Working Visit to the Scientific Computing Laboratory of the Institute of Physics Belgrade

From November 24 to 28, 2025, staff from the Laboratory of Topological Methods in Dynamics at HSE University Nizhny Novgorod visited the Scientific Computing Laboratory of the Institute of Physics Belgrade, National Institute of the Republic of Serbia. The visit was organized in the framework of the project “International Academic Cooperation” HSE University on the topic "Machine Learning and Nonlinear Dynamics: Intersection, Interplay, and Synthesis."

The program of the visit included meetings at the Institute of Physics in Belgrade, introductions to the research team and other institute groups, discussions of joint research projects, a joint scientific and educational seminar "Machine Learning and Nonlinear Dynamics: Intersection, Interplay, and Synthesis," and a series of presentations at a research seminar at the Scientific Computing Laboratory. On the final day of the visit, the results were summarized, having discussed future plans for collaboration.

Prof. Igor Franović

— I think that investigation of the learning dynamics of saddle-to-saddle type may be quite important for an interesting synthesis between machine learning and heteroclinic-type dynamics suggested to shape cognitive brain processes. I also believe that the research on pancreatic beta cells holds great promise, especially in the context of explaining the onset of diabetes, which is now a hot topic.

During the working visit, staff from the Laboratory of Topological Methods in Dynamics presented their research results on the project's topic and made the following talks:

• N.V. Stankevich - "Multistability in Dynamic Systems. Prediction of Multistability with Machine Learning";
• V.V. Klinshov - "Extreme Multistability and Complex Landscape of the Cost Function in a Simple Machine Learning Task";
• E.R. Bagautdinova - "Comparative Analysis of Multistable Properties of the Hodgkin–Huxley Model and Its Neural Network Map";
• T.I. Miakov - "Neural Network Maps for Different Dynamic Systems."

The works sparked heated discussion, leading to two joint research projects on the impact of noise on bistable models and neural network maps, planned for implementation in the next year.

During free time, the colleagues took a stroll around Belgrade and discovered a number of striking similarities between Belgrade and Nizhny Novgorod. Both cities are located at the confluence of two major rivers—the Oka and Volga in the Russian metropolis, and the Danube and Sava in the Serbian capital—and also share hilly historic centers with fortresses (the Nizhny Novgorod Kremlin and Kalemegdan in Belgrade). Furthermore, the cities share similar infrastructure: similar bridges, the division to the old historical city center and the new modern parts of the city, as well as the No. 2 circular tram line.

As part of the cultural program, the colleagues visited the Nikola Tesla Museum, which preserves the memory of the great scientist and displays his numerous inventions and scientific works. The visit provided an opportunity to delve deeper into the life and scientific achievements of the great Serbian inventor, whose discoveries literally lit the world and significantly influenced the development of modern science and technology.

The working visit took place in a friendly atmosphere, enabled productive planning of future work, and contributed to strengthening cooperation between HSE University and the Institute of Physics in Belgrade, as well as expanding joint research in the fields of nonlinear dynamics and machine learning.