Simulating multilevel quantum systems using the density matrix formalism involves solving large systems of coupled differential equations, often requiring computationally intensive methods such as Runge–Kutta, Monte Carlo, or physics-informed neural networks. These solvers demand high-performance computing resources because of the complexity and size of the underlying matrices. This study evaluates key aspects of simulation performance, such as numerical accuracy, scalability, parallel efficiency, or resource utilization, to develop efficient computational strategies. To address infrastructure overhead and improve accessibility, a Function-as-a-Service-based platform is proposed to enable elastic, on-demand execution of quantum simulation workflows. A monitoring and orchestration strategy is integrated into the platform to optimize resource use across varying problem dimensionalities. By abstracting the complexity of traditional high-performance computing environments, the proposed Function-as-a-Service-based approach facilitates efficient and scalable quantum simulations while reducing user-side infrastructure management. The results provide a foundation for designing accessible, high-fidelity simulations for complex quantum systems on modern cloud-based or hybrid high-performance computing platforms.
This article presents a comprehensive analysis of the long-term economic growth trajectory in Armenia. The research utilizes the Long-Term Growth Model (LTGM) to assess the impact of total factor productivity (TFP), physical capital, and human capital on per capita GDP dynamics through 2100. The results of scenario-based modeling made it possible to identify the projected decline in the working-age population as the key long-term constraint on growth. However, the existing levels of physical capital and the quality of human capital are deemed sufficient to sustain economic expansion. The calculations indicate that under a pessimistic TFP growth scenario, per capita GDP growth could decline to 0.5
An interval t-coloring of a graph G is a proper edge-coloring with colors 1,… ,t such that the colors on the edges incident to every vertex of G are colored by consecutive colors. A graph G is called interval colorable if it has an interval t-coloring for some positive integer t. Let 𝔑 be the set of all interval colorable graphs. For a graph G∈𝔑 , we denote by w(G) and W(G) the minimum and maximum number of colors in an interval coloring of a graph G, respectively. In this paper we present some new sharp bounds on W(G□ H) for graphs G and H satisfying various conditions. In particular, we show that if G,H∈𝔑 and H is an r-regular ( r∈ℕ ) graph, then W(G□ H)≥ W(G)+W(H)+r . We also derive a new upper bound on W(G) for interval colorable connected graphs with additional distance conditions. Based on these bounds, we improve known lower and upper bounds on W(C_2n_1□ C_2n_2□⋯□ C_2n_k) for k-dimensional tori C_2n_1□ C_2n_2□⋯□ C_2n_k and on W(K_2n_1□ K_2n_2□⋯□ K_2n_k) for Hamming graphs K_2n_1□ K_2n_2□⋯□ K_2n_k , and these new bounds coincide with each other for hypercubes. Finally, we give several results on interval colorings of Fibonacci cubes _n.
Speech Emotion Recognition in Russian remains underexplored compared to English, largely due to limited multimodal resources, heterogeneous annotation schemes, and the absence of strong baselines across datasets. To address these challenges, we propose a unified bimodal audio-text architecture that integrates speech representations with text embeddings and a cross-modal Co-Attention fusion module. Central to the framework is an auxiliary prototype-based objective that introduces trainable class-specific emotion prototypes and aligns acoustic embeddings with these anchors through cosine-based classification. While prototype-based representation learning has been explored previously in emotion recognition, we adapt this formulation to multimodal Russian SER and investigate its effectiveness as a stable auxiliary supervision mechanism for heterogeneous low-resource datasets. This objective provides an additional structural signal for the audio encoder without relying on batch-dependent pair construction. We evaluate the system on four Russian SER corpora, corresponding to five evaluation subsets: RESD, REPV, SCES-1, Dusha Crowd, and Dusha Podcast, covering both acted and spontaneous speech from adult and child speakers. Extensive experiments demonstrate that the proposed model achieves strong and consistent improvements over existing baselines, with multimodal fusion yielding substantial gains on datasets containing informative transcripts, whereas the audio-only variant remains competitive on lexically constrained corpora. Ablation studies further highlight the effectiveness of the auxiliary prototype-based objective. Overall, this work provides one of the first systematic multimodal evaluations across multiple Russian SER datasets and demonstrates the effectiveness of prototype-guided auxiliary supervision in this setting.
This paper proposes a numerical approach to the problem of inverse quantum control with a focus on the automated formation of external laser control fields that ensure high-precision and controlled population transfer in multilevel quantum systems. The method is based on optimizing the parameters of control pulses using gradient algorithms in combination with automatic differentiation, which allows for effective control of the physical characteristics of laser fields and increases the stability of the transfer process. The obtained results provide a foundation for the development of more general quantum control methods focused on experimental implementation on real physical platforms.
