Vladimir State University

Waste highly-oxidized Nd–Fe–B-based permanent magnets were subjected to milling and subsequent treatment with thermal plasma jet. X-ray diffraction (XRD) analysis, scanning electron microscopy (SEM), and energy dispersive analysis are used to characterize the phase composition of powder particles solidified after the treatment. The separation of heavy and light rare-earth metal constituents of a permanent magnet alloy is demonstrated. The treatment with thermal plasma jet followed by rapid cooling during solidification of powder particles can be suggested as possible recycling process of end-of-life permanent magnets.

This paper considers an interactive LLM-based agent designed to construct a formal JSON model of a target system from a multi-turn dialogue with a user. For such an agent, we formulate a context management problem with an explicit separation between the textual dialogue history and the structured system state. We propose a cost-aware algorithm that automatically decides when to compress a part of the dialogue into an updated JSON state and then rely on this state instead of repeatedly sending a growing history fragment. The key idea is to compare the accumulated cost of transmitting the uncompressed dialogue segment with the estimated cost of a single state-update step, including both input and output tokens; an update is performed only when this ratio becomes economically favorable. The algorithm is implemented in a prototype agent based on the GigaChat model and evaluated on three modeling scenarios: client flows in an online school, financial flows in machine manufacturing, and a predator–prey population model. Across dialogues of length 20, 32, and 50 turns, the experiments demonstrate a reduction in total token consumption from 19.63 % to 59.54 % compared to a non-optimized baseline, while preserving the dialogue trajectory and perceived answer quality.

This paper presents a scalable approach to inter-module communication in high-complexity digital signal processing (DSP) systems based on encapsulating payload data and configuration parameters into unified atomic packets. The proposed approach employs specialized hardware wrappers and flow-control mechanisms to enable the construction of flexible processing pipelines while ensuring synchronization between data and configuration. The solution decouples computational cores from variable-latency pipelines, thereby unifying the integration of heterogeneous processing blocks. The feasibility and efficiency of the proposed approach are validated through the implementation of a Long Term Evolution (LTE) baseband physical-layer subsystem and experimental hardware verification on a software-defined radio (SDR) platform.

The use of bimetallic materials composed of copper alloys and steel offers operational and economic advantages. Sintering copper powder onto a steel substrate serves as a relatively simple and effective method for producing such bimetallic materials, with the addition of tin and lead during sintering enhancing anti-friction properties. However, a key limitation of this method is the relatively low hardness of the copper alloy layer. Moreover, lead has a tendency to form large inclusions at the grain boundaries of copper, which can act as stress concentrators and initiate cracks, ultimately leading to material failure. Wave deformation hardening (WDH) has been identified as a strategy to address this issue. WDH has shown to be effective for hardening metallic materials at room temperature and when preheated. In this study, WDH was applied to copper alloy coatings on steel substrates in bimetallic samples heated to various temperatures, with the results showing that WDH significantly increased the hardness of the copper alloy layer. The size of large lead inclusions in the copper alloy microstructure was also reduced, while the number of smaller inclusions increased, resulting in a more uniform distribution. The most favorable results were obtained at a WDH temperature of 300°C, which produced more than a threefold increase in hardening compared to room-temperature treatment. WDH effectively improves the mechanical properties of bimetallic materials.

The article is devoted to the analysis and systematization of pedagogical experience in the formation of engineering thinking of students. The author suggests a typology of educational practices, including contextual, design, research, maker, and digital formats. It is proved that bridging the gap between educational tasks and real engineering activities is possible through the creation of a single educational ecosystem integrating these approaches.