Volume 34, Issue 4

The quantitative analysis of recurrence plots while applied to mathematical models was shown to be an effective tool in recognizing a frontier between deterministic chaos and random processes. In nature, however, unlike mathematical models, deterministic processes are closely intertwined with random influences. As a result, the non-structural distributions of points on the recurrence plots, which are typical of random processes, are inevitably superimposed on the aperiodic structures characteristic of chaos. Taking into account that the stochastic impacts are an inherent feature of the dynamics of populations in the wild, we present here the results of the analysis of recurrence plots in order to reveal the extent to which irregular phytoplankton oscillations in the Naroch Lakes, Belarus, are susceptible to stochastic impacts. We demonstrate that numerical assessments of the horizon of predictability T pr of the dynamics under study and the average number P d of the points that belong to the diagonal segments on the recurrence plots can furnish insights into the extent to which the dynamics of both model and phytoplankton populations are affected by random components. Specifically, a comparative analysis of the values of T pr and P d for the time series of phytoplankton and the time series of random processes allows us to conclude that random components of the phytoplankton dynamics in the Naroch Lakes do not prevent recognition of chaotic nature of these dynamics.

The problem of modelling 2D isotropic turbulence in a periodic rectangular domain excited by the forcing pattern of prescribed spatial scale is considered. This setting could be viewed as the simplest analogue of the large scale quasi-2D circulation of the ocean and the atmosphere. Since the direct numerical simulation (DNS) of this problem is usually not possible due to the high computational costs we explore several possibilities to construct coarse approximation models and corresponding subgrid closures (deterministic or stochastic). The necessity of subgrid closures is especially important when the forcing scale is close to the cutoff scale of the coarse model that leads to the significant weakening of the inverse energy cascade and large scale component of the system dynamics. The construction of closures is based on the a priori analysis of the DNS solution and takes into account the form of a spatial approximation scheme used in a particular coarse model. We show that the statistics of a coarse model could be significantly improved provided a proper combination of deterministic and stochastic closures is used. As a result we are able to restore the shape of the energy spectra of the model. In addition the lagged auto correlations of the model solution as well as its sensitivity to external perturbations fit the characteristics of the DNS model much better.

In this research, we propose a deep learning based approach for speeding up the topology optimization methods. The problem we seek to solve is the layout problem. The main novelty of this work is to state the problem as an image segmentation task. We leverage the power of deep learning methods as the efficient pixel-wise image labeling technique to perform the topology optimization. We introduce convolutional encoder-decoder architecture and the overall approach of solving the above-described problem with high performance. The conducted experiments demonstrate the significant acceleration of the optimization process. The proposed approach has excellent generalization properties. We demonstrate the ability of the application of the proposed model to other problems. The successful results, as well as the drawbacks of the current method, are discussed.

The paper addresses a numerical approach for calculating three-fluid hydrodynamics on Eulerian grids with taking into account surface tension and viscous effects. The medium considered consists of three different compressible fluids separated with interfaces. The fluids are assumed to be immiscible. The three-fluid flow is described by the reduced equilibrium model derived from the non-equilibrium three-phase model by performing an asymptotic analysis in the limit of zero relaxation time. To simulate surface tension effects, we extend the continuum surface force (CSF) model of two-fluid incompressible flow to the case of compressible three-fluid flow. A thermodynamically consistent surface energy of the compressible three-fluid flow is obtained by means of splitting the surface tension between distinct fluids into pairs of specific phase related surface tensions. Some aspects of the numerical method for solving the system of governing equations of the considered three-fluid model are discussed. Numerical results presented demonstrate the accuracy and robustness of the proposed model in simulating dynamics of interfaces and surface tension effects.