
This paper proposes methods for obtaining sufficient conditions for asymptotic stability and instability for systems of differential equations containing linear delay. By changing the argument, such systems are reduced to systems with constant delay, which, however, contain an exponential factor on the right-hand side, i.e., are not Lipschitz continuous. Based on these conditions, several systems of linear differential equations are studied, and for one of them, an algorithm for stabilization over an infinite time interval is proposed. It is shown that similar sufficient conditions are also valid for singularly perturbed systems with constant delay and a small parameter at the derivative. Based on the obtained asymptotic stability conditions, an algorithm for stabilizing certain systems with linear delay is proposed. This algorithm can also be used for calming the singular systems under study. Sufficient conditions for asymptotic stability are obtained for systems of differential equations containing linear delay. Systems of this type are encountered in problems in mechanics, physics, biology, economics, and queuing theory. Based on the stability and instability conditions obtained by the authors, the asymptotic behavior of certain systems of linear differential equations is studied, with one of them stabilized over an infinite time interval.
A new hp-version of the least-squares collocation method (hp-LSCM) is proposed, implemented, and verified for the numerical solution of elliptic problems with second-order highest derivatives on refined triangular grids generated in Gmsh. The method employs approximation spaces of polynomial degrees p = 2, 3, 5, 6, 8. The unknown coefficients are obtained by solving overdetermined sparse systems of linear algebraic equations using an orthogonal factorization method provided by the SuiteSparse library, combined with CUDA-based parallelization. The developed hp-LSCM is tested on the Dirichlet problem for the Poisson equation in a square domain and on the bending of a clamped annular plate within the Reissner-Mindlin theory. Solutions with large gradients and limited smoothness are considered. A comparison is made with the p = 4 version of the collocation method (based on the LSCM), which leads to a square system matrix. Additionally, the conditioning and accuracy of the solutions are analyzed with respect to the number of equations in the approximate problem and the grid refinement parameter.
A model of local saturation of a "hot" viscous incompressible fluid under the action of a constant pressure gradient from the "free" surface of a "cold" porous matrix is presented using the Darcy-Brinkman equations and a single-temperature energy formulation. It is shown that for laminar pore flow, the velocity is virtually constant, with the exception of an insignificant initial time interval. This allowed us to apply a hydrodynamic idealization of displacement and consider a non-conjugate thermal initial-boundary value subproblem for a semi-bounded isotropic fine-grained porous medium, replacing the infinitely extending "free" surface with a bounded region whose scale significantly exceeds the local saturation zone. An analytical solution is obtained by combining the one-sided integral Laplace transform and the finite integral cosine Fourier transform. A calculation examples is presented quantitatively characterizing the migration of temperature inhomogeneity in a porous layer
This article examines the development of an optical character recognition (OCR) system for the Mansi language, a resource-poor Finno-Ugric language with a unique set of diacritics. The primary objective of the study is to adapt existing OCR technologies to the specifics of the Mansi script, which is characterized by a limited volume of digitized texts and the presence of specific graphic symbols. To address this challenge, a comprehensive approach was developed, including generating an extensive synthetic dataset taking into account font variability and Unicode normalization, further training the Tesseract 5 model using transfer learning based on a pre-trained Russian-language model, and evaluating recognition quality using the CER and WER metrics. The resulting specialized model demonstrated a CER value of 0,85%, which is twenty times higher than the baseline model (18,5%). The developed model is implemented as a public web service and is openly accessible, enabling the automated digitization of printed sources in the Mansi language and facilitating the preservation of the cultural heritage of the indigenous peoples of the North.