Ferroelectric crystallites embedded into a dielectric matrix experience temperature-dependent elastic strains caused by differences in the thermal expansion of the crystallites and the matrix. Owing to the electrostriction, these lattice strains may affect polarization states of ferroelectric inclusions significantly, making them different from those of a stress-free bulk crystal. Here, using a nonlinear thermodynamic theory, we study the mechanical effect of elastic matrix on the phase states of embedded single-domain ferroelectric nanocrystals. Their equilibrium polarization states are determined by minimizing a special thermodynamic potential that describes the energetics of an ellipsoidal ferroelectric inclusion surrounded by a linear elastic medium. To demonstrate the stability ranges of such states for a given material combination, we construct a phase diagram, where the inclusion's shape anisotropy and temperature are used as two parameters. The 'shape-temperature' phase diagrams are calculated numerically for PbTiO and BaTiO nanocrystals embedded into representative dielectric matrices generating tensile (silica glass) or compressive (potassium silicate glass) thermal stresses inside ferroelectric inclusions. The developed phase maps demonstrate that the joint effect of thermal stresses and matrix-induced elastic clamping of ferroelectric inclusions gives rise to several important features in the polarization behavior of PbTiO and BaTiO nanocrystals. In particular, the Curie temperature displays a nonmonotonic variation with the ellipsoid's aspect ratio, being minimal for spherical inclusions. Furthermore, the diagrams show that the polarization orientation with respect to the ellipsoid's symmetry axis is controlled by the shape anisotropy and the sign of thermal stresses. Under certain conditions, the mechanical inclusion-matrix interaction qualitatively alters the evolution of ferroelectric states on cooling, inducing a structural transition in PbTiO nanocrystals and suppressing in BaTiO inclusions some transformations occurring in their bulk counterpart. The constructed phase maps open the possibility to calculate dielectric properties of strained PbTiO and BaTiO nanocrystals and ferroelectric nanocomposites comprising such crystallites.

A charge flow through a magnetic tunnel junction (MTJ) leads to the generation of a spinpolarized current which exerts a spin-transfer torque (STT) on the magnetization. When the density of applied direct current exceeds some critical value, the STT excites high-frequency magnetization precession in the "free" electrode of MTJ. Such precession gives rise to microwave output voltage and, furthermore, can be employed for spin pumping into adjacent normal metal or semiconductor. Here we describe theoretically the spin dynamics and charge transport in the CoFeB/MgO/CoFeB/Au tunneling heterostructure connected to a constant-current source. The magnetization dynamics in the free CoFeB layer with weak perpendicular anisotropy is calculated by numerical integration of the Landau-Lifshitz-Gilbert-Slonczewski equation accounting for both STT and voltage controlled magnetic anisotropy associated with the CoFeB|MgO interface. It is shown that a large-angle magnetization precession, resulting from electrically induced dynamic spin reorientation transition, can be generated in a certain range of relatively low current densities. An oscillating spin current, which is pumped into the Au overlayer owing to such precession, is then evaluated together with the injected spin current. Considering both the driving spin-polarized charge current and the pumped spin current, we also describe the charge transport in the CoFeB/Au bilayer with the account of anomalous and inverse spin Hall effects. An electric potential difference between the lateral sides of the CoFeB/Au bilayer is calculated as a function of distance from the CoFeB|MgO interface. It is found that this transverse voltage signal in Au is large enough for experimental detection, which indicates significant efficiency of the proposed current-driven spin injector. arXiv:1904.10361v1 [cond-mat.mes-hall]

Conventional approaches to detect patterns in neuronal firing are template based. As the pattern length increases, the number of trial patterns to be tested leads to strongly divergent computational costs. To remedy this problem, we propose a different statistical approach, based on the correlation integral. Applications of our method to model and neuronal data demonstrate its reliability, even in the presence of noise. Additionally, our investigation provides interesting insights into the nature of correlation-integral anomalies.