Vol 15, No 3 (2024)
Simulation the response of graphene to an external electric field using the exact tight-binding model
Abstract
Numerical simulation of the interaction of electromagnetic radiation with graphene allows us to reproduce fast nonlinear processes and their observed manifestations. The paper presents the results obtained in the process of developing a software solution for calculating the observed parameters of such processes.In graphene physics, the massless fermion approximation is classical. However, in the study of processes with high energy density, model based on this approximation are beyond the limits of their applicability and the results obtained on their basis can not be considered reliable. To solve this problem, a transition to a substantially more accurate model based on a strict account of the nearest-neighbor interaction in the crystal lattice (tight-binding model) has been made.Comparative testing of these two models shows that at low energy characteristics of the external perturbation the results coincide. However, as the energy characteristics of the affecting electromagnetic field increase, the divergence of the results becomes apparent and grows.The new exact model has a more complex mathematical formulation and requires more computational resources. When using the same hardware configuration it is expressed in the increase of counting time. Relative and absolute values for a number of examples are given.The obtained results allow us to expand the range of parameters for modeling of nonlinear processes in the considered material, for example, generation of high-frequency harmonics and ensure its reliability.
Program Systems: Theory and Applications. 2024;15(3):3-22
3-22
Mathematical modeling and research of the optimal configuration of an optical stereo system consisting of two flat mirrors
Abstract
The paper is devoted to mathematical modeling and optimization of optical stereo system configuration, consists of video camera and two flat mirrors. The difference between this research and previous researches is the consideration of a large number of restrictions on the configuration of the optical system: the size of the stereo base, the size of the mirrors, overall dimensions of the optical system, the absence of double reflection of light rays, preventing the situation when the video camera is reflected in the mirrors. A conditional optimization problem is formulated to find the optimal configuration of the considered optical system. The perimeter of the rectangle limiting the dimensions of the optical system was chosen as the target function. Numerical solution to the problem was found using the SciPy package. The results obtained expand the theory of computer vision and can be used in the creation and research of computer vision systems for robotic systems.
Program Systems: Theory and Applications. 2024;15(3):23-52
23-52
Application of Siamese neural networks to classify plant biomass by visual state
Abstract
This paper proposes a method for classifying plant biomass by visual condition using images captured in a specially designed greenhouse and Siamese architecture artificial neural network technologies. Criteria for various states of plant biomass have been determined. We have generated our own dataset for training Siamese neural networks, containing samples of biomass states in the form of textures. As a result, a training accuracy of 91.6% and an average classification accuracy of individual biomass states of 73.6%.
Program Systems: Theory and Applications. 2024;15(3):53-74
53-74
Recovering text sequences using deep learning models
Abstract
This article presents the results of the formation, training and performance evaluation of models with the Encoder-Decoder and Sequence-To-Sequence (Seq2Seq) architectures for solving the problem of supplementing incomplete texts. Problems of this type often arise when restoring the contents of documents from their low-quality images. The studies conducted in the work are aimed at solving the practical problem of forming electronic copies of scanned documents of the «Roskadastr» PLC, the recognition of which is difficult or impossible with standard means.The formation and study of models was carried out in Python using the high-level API of the Keras package. A dataset consisting of several thousand pairs was formed for the purpose of training and studying the models. Each pair in this set represented an incomplete and corresponding full text. To evaluate the quality of the models, the values of the loss function and the accuracy, BLEU and ROUGE-L metrics were calculated. Loss and accuracy made it possible to evaluate the effectiveness of the models at the level of predicting individual words. The BLEU and ROUGE-L metrics were used to evaluate the similarity between the full and reconstructed texts. The results showed that both the Encoder-Decoder and Seq2Seq models cope with the task of reconstructing text sequences from their fixed set, but the Seq2Seq transformer-based model achieves better results in terms of training speed and quality.
Program Systems: Theory and Applications. 2024;15(3):75-110
75-110