau + higher-order terms approximating the delay shift, though this is typically limited to low-order approximations for s and does not yield exact solutions. For linear DDEs with constant coefficients, the Laplace transform provides an exact solution in the transform domain. Consider the scalar equation \dot{x}(t) = a x(t) + b x(t - \tau) + g(t) for t \geq 0, with initial function \phi on [-\tau, 0]. Applying the Laplace transform \hat{x}(s) = \int_0^\infty e^{-st} x(t) \, dt yields \hat{x}(s) = \frac{\phi(0) + b e^{-s \tau} \int_{-\tau}^0 e^{-s \theta} \phi(\theta) \, d\theta + \hat{g}(s)}{s - a - b e^{-s \tau}}, where the numerator accounts for the initial history and forcing term, and the denominator is the characteristic function. The solution x(t) is recovered via inverse Laplace transform, often involving residues or series expansions, as in the homogeneous case where poles determine the behavior. For systems, matrix forms extend this approach, sometimes using the matrix Lambert W function for explicit inversion. Despite these tools, analytical approaches are feasible only for linear constant-coefficient DDEs or simple nonlinear cases with constant delays, as variable delays, multiple delays, or strong nonlinearities preclude closed-form expressions and necessitate infinite expansions or piecewise constructions. No general analytical method exists for arbitrary DDEs, highlighting the need for numerical or qualitative alternatives in complex scenarios.

Numerical Methods

Numerical methods for delay differential equations (DDEs) approximate solutions by discretizing the time domain while incorporating the history function and delayed terms, extending techniques from ordinary differential equations (ODEs) to handle the infinite-dimensional nature of DDEs. These methods must address challenges such as evaluating the solution at non-grid points corresponding to delays and propagating discontinuities in derivatives that arise when the integration time aligns with delay intervals. Discretization approaches commonly adapt one-step methods to DDEs. Runge-Kutta methods are extended using continuous formulations that generate dense output polynomials for interpolating the history at arbitrary delay points, ensuring the approximation maintains the method's ; for instance, a fourth-order continuous Runge-Kutta scheme uses stage derivatives to fit Hermite interpolants with local error bounded by O(h^5). Collocation methods, which enforce the DDE at specific nodes within each step using piecewise polynomials, similarly provide high-order dense output and are effective for both retarded and neutral DDEs, with order up to $2s-1 for s-stage Radau . Specialized solvers implement these discretizations with tailored features. The dde23 algorithm in utilizes a variable-step, variable-order explicit Runge-Kutta method (orders 1–3) for constant-delay DDEs, incorporating automatic detection and propagation of discontinuities at delay multiples of the step size to ensure accurate derivative jumps without halting integration. For DDEs, which involve delayed derivatives and often exhibit , the Radau IIA method (e.g., three-stage order 5) solves the implicit system via simplified iterations, handling the neutral term through approximation of delayed derivatives and adapting steps for small or vanishing delays. Error analysis for these methods indicates that global convergence order matches the p under conditions on the functional, but of delayed values can degrade the effective order to p-1 for low-order explicit schemes if delays are not resolved finely. issues arise prominently in long-delay problems, where the ratio of delay to interval amplifies eigenvalue spreads, requiring implicit methods like Radau to avoid and achieve efficient computation with step sizes independent of severity. Post-2020 developments have incorporated to enhance numerical strategies for DDEs, particularly in parameter estimation. Neural Delay Differential Equations (NDDEs), which parameterize the and delay kernels as neural networks within a continuous-depth , facilitate data-driven of unknown delays and parameters by optimizing via , outperforming standard Neural ODEs on delayed tasks such as synthetic time-series fitting.

Linear Delay Differential Equations

Characteristic Equation

For linear constant-coefficient delay differential equations of the form \dot{x}(t) = A x(t) + B x(t - \tau), where x(t) \in \mathbb{R}^n, A, B \in \mathbb{R}^{n \times n}, and \tau > 0 is a constant delay, solutions can be sought in the exponential form x(t) = e^{\lambda t} v with \lambda \in \mathbb{C} and constant vector v \neq 0. Substituting this ansatz yields \lambda e^{\lambda t} v = A e^{\lambda t} v + B e^{\lambda (t - \tau)} v = e^{\lambda t} (A v + B e^{-\lambda \tau} v), which simplifies to (\lambda I - A - B e^{-\lambda \tau}) v = 0. For nontrivial solutions, the characteristic matrix must be singular, leading to the characteristic equation \Delta(\lambda) = \det(\lambda I - A - B e^{-\lambda \tau}) = 0. The roots \lambda_k of this equation form the of the system, consisting of infinitely many eigenvalues due to the transcendental nature of the term. The real parts \operatorname{Re}(\lambda_k) govern the short-term growth or of solutions, as modes with \operatorname{Re}(\lambda_k) > 0 contribute to while those with \operatorname{Re}(\lambda_k) < 0 decay. In general, the characteristic function \Delta(\lambda) is a quasipolynomial of the form \sum_{k=0}^n P_k(\lambda) e^{-k \lambda \tau} = 0, where the P_k(\lambda) are polynomials of degree at most n-k, arising from the expansion of the determinant. A simple scalar example illustrates this: consider \dot{x}(t) + a x(t) + b x(t - \tau) = 0, or equivalently \dot{x}(t) = -a x(t) - b x(t - \tau). The characteristic equation reduces to \lambda + a + b e^{-\lambda \tau} = 0. The roots \lambda_k can be plotted in the complex plane, revealing chains of eigenvalues asymptotic to vertical lines \operatorname{Re}(\lambda) \approx \frac{1}{\tau} \ln |b|, with the rightmost roots determining stability thresholds. This framework extends the characteristic equation for ordinary differential equations \dot{x}(t) = A x(t), where \det(\lambda I - A) = 0 yields finitely many eigenvalues from a polynomial determinant. The delay introduces the transcendental exponential factor, transforming the equation into a quasipolynomial with infinitely many roots distributed throughout the complex plane.

Stability Analysis

The asymptotic stability of the zero solution to a linear delay differential equation is equivalent to all characteristic roots having negative real parts, which is a necessary and sufficient condition for linear systems. This spectral criterion extends the classical result from ordinary differential equations to the infinite-dimensional setting of functional differential equations, where the location of roots in the complex plane determines the long-term behavior of solutions. Hopf bifurcations arise in linear delay differential equations as the delay parameter \tau is varied, occurring when a pair of complex conjugate characteristic roots crosses the imaginary axis, typically leading to the onset of periodic solutions in the nonlinear case. The bifurcation condition requires a purely imaginary root \lambda = i\omega with \omega > 0, and for equations of the form \dot{x}(t) = a x(t) + b x(t - \tau), \omega = \sqrt{b^2 - a^2} assuming |b| > |a| and signs such that a stability switch occurs, with the critical delays given by \tau_k = \frac{1}{\omega} \arccos\left( -\frac{a}{b} \right) + \frac{2k\pi}{\omega}, \quad k = 0, 1, 2, \dots, where the branch of arccos is chosen such that \sin(\omega \tau_k) = -\omega / b. Stability criteria for linear delay differential equations are classified as delay-independent or delay-dependent, depending on whether they hold for all \tau \geq 0 or only for bounded ranges of \tau. Delay-independent ensures asymptotic regardless of delay size, with a representative sufficient condition for the system \dot{x}(t) = A x(t) + B x(t - \tau) being that A is Hurwitz and \|B\| < \|A\| in appropriate matrix norms, guaranteeing all remain in the left half-plane. In contrast, delay-dependent criteria, such as analogs of the adapted to quasi-polynomials, provide regions for specific \tau intervals by analyzing root crossings. Key tools for stability analysis include the argument principle, which counts the number of characteristic roots with positive real parts by evaluating the change in argument of the along the imaginary axis contour. Pseudospectral methods offer computational approaches to approximate the and boundaries efficiently, particularly for high-dimensional systems, by discretizing the infinitesimal generator.

Applications

In Biological and Physical Systems

Delay differential equations (DDEs) are particularly valuable in modeling biological systems where time lags arise naturally from processes such as maturation, incubation, or response delays. In , a classic example is Nicholson's blowflies equation, which captures the oscillatory behavior of blowfly populations due to delays in larval development: \dot{x}(t) = -\delta x(t) + P e^{-a x(t - \tau)}, where x(t) represents the adult population, \delta is the adult death rate, P is the birth rate parameter, a scales , and \tau is the delay. This model, originally proposed by Nicholson in 1954 and analyzed as a DDE by Gurney, Blythe, and Nisbet in 1980, demonstrates how delays can induce sustained oscillations, mirroring observed population cycles in controlled experiments. In epidemiology, DDEs extend compartmental models like the SIR framework to account for intrinsic delays, such as the latent period before infectiousness. A simplified delay SIR model for the infected compartment is: \dot{I}(t) = \beta S(t) I(t) - \gamma I(t - \tau), where I(t) is the number of infectious individuals, S(t) is susceptible, \beta is the transmission rate, \gamma is the recovery rate, and \tau represents the incubation delay. This formulation, introduced by Hethcote and van den Driessche in 1989, reveals how delays can destabilize equilibria and promote epidemic waves, as seen in analyses of diseases like tuberculosis with prolonged latency. More recent applications include COVID-19 models incorporating delays for incubation and quarantine, as analyzed in studies up to 2023, revealing insights into wave dynamics. Physiological systems also benefit from modeling to incorporate delays in signaling pathways. In glucose-insulin regulation, delays occur between insulin secretion and its effect on , leading to models like those by Sturis et al. (1991), where the dynamics exhibit ultradian oscillations akin to observed blood glucose rhythms in humans. Similarly, neuronal models extend the Hodgkin-Huxley framework to include synaptic transmission delays, as explored in neuronal models with delays, e.g., by Longtin and others, which can generate irregular firing patterns and synchronization in neural networks, relevant to conditions like . In physical systems, DDEs arise in scenarios with delayed or effects. For instance, the wave equation with delayed conditions, \partial_{tt} u(x,t) = \partial_{xx} u(x,t) for $0 < x < 1, subject to u(1,t) = f(u(1,t-\tau)), models wave propagation in materials where responses lag due to , as studied by Nicaise and Pignotti (2006); such delays can lead to spatiotemporal instabilities. In vibration dynamics, hereditary materials with viscoelastic are modeled by DDEs like \ddot{x}(t) + a \dot{x}(t) + b x(t) + \int_{-\infty}^t k(t-s) \dot{x}(s) \, ds = 0, where the integral term approximates delay effects, capturing damped oscillations in polymers as analyzed by Drozdov (1996).

In Engineering and Control Theory

In engineering and control theory, delay differential equations (DDEs) are essential for modeling time lags in feedback systems, where delays arise from signal transmission, processing, or actuation, potentially leading to instability or oscillations. A prominent application is in process control, where the Smith predictor compensates for dead-time delays in systems like chemical reactors or heating processes. Introduced by Smith in 1959, this model-based controller uses an internal delay-free model of the plant to predict future outputs, allowing the feedback loop to behave as if no delay were present, thereby improving stability and performance in proportional-integral-derivative (PID) control setups. Stability analysis in delayed feedback loops often relies on the characteristic equation derived from linear DDEs to design robust controllers, ensuring that roots lie in the left half-plane despite delays. For instance, in servo mechanisms or robotic arms, distributed delays from multi-stage processing can be mitigated using Lyapunov-based methods or predictor-corrector schemes, which expand the stable gain margins compared to delay-free systems. In networked control systems, DDEs model congestion control protocols like TCP/IP, where round-trip time delays govern window size adjustments in fluid approximations. These models, formulated as delay equations for sending rates and queue lengths, reveal stability conditions under varying traffic loads, enabling optimizations that reduce and . In , neutral DDEs describe wave propagation in transmission lines, incorporating in both state and derivative terms to capture high-frequency effects and reflections. These equations arise in partial element (PEEC) models for interconnects, aiding the design of networks by predicting voltage transients. Recent applications post-2020 include autonomous , where sensor-actuator in path-tracking control are compensated using controllers with for path tracking under communication lags up to 500 ms. Similarly, in , distributed delay DDEs optimize amid multiple delivery lags, quantifying the and stabilizing demand-supply oscillations through predictive ordering policies.

References

  1. [1]
    Delay-differential equations - Scholarpedia
    Aug 2, 2015 · Delay differential equations differ from ordinary differential equations in that the derivative at any time depends on the solution.Introduction · Derivative Discontinuities · Available Delay Differential... · An Example
  2. [2]
    [PDF] A Brief Introduction to Delay Differential Equations and Applications
    ”Delay Differential Equations and Applications. 1. J.K. Hale, History Of Delay Equations. 2. J. Arino and P. van den Driessche, Time Delays in Epidemic Models:.
  3. [3]
    [PDF] Approximation Of Continuously Distributed Delay Differential ...
    Aug 1, 2017 · 1.1 DELAY DIFFERENTIAL EQUATION (DDE). Definition. A differential ... The general form of the time dependent DDE for the function x(t) ...
  4. [4]
    [PDF] Delay-Differential Equations - FSU Math
    Figure 2: Solution to delayed-logistic equation with α = 2. is capable of producing periodic motion, in contrast to a single ordinary differential equation.
  5. [5]
    [PDF] arXiv:1903.02098v1 [math.NA] 5 Mar 2019
    Mar 5, 2019 · Thus, the equation 1 is a delay differential equation of neutral type with constant delays [4], the general form for which would be. ˙x(t) ...<|control11|><|separator|>
  6. [6]
    [PDF] Galerkin Approximations of General Delay Differential Equations ...
    Delay differential equations (DDEs) are often used to model systems with time-delayed effects, and they have found applications in fields such as climate ...<|control11|><|separator|>
  7. [7]
    Differential-Difference Equations - RAND
    Differential-Difference Equations. Richard Ernest Bellman, Kenneth L. Cooke. ResearchPublished 1963. Download PDF.Missing: delay monograph<|separator|>
  8. [8]
    Theory of Functional Differential Equations - SpringerLink
    © 1977. 2nd edition; View latest edition. Accessibility Information. Theory of Functional Differential Equations. Overview. Authors: Jack K. Hale. Jack K. Hale.
  9. [9]
    [PDF] Delay Differential Equations in Single Species Dynamics
    In this survey, we shall review various delay differential equations models arising from studying single species dynamics. Let x(t) denote the population ...
  10. [10]
    Ordinary and Delay Differential Equations - SpringerLink
    Book Title: Ordinary and Delay Differential Equations · Authors: R. D. Driver · Series Title: Applied Mathematical Sciences · Publisher: Springer New York, NY.
  11. [11]
    Embedding Model Reduction Method for Time-Delay Differential ...
    The approach is to first change the DAE system into a parametric ordinary differential equation (ODE) system via the ε-embedding technique.Missing: seminal | Show results with:seminal
  12. [12]
    [PDF] Solution of Systems of Linear Delay Differential Equations via ...
    In this paper, the Laplace transform approach for solving linear systems of DDEs is developed using the matrix Lam- bert W function method.Missing: seminal | Show results with:seminal
  13. [13]
    [PDF] Solving delay differential equations
    In this chapter we consider a broad class of delay differential equations and their numerical integration. The literature on this subject is very wide; ...
  14. [14]
    Solving DDEs in Matlab - ScienceDirect.com
    We have written a program, dde23, to solve delay differential equations (DDEs) with constant delays in Matlab. In this paper we discuss some of its features.
  15. [15]
    [PDF] Implementing Radau IIA methods for stiff delay differential equations
    Feb 5, 2001 · Radau IIA methods are implicit Runge-Kutta methods, using collocation based on Radau nodes, with 'stiff accuracy' and an invertible coefficient ...
  16. [16]
    Stiff delay equations - Scholarpedia
    Nov 9, 2007 · The algorithm extends Hairer and Wanner's Radau5 code, and is based on the 3 stage Radau IIA collocation method. Snddelm: a Fortran77 code by ...<|control11|><|separator|>
  17. [17]
    [2102.10801] Neural Delay Differential Equations - arXiv
    Feb 22, 2021 · Here we propose a new class of continuous-depth neural networks with delay, named as Neural Delay Differential Equations (NDDEs).
  18. [18]
    Introduction to Functional Differential Equations - SpringerLink
    One major change was a complete new presentation of lin ear systems (Chapters 6~9) for retarded and neutral functional differential equations.
  19. [19]
    Pseudospectral methods for stability analysis of delayed dynamical ...
    Nov 28, 2013 · The objective is the efficient computation of stability charts for varying or uncertain system parameters. A benchmark set of tests is provided ...
  20. [20]
    Smith predictor based control of multi time-delay processes
    Early work by Smith (1959) introduced the predictor concept, which has since become the foundation for many time-delay compensation techniques. However, the ...
  21. [21]
    Stability and stabilization of delay differential systems - ScienceDirect
    The system parameters and the delays are assumed to be imperfectly known. The instantaneous state vector is perturbed by a bounded external disturbance vector.
  22. [22]
    https://doi.org/10.1016/j.jde.2024.06.029
  23. [23]
    A Robust Path Tracking Controller for Autonomous Mobility ... - MDPI
    This paper proposes a Lyapunov-based backstepping steering control algorithm to compensate for control delays in autonomous mobility systems.Missing: post- | Show results with:post-
  24. [24]
    A Quality Decision Model Considering the Delay Effects in a Dual ...
    Firstly, we constructed a delayed differential equation for the effect of quality on goodwill based on the Nerlove–Arrow model for a two-channel supply chain in ...Missing: post- | Show results with:post-