Transformation of alfvén into slow magnetosonic disturbances in the magnetosphere due to reflection from magnetically conjugate ionospheric regions

We perform numerical and analytical studies of MHD phenomena within the framework of a one-dimensional model of the magnetosphere bounded by conjugate reflection surfaces. Specific features of the nonlinear transformation of an Alfvén wave into a slow magnetosonic one upon propagation along the magnetic fieldB ₀ are considered. Effects of reflection of a nonlinear magnetospheric wave from magnetically conjugate ionospheres are analyzed in detail. It is shown that the main effect for the case of a nonlinear Alfvén wave in a low-temperature plasma is the plasma sweeping off the volumes close to the reflecting boundaries. Numerical results which demonstrate the generation dynamics of the slow magnetosonic wave and plasma sweeping upon reflection of a powerful Alfvén wave from the ionosphere are in qualitative agreement with the analytical estimates. In a cold plasma, the effect is so significant that it requires a numerical study even if the initial Alfvén wave is linear. The key role of the ratio between the Alfvén and sound velocities for the value of the density disturbance is revealed. Pedagogical University, Nizhny Novgorod, Russia. Translated from Izvestiya Vysshikh Uchebnykh Zavedenii, Radiofizika, Vol. 43, No. 4, pp. 285–295, April, 2000.     

Research system of decision making by composite image fragments using neuron-like algorithms

Base models of homogeneous neuron-like media and the corresponding systems of spatiotemporal solutions are considered. Ways of using these models and solutions as the base models for designing hierarchic systems of parallel processing of complex information flows with the help of neuron-like algorithms are studied. The version of a decision-making system with tunable architecture at a coarsegrain level (for adapting the system to the processing purposes and type of image) and universal neuron-like subsystems at a fine-grain level (for providing a wide variety of operations needed to extract local and global features and to encode individual fragments or the whole of the image) is developed. Examples of the transformation processes in the images of biological objects (growing neurons, erythrocytes, fingerprints) in a hierarchic decision-making system are demonstrated.Institute of Applied Physics, Russian Academy of Sciences. Translated from Izvestiya Vysshikh Uchebnykh Zavedenii, Radiofizika, Vol. 37, No. 8, pp. 961–986, August, 1994.     

Influence of the solar-wind magnetic field on the magnetosheath turbulence behind the bow shock

The experiment Sura–-WIND (1996–1997) on radio-raying of geodisturbed solar-wind region is interpreted in terms of modern knowledge of an interaction between the magnetized solar wind and the Earth's magnetosphere. Characteristics of the scattered signal at 9 MHz, determined by a plasma turbulence level with scales about 100 km, are statistically related to in situ measurements of solar wind parameters such as plasma density and the orientation and magnitude of the interplanetary magnetic field (IMF) onboard WIND spacecraft. The dependence of the scintillation index of the detected scattered signal, characterising the average turbulence level of the Earth's magnetosheath behind the bow shock, on the IMF orientation and magnitude is revealed. To verify the relation obtained, modern nonlinear correlation techniques based on the theory of artificial neural networks (ANN) are applied. The results obtained using a three-layer ANN with error backpropagation confirm an essential IMF influence on the plasma turbulence in the magnetosheath.     

Visual image processing by neural networks with nonlocal coupling

The possibilities of using homogeneous artificial neural networks both for encoding and back reconstruction of visual images are investigated. A conjugate reverse neural network with an opposite time direction is used for the image reconstruction. Information-entropic characteristics of the image and its different-level histograms are used for quality monitoring of the dynamic image processing. Experimental data on the dynamic behavior of these values in computer simulation for pure images are reported. Comparison with the thermodynamic entropy, subject to a monotonic change, is made. Special cases of nonmonotonic entropy change are considered and examples of the dynamic processing of images with code redundancy are demonstrated.Radiophysical Research Institute. Translated from Izvestiya Vysshikh Uchebnykh Zavedenii, Radiofizika, Vol. 37, No. 9, pp. 1195–1213, September, 1994.