Search Results (47)

The present work is devoted to the analysis of a time-fractional Gardner equation arising in the modeling of nonlinear plasma waves in media endowed with memory and anomalous transport effects. Building on a physically motivated soliton profile, we construct a finite-time fractional ansatz in which the integer-order time variable is replaced by a fractional reparametrization that encodes the Caputo memory kernel. Within this framework, the governing evolution equation is not treated via a formal infinite expansion but rather via a finite approximation, whose quality is assessed directly via the associated residual. The Caputo fractional derivative is evaluated by a strong finite-difference formula that is second-order accurate in time and preserves the nonlocal convolution structure of the fractional operator. This combination of a finite fractional ansatz and a strong Caputo discretization allows us to compute the residual of the time analytically fractional Gardner equation and to use it as a quantitative diagnostic of accuracy and consistency. Two representative classes of nonlinear structures supported by the Gardner equation are examined in detail: a smooth solitary-wave profile and a cnoidal-wave configuration. For each example, the approximate fractional solution is generated, the corresponding residual is evaluated in space–time, and global and final-time residual norms are determined to quantify the influence of the fractional order on the wave dynamics and on the quality of the approximation. The numerical results show that the proposed residual-controlled approach yields residual magnitudes that remain one to two orders of magnitude smaller than those associated with truncated residual power-series approximations constructed from the same data, while preserving the expected qualitative features of fractional solitary and cnoidal waves in non-Markovian plasma environments. Full article

Low-power NPUs enable on-device LLM inference through efficient integer and fixed-point algebra, yet their lack of native exponential support makes Transformer softmax a critical performance bottleneck. Existing NPU kernels approximate $e^x$ using uniform piecewise polynomials to enable O(1) SIMD indexing, but this wastes computation by applying high-degree arithmetic indiscriminately in every segment. Conversely, fully adaptive approaches maximize statistical fidelity but introduce pipeline stalls due to comparator-based boundary search. To bridge this gap, we propose an attention distribution-aware softmax that uses Particle Swarm Optimization (PSO) to define non-uniform segments and variable polynomial degrees, prioritizing finer granularity and lower arithmetic complexity in attention-dense regions. To ensure efficiency, we snap boundaries into a 128-bin LUT, enabling O(1) retrieval of segment parameters without branching. Inference measurements show that this favors low-degree execution, minimizing exp-kernel overhead. Using TinyLlama-1.1B-Chat as a testbed, the proposed weighted design reduces cycles per call exp kernel (CPC) by 18.5% versus an equidistant uniform Degree-4 baseline and 13.1% versus uniform Degree-3, while preserving ranking fidelity. These results show that grid-snapped, variable-degree approximation can improve softmax efficiency while largely preserving attention ranking fidelity, enabling accurate edge LLM inference. Full article

We introduce the discrete asymmetric double Lindley distribution, a new two-parameter family on the integer line designed to model signed counts and net changes with flexible asymmetric tail behavior. This statistical model is obtained by merging two Lindley-type linear-geometric kernels on the negative and non-negative half-lines, with tail decay rates that are coupled through a simple two-parameter mechanism. This construction yields an analytically tractable probability mass function with an explicit normalizing constant, as well as closed-form expressions for the cumulative distribution function and one-sided tail probabilities. We further provide a transparent stochastic representation based solely on Bernoulli and geometric random variables, leading to an exact and efficient simulation algorithm that is convenient for Monte Carlo studies and validating numerical likelihood routines. Graphical illustrations highlight the role of the asymmetry parameter in controlling the imbalance between the two tails and the resulting skewness on $\mathbb{Z}$. The proposed family offers a practical and interpretable alternative to existing integer-line models for asymmetric discrete data, with direct applicability to likelihood-based inference and real-world datasets. Full article

Neural field dynamics in the cerebral cortex exhibit complex spatiotemporal patterns inadequately captured by classical integer-order diffusion models that assume exponentially decaying spatial interactions. This study establishes a stochastic fractional FitzHugh–Nagumo framework incorporating power-law spatial correlations through fractional Laplacian operators, providing explicit parameterization of non-local cortical connectivity characteristics. The inverse problem of estimating fractional orders and model parameters from electroencephalographic data is addressed through multi-objective optimization with rigorous train–test validation. Systematic sensitivity analysis across the parameter space $({\alpha }_u,{\alpha }_v)\in [1.0,2.0]\times [1.0,2.0]$ identifies optimal subdiffusive characteristics at ${\alpha }_u={\alpha }_v=1.5$, corresponding to power-law spatial kernels $C\left(x\right)\sim {\left|x\right|}^{-1.5}$ consistent with anatomical connectivity measurements. The optimized model achieves out-of-sample performance $R^2=0.973$ on held-out test data, approaching the measurement noise ceiling. While classical FitzHugh–Nagumo models achieve comparable test accuracy, the fractional framework provides enhanced interpretability through explicit spatial interaction parameterization. The fractional orders serve as quantitative biomarkers of cortical network organization, enabling data-driven characterization across brain states and neurological conditions. The methodology establishes computational foundations for clinical applications in epilepsy monitoring, neurodegenerative disease detection, and brain–computer interfaces. Full article

In this study, nonlinear fractional Korteweg–de Vries (KdV) type equations with nonlocal operators are studied using Mittag–Leffler kernels and exponential decay. The KdV equations are well known for its use in modeling ion-acoustic waves in plasma, oceanic dynamics, and shallow-water waves. As a result, mathematicians are working to examine modified and generalized versions of the basic KdV equation. In order to find the solutions of nonlinear fractional KdV equations, an extension of this concept is described in the current paper. The solution of fractional KdV equations is carried out using the well-known natural transform decomposition method (NTDM). To evaluate the problem, we employ the fractional operator in the Caputo–Fabrizio (CF) and the Atangana–Baleanu–Caputo sense (ABC) manner. Nonlinear terms can be handled with Adomian polynomials. The main advantage of this novel approach is that it might offer an approximate solution in the form of convergent series using easy calculations. The dynamical behavior of the resulting solutions have been demonstrated using graphs. Numerical data is represented visually in the tables. The solutions at various fractional orders are found and it is proved that they all tend to an integer-order solution. Additionally, we examine our findings with those of the iterative transform method (ITM) and the residual power series transform method (RPSTM). It is evident from the comparison that our approach offers better outcomes compared to other approaches. The results of the suggested method are very accurate and give helpful details on the real dynamics of each issue. The present technique can be expanded to address other significant fractional order problems due to its straightforward implementation. Full article

This study employs the Atangana–Baleanu–Caputo fractional derivative within the Moore–Gibson–Thompson heat conduction model to analytically investigate the thermoelastic vibrations in solid medium-containing voids. The ABC–MGT formulation incorporates a non-singular Mittag–Leffler memory kernel, facilitating the modeling of tempered hereditary relaxation in voided thermoelastic media, thereby producing more realistic attenuation and phase lag characteristics in transient responses than conventional integer-order models. Specifically, our novelty lies in developing a coupled thermoelastic–void formulation within an ABC–MGT heat conduction framework, deriving the full governing system and boundary-value solution in the Laplace domain, and providing a systematic parametric analysis showing how the ABC order changes attenuation, phase lag, and stress/void interactions. This approach enables a precise analytical resolution of the problem. The analysis indicates that the presence and size of voids substantially impact the system response variables, with smaller apertures yielding reduced magnitudes. Thus, this analytical investigation introduces a novel methodology for addressing the complex challenges associated with advanced functional materials and high-performance engineering structures. Full article

Conventional sensing expends energy at three stages: powering dedicated sensors, transmitting measurements, and executing computationally intensive inference. Wireless sensing re-purposes WiFi channel state information (CSI) inherent in every packet, eliminating extra sensors and uplink traffic, though reliance on deep neural networks (DNNs) often trained and run on graphics processing units (GPUs) can negate these gains. This review highlights two core energy efficiency levers in CSI-based wireless sensing. First ambient CSI harvesting cuts power use by an order of magnitude compared to radar and active Internet of Things (IoT) sensors. Second, integrated sensing and communication (ISAC) embeds sensing functionality into existing WiFi links, thereby reducing device count, battery waste, and carbon impact. We review conventional handcrafted and accuracy-first methods to set the stage for surveying green learning strategies and lightweight learning techniques, including compact hybrid neural architectures, pruning, knowledge distillation, quantisation, and semi-supervised training that preserve accuracy while reducing model size and memory footprint. We also discuss hardware co-design from low-power microcontrollers to edge application-specific integrated circuits (ASICs) and WiFi firmware extensions that align computation with platform constraints. Finally, we identify open challenges in domain-robust compression, multi-antenna calibration, energy-proportionate model scaling, and standardised joules per inference metrics. Our aim is a practical battery-friendly wireless sensing stack ready for smart home and 6G era deployments. Full article

The spread of computer viruses poses a critical threat to networked systems and requires accurate modeling tools. Classical integer-order approaches had failed to capture memory effects inherent in real digital environments. To address this, we developed a four-compartment fractional-order model using the Atangana–Baleanu–Caputo (ABC) derivative with Mittag-Leffler kernels. We established fundamental properties such as positivity, boundedness, existence, uniqueness, and Hyers–Ulam stability. Analytical solutions were derived via Laplace transform and homotopy series, while the Variation-of-Parameters Method and a dedicated numerical scheme provided approximations. Simulation results showed that the fractional order strongly influenced infection dynamics: smaller orders delayed peaks, prolonged latency, and slowed recovery. Compared to classical models, the ABC framework captured realistic memory-dependent behavior, offering valuable insights for designing timely and effective cybersecurity interventions. The model exhibits structural symmetries: the infection flux depends on the symmetric combination $L+I$ and the feasible region (probability simplex) is invariant under the flow. Under the parameter constraint $\delta =\theta$ (and equal linear loss terms), the system is equivariant under the involution $(L,I)\mapsto (I,L)$, which is reflected in identical Hyers–Ulam stability bounds for the latent and infectious components. Full article

Continuous multi-day extremely low or high new energy outputs have posed significant challenges in relation to power supply and new energy accommodations. Conventional reservoir hydropower, with the advantage of controllability and the storage ability of reservoirs, can represent a reliable and low-carbon flexibility resource to safeguard against continuous extreme new energy fluctuations. This paper proposes a mid-term scheduling method for reservoir hydropower to enhance our ability to regulate continuous extreme new energy fluctuations. First, a data-driven scenario generation method is proposed to characterize the continuous extreme new energy output by combining kernel density estimation, Monte Carlo sampling, and the synchronized backward reduction method. Second, a two-stage stochastic hydropower–new energy complementary optimization scheduling model is constructed with the reservoir water level as the decision variable, ensuring that reservoirs have a sufficient water buffering capacity to free up transmission channels for continuous extremely high new energy outputs and sufficient water energy storage to compensate for continuous extremely low new energy outputs. Third, the mathematical model is transformed into a tractable mixed-integer linear programming (MILP) problem by using piecewise linear and triangular interpolation techniques on the solution, reducing the solution complexity. Finally, a case study of a hydropower–PV station in a river basin is conducted to demonstrate that the proposed model can effectively enhance hydropower’s regulation ability, to mitigate continuous extreme PV outputs, thereby improving power supply reliability in this hybrid renewable energy system. Full article

Traditional operational calculus, while intuitive and effective in addressing problems in physical fractal spaces, often lacks the rigorous mathematical foundation needed for fractional operations, sometimes resulting in inconsistent outcomes. To address these challenges, we have developed a universal framework for defining the fractional calculus operators using the generalized fractional calculus with the Sonine kernel. In this framework, we prove that the α-th power of a differential operator corresponds precisely to the α-th fractional derivative, ensuring both accuracy and consistency. The relationship between the fractional power operators and fractional calculus is not arbitrary, it must be determined by the specific operator form and the initial conditions. Furthermore, we provide operator representations of commonly used fractional derivatives and illustrate their applications with examples of fractional power operators in physical fractal spaces. A superposition principle is also introduced to simplify fractional differential equations with non-integer exponents by transforming them into zero-initial-condition problems. This framework offers new insights into the commutative properties of fractional calculus operators and their relevance in the study of fractal structures. Full article

This paper explores a general class of singular kernels with the objective of designing new families of uniformly continuous sliding mode controllers. The proposed controller results from filtering a discontinuous switching function by means of a Sonine integral, producing a uniformly continuous control signal, preserving finite-time sliding motion and robustness against continuous but unknown and not necessarily integer-order differentiable disturbances. The principle of dynamic memory resetting is considered to demonstrate finite-time stability. A set of sufficient conditions to design singular kernels, preserving the above characteristics, is presented, and several examples are exposed to propose new families of continuous sliding mode approaches. Simulation results are studied to illustrate the feasibility of some of the proposed schemes. Full article

The principal aim of this paper is to introduce the extended Voigt-type function ${\mathbb{V}}_{\mu ,\nu }(x,y)$ and its counterpart extension ${\mathbb{W}}_{\mu ,\nu }(x,y)$, involving the Neumann function $Y_{\nu }$ in the kernel of the representing integral. The newly defined integral reduces to the classical Voigt functions $K(x,y)$ and $L(x,y)$, and to their generalizations by Srivastava and Miller, by the unification of Klusch. Following an approach by Srivastava and Pogány, we also present the multiparameter and multivariable versions ${\mathbb{V}}_{\mu ,\nu }^{\left(r\right)}(x,y),{\mathbb{W}}_{\mu ,\nu }^{\left(r\right)}(x,y)$ and the r positive integer of the initial extensions ${\mathbb{V}}_{\mu ,\nu }(x,y),{\mathbb{W}}_{\mu ,\nu }(x,y)$. Several computable series expansions are obtained for the discussed Voigt-type functions in terms of Humbert confluent hypergeometric functions ${\Psi }_2^{\left(r\right)}$. Furthermore, by transforming the input extended Voigt-type functions by the Grünwald–Letnikov fractional derivative, we establish representation formulae in terms of the associated Legendre functions of the second kind $Q_{\eta }^{-\nu }$ in the two-parameter and two-variable cases. Finally, functional bounding inequalities are given for ${\mathbb{V}}_{\mu ,\nu }(x,y)$ and ${\mathbb{W}}_{\mu ,\nu }(x,y)$. Particularly interesting results are presented for the Neumann function $Y_{\nu }$ and for the Struve ${\mathbf{H}}_{\nu }$ function in the form of several functional bounds. The article ends with a thorough discussion and closing remarks. Full article

In this paper, by replacing the exponential memory kernel function of a tabu learning single-neuron model with the power-law memory kernel function, a novel Caputo’s fractional-order tabu learning single-neuron model and a network of two interacting fractional-order tabu learning neurons are constructed firstly. Different from the integer-order tabu learning model, the order of the fractional-order derivative is used to measure the neuron’s memory decay rate and then the stabilities of the models are evaluated by the eigenvalues of the Jacobian matrix at the equilibrium point of the fractional-order models. By choosing the memory decay rate (or the order of the fractional-order derivative) as the bifurcation parameter, it is proved that Hopf bifurcation occurs in the fractional-order tabu learning single-neuron model where the value of bifurcation point in the fractional-order model is smaller than the integer-order model’s. By numerical simulations, it is shown that the fractional-order network with a lower memory decay rate is capable of producing tangent bifurcation as the learning rate increases from 0 to 0.4. When the learning rate is fixed and the memory decay increases, the fractional-order network enters into frequency synchronization firstly and then enters into amplitude synchronization. During the synchronization process, the oscillation frequency of the fractional-order tabu learning two-neuron network increases with an increase in the memory decay rate. This implies that the higher the memory decay rate of neurons, the higher the learning frequency will be. Full article

Gaussian filtering, being a convolution with a Gaussian kernel, is a widespread technique in image analysis and computer vision applications. It is the traditional approach for noise reduction. In some cases, performing the exact convolution can be computationally expensive and time-consuming. To address this problem, approximations of the convolution are often used to achieve a balance between accuracy and computational efficiency, such as with running sums, Bell blur, Deriche approximation, etc. At the same time, modern computing devices support data parallelism (vectorization) via Single Instruction Multiple Data (SIMD) and can process integer numbers faster than floating-point approaches. In this paper, we describe several methods for approximating a Gaussian filter, implement the SIMD and quantized versions, and compare them in terms of speed and accuracy. The experiments were performed on central processing units with a x86_64 architecture using a family of SSE SIMD extensions and an ARMv8 architecture using the NEON SIMD extension. All the optimized approximations demonstrated 10–20× speedup while maintaining the accuracy in the range of 1 × ${10}^{-5}$ or higher. The fastest method is a trivial Stack blur with a relatively high error, so we recommend using the second-order Vliet–Young–Verbeek filter and quantized Bell blur and running sums as more accurate and still computationally efficient alternatives. Full article

This paper focuses on examining numerical solutions for fractional-order models within the context of the coupled multi-space Korteweg-de Vries problem (CMSKDV). Different types of kernels, including Liouville-Caputo fractional derivative, as well as Caputo-Fabrizio and Atangana-Baleanu fractional derivatives, are utilized in the examination. For this purpose, the nonstandard finite difference method and spectral collocation method with the properties of the Shifted Vieta-Lucas orthogonal polynomials are employed for converting these models into a system of algebraic equations. The Newton-Raphson technique is then applied to solve these algebraic equations. Since there is no exact solution for non-integer order, we use the absolute two-step error to verify the accuracy of the proposed numerical results. Full article