Sort:
Open Access Research Article Issue
A physics-informed Chebyshev spectral neural network for distributed-order fractional optimal control problems
AIMS Mathematics 2026, 11(7): 23308-23343
Published: 15 July 2026
Abstract PDF (2.1 MB) Collect
Downloads:11

In this paper, a physics-informed Chebyshev spectral neural network (PI-CSNN) framework was proposed for solving distributed-order fractional optimal control problems (DOFOCPs). The proposed approach combines Chebyshev spectral approximations with physics-informed residual minimization to construct an accurate computational framework for DOFOCPs. The distributed-order fractional operators are incorporated directly into the neural-spectral formulation, allowing the method to treat both single-order and distributed-order optimal control models within a unified framework. Several numerical examples involving solutions with limited regularity and time-invariant and time-varying DOFOCPs were presented to evaluate the performance of the proposed method. The numerical results demonstrated high accuracy, systematic convergence, and reliable approximations of the state, adjoint, and control variables. Comparisons with existing numerical methods showed that the proposed PI-CSNN framework achieves competitive optimal cost values and close agreement with benchmark solutions. Overall, the proposed PI-CSNN framework provides an accurate and flexible computational methodology for DOFOCPs.

Open Access Research Article Issue
Numerical and deep learning methods for diffusion model with fractional Laplacian operator and its application to signal denoising
AIMS Mathematics 2026, 11(3): 7235-7263
Published: 15 March 2026
Abstract PDF (1.7 MB) Collect
Downloads:20

In modern signal processing applications, denoising signals remain an essential task, as it is difficult to reduce noise without compromising essential structural information. While standard Laplacian operators are limited in their capacity to handle long-range interactions and maintain fine-scale features, classical diffusion models offer a sound mathematical foundation for signal smoothing. To overcome these limitations, in this study, we developed a novel numerical and deep learning approach driven by a nonlocal fractional Laplacian operator of the form ( − Δ ) s ( 0 < s < 1 ), which captures long-range interactions within signals. To accurately approximate the fractional Laplacian dynamics, a classical second-order finite difference (FD) discretization of the Laplacian was constructed in space, and the fractional operator was defined through its spectral matrix power. The resulting semi-discrete system was integrated in time using an unconditionally stable Crank-Nicolson (CN) scheme. In addition to this model-based method, a convolutional neural network (CNN) architecture was employed as a refinement step, trained to learn the nonlinear mapping from noisy to clean signals. To evaluate the denoising capability and effectiveness of the proposed methods, different input signals contaminated with additive white Gaussian noise at prescribed signal-to-noise (SNR) ratios were considered. The SNR and root mean square error (RMSE) were used as quantitative performance measures. The results showed that the FD-CN and the CNN methods perform well in conjunction. The numerical scheme smooths the data in a numerically controlled way, and the CNN captures complex local structures and adapts to non-Gaussian noise patterns. In addition, the proposed framework was extended to multichannel real-world ECG signals, demonstrating its robustness in handling correlated noise across multiple leads. The multichannel experiments confirmed that the FD-CN and CNN-based approaches remain effective under realistic, real-data conditions.

Open Access Research Article Issue
Two-dimensional numerical modeling and simulation of heat and moisture transfer during drying
AIMS Mathematics 2026, 11(3): 8762-8791
Published: 15 March 2026
Abstract PDF (658.6 KB) Collect
Downloads:15

This study explores a two-dimensional model for transient heat and moisture transfer in a rectangular domain subject to convective boundary conditions. The governing equations consist of parabolic partial differential equations, which are derived from Fourier's law of heat conduction and Fick's law of mass diffusion. The model equations are discretized using central finite-difference approximations in space and a Crank–Nicolson (CN) scheme in time. To efficiently handle the multidimensional diffusion operators, an operator splitting strategy based on the Peaceman–Rachford alternating direction implicit (ADI) method is employed. This approach decomposes the two-dimensional diffusion operator into a sequence of one-dimensional subproblems, each associated with a single spatial direction, which significantly reduces the computational complexity of the fully implicit scheme while retaining second-order accuracy in time and space and unconditional stability for linear diffusion operators. The numerical scheme is shown to be unconditionally stable for the diffusion operators and computationally efficient for long-time simulations. The numerical results for the temperature and moisture fields are validated through comparison with available analytical solutions and experimental data from the literature, demonstrating very good agreement in all cases. In addition, a structure-preserving iteration method (SPIM) applied to the unsplit CN formulation is used as a benchmark reference, confirming that the proposed ADI–CN splitting achieves accuracy comparable to the unsplit CN scheme while maintaining competitive computational efficiency.

Open Access Research Article Issue
Long time behavior of higher-order delay differential equation with vanishing proportional delay and its convergence analysis using spectral method
AIMS Mathematics 2022, 7(4): 4946-4959
Published: 15 April 2022
Abstract PDF (306.5 KB) Collect
Downloads:3

Delay differential equations (DDEs) are used to model some realistic systems as they provide some information about the past state of the systems in addition to the current state. These DDEs are used to analyze the long-time behavior of the system at both present and past state of such systems. Due to the oscillatory nature of DDEs their explicit solution is not possible and therefore one need to use some numerical approaches. In this article, we developed a higher-order numerical scheme for the approximate solution of higher-order functional differential equations of pantograph type with vanishing proportional delays. Some linear and functional transformations are used to change the given interval [0, T] into standard interval [-1, 1] in order to fully use the properties of orthogonal polynomials. It is assumed that the solution of the equation is smooth on the entire domain of interval of integration. The proposed scheme is employed to the equivalent integrated form of the given equation. A Legendre spectral collocation method relative to Gauss-Legendre quadrature formula is used to evaluate the integral term efficiently. A detail theoretical convergence analysis in L∞ norm is provided. Several numerical experiments were performed to confirm the theoretical results.

Open Access Research Article Issue
Stability analysis and simulations of tumor growth model based on system of reaction-diffusion equation in two-dimensions
AIMS Mathematics 2024, 9(5): 11560-11579
Published: 15 May 2024
Abstract PDF (1.1 MB) Collect
Downloads:8

In this study, we introduce a novel framework for exploring the dynamics of tumor growth and an evolution model for two-stage carcinogenic mutations in two-dimensions based on a system of reaction-diffusion equations. It is shown theoretically that the system is globally stable in the absence of both delay and diffusion. The inclusion of diffusion does not destabilize the system, while including delay does capture the key elements of how normal cells convert into cancer cells. To further validate these results, several numerical experiments are performed for different parameter values involved in the model equation. These parameter values are chosen in the sense that they have some biological meanings using the steady states of the equilibrium points. For the purpose of simulation, a stable Euler scheme is used for temporal discretization, while a Fourier spectral method is used for space variables, which is a natural choice due to the periodic boundary conditions in the model equation. The numerical simulation results further confirm our theoretical justification.

Open Access Research Article Issue
Advanced machine learning technique for solving elliptic partial differential equations using Legendre spectral neural networks
Electronic Research Archive 2025, 33(2): 826-848
Published: 15 February 2025
Abstract PDF (1 MB) Collect
Downloads:18

In this work, a novel approach based on a single-layer machine learning Legendre spectral neural network (LSNN) method is used to solve an elliptic partial differential equation. A Legendre polynomial based approach is utilized to generate neurons that fulfill the boundary conditions. The loss function is computed by using the error back-propagation principles and a feed-forward neural network model combined with automatic differentiation. The main advantage of using this methodology is that it does not need to solve a system of nonlinear and nonsparse equations compared with other traditional numerical schemes, which makes this algorithm more convenient for solving higher-dimensional equations. Further, the hidden layer is eliminated with the help of a Legendre polynomial to enlarge the input pattern. The neural network's training accuracy and efficiency were significantly enhanced by the innovative sampling technique and neuron architecture. Moreover, the Legendre spectral approach can handle equations on more complex domains because of numerous networks. Several test problems were used to validate the proposed scheme, and a comparison was made with other neural network schemes consisting of the physics-informed neural network (PINN) scheme. We found that our proposed scheme has a very good agreement with PINN, which further enhances the reliability and efficiency of our proposed method. The absolute and relative error in both L 2 and L ∞ between exact and numerical solutions are provided, which shows that our numerical method converges exponentially.

Open Access Research Article Issue
Dynamics and simulations of stochastic COVID-19 epidemic model using Legendre spectral collocation method
AIMS Mathematics 2023, 8(2): 4220-4236
Published: 15 February 2023
Abstract PDF (488.3 KB) Collect
Downloads:13

The aim of this study is to investigate the dynamics of epidemic transmission of COVID-19 SEIR stochastic model with generalized saturated incidence rate. We assume that the random perturbations depends on white noises, which implies that it is directly proportional to the steady states. The existence and uniqueness of the positive solution along with the stability analysis is provided under disease-free and endemic equilibrium conditions for asymptotically stable transmission dynamics of the model. An epidemiological metric based on the ratio of basic reproduction is used to describe the transmission of an infectious disease using different parameters values involve in the proposed model. A higher order scheme based on Legendre spectral collocation method is used for the numerical simulations. For the better understanding of the proposed scheme, a comparison is made with the deterministic counterpart. In order to confirm the theoretical analysis, we provide a number of numerical examples.

Open Access Research Article Issue
Stability analysis of fractional two-dimensional reaction-diffusion model with applications in biological processes
AIMS Mathematics 2025, 10(5): 11732-11756
Published: 15 May 2025
Abstract PDF (1.2 MB) Collect
Downloads:59

We study the stability of a two-dimensional fractional reaction-diffusion system under the Caputo differential operator in time. Our model is based on the Grey-Scott model, a well-known coupled reaction-diffusion system that describes the interaction between two chemical species. The diffusion term captures the species special spread, while the nonlinear term in the system describes the chemical reaction, resulting in a wide range of difficult, self-organizing patterns, including spots, stripes, or spirals, depending on the parameter values. We derive conditions for local stability of the homogeneous equilibrium by linearizing the system and analyzing the eigenvalues of the Jacobian. Furthermore, we construct appropriate Lyapunov functionals to establish global asymptotic stability of the discrete model under suitable conditions. This approach seeks to provide a robust framework for analyzing complex dynamical behaviors in systems governed by fractional-order in-time reaction-diffusion systems. The numerical simulations employ the Chebyshev spectral method for spatial discretization and the L 1 scheme for fractional time derivatives. These simulations validate the theoretical findings, demonstrating the model's ability to replicate intricate patterns often observed in reaction-diffusion systems. The results suggest that the fractional-order framework enhances the understanding of pattern formation in such systems, making this model a valuable tool for studying anomalous diffusion and non-local dynamics in biological and chemical processes.

Open Access Research Article Issue
An innovative perspective on fractional inequalities through fractional operators and extended convexity
AIMS Mathematics 2026, 11(2): 4008-4042
Published: 09 February 2026
Abstract PDF (1.1 MB) Collect
Downloads:31

The theory of integral inequalities is significantly advanced by the relationship between fractional calculus and convexity. The current study used extended convex functions and Katugampola fractional operators to derive Hermite-Hadamard, Fejér-Hermite-Hadamard-type inequalities as well as a few other fractional integral inequalities. We used two extended-type convex functions: log-convex and exponentially trigonometric convex functions. Our conclusions were supported by tabular data and graphical representations, which offer numerical and visual validation of the explored results. This study improves mathematical analysis and expands the scope of these inequalities by emphasizing their applicability across different forms of convexity. The knowledge acquired is highly valuable for theoretical investigation as well as real-world applications in a variety of scientific fields.

Open Access Research Article Issue
Legendre spectral-Monte Carlo method and its error analysis for nonlinear stochastic Itô–Volterra integral equation
Networks and Heterogeneous Media 2025, 20(5): 1524-1544
Published: 05 January 2026
Abstract PDF (520.7 KB) Collect
Downloads:33

Nonlinear stochastic Itô–Volterra integral equations (NSIVIEs) represent systems whose current state is influenced by random fluctuations and is dependent on previous information. These equations appear in many real-world scenarios, including engineering systems, biological processes, financial markets, heterogeneous media, complex transport phenomena, and viscoelastic materials. Strong numerical frameworks are required because analytical solutions for these equations are rarely available, particularly when nonlinearities and random fluctuations are present. To effectively solve NSIVIEs, in this study we propose a new hybrid numerical framework that combines Monte Carlo simulation and Legendre spectral collocation. By using orthogonal polynomial basis functions to approximate the solution, this method provides spectral accuracy while handling the hereditary memory component of the Volterra equation through a high-order Legendre spectral collocation method. A precise statistical treatment of the random fluctuations is made possible by simultaneously addressing the stochastic Itô noise through Monte Carlo sampling across numerous independent realizations. We perform a thorough convergence analysis and obtain explicit error bounds that measure the decrease in approximation error with increasing spectral resolution and Monte Carlo sample count. Numerical experiments show that the method can accurately reproduce complex stochastic behaviors and validate theoretical predictions.

Total 10