Aleksejus Kononovicius

Senior researcher at Institute of Theoretical Physics and Astronomy (Faculty of Physics, Vilnius University). Research interests: stochastic and agent-based modeling, complex systems, Econophysics, Sociophysics, Statistical Physics and Data Science. Contributor to Physics of Risk science blog.

Highlights

Intro: My name is Aleksejus Kononovicius (lt. Kononovičius). I hold PhD in Physics and currently work as a senior researcher at Institute of Theoretical Physics and Astronomy (Faculty of Physics, Vilnius University). Besides conducting research in Econophysics and Sociophysics, which involves stochastic and agent-based modeling as well as a bit of Data Science, I blog about my research interests on Physics of Risk.

Recent papers:

• Delayed interactions in the noisy voter model through the periodic polling mechanism. Here we consider a variation of the noisy voter model in which interactions occur (or, peer pressure is exerted) only through the periodic public polls. We have shown that periodic polling mechanism has a stabilizing effect on the system (variance of the stationary opinion distribution decreases as polling period grows larger). When there is also a delay in announcing the poll outcome, the scaling behavior of stationary opinion distribution is non-trivial. Namely, in this case there is an optimal delay for which the stabilizing effect is strongest. Furthermore, this paper introduces a novel type of delay, which effectively freezes not the opinions of agents themselves (already explored in other works), but the perception of their peers opinions.
• 1/f noise from the sequence of nonoverlapping rectangular pulses. Typically models of 1/f noise in solid state matter involve point processes, or telegraph-like processes. E.g., electrical charge getting trapped for a period of time and then getting released. Here we have examined the conditions necessary to observe pure 1/f noise: gap duration must be power-law distributed, while pulse duration must be long in comparison to gap duration. Otherwise 1/f noise is perverted by logarithmic dependence. In comparison to earlier similar works we have shown that low frequency cutoff will still be present albeit at an extremely low frequency.
• Anomalous diffusion and long-range memory in the scaled voter model. Scaled Brownian motion is a Markovian model mimicking long-range memory properties of fractional Brownian motion, such as anomalous diffusion and first passage time distribution. Here we introduce scaled voter model, a generalization of voter model in which herding intensity parameter is a power-law function of time. This effectively changes stochastic differential equation of the process by replacing standard Brownian motion with scaled Brownian motion. We show that scaled voter model exhibits statistical signatures of long-range memory and analytically analyze them.

Contact me: ,