Identification of the best architecture of a multilayer perceptron in modeling daily total ozone concentration over Kolkata,India

Autoregressive neural network (AR-NN) models of various orders have been generated in this work for the daily total ozone (TO) time series over Kolkata (22.56°N, 88.5°E). Artificial neural network in the form of multilayer perceptron (MLP) is implemented in order to generate the AR-NN models of orders varying from 1 to 13. An extensive variable selection method through multiple linear regression (MLR) is implemented while developing the AR-NNs. The MLPs are characterized by sigmoid non-linearity. The optimum size of the hidden layer is identified in each model and prediction are produced by validating it over the test cases using the coefficient of determination (R ²) and Willmott’s index (WI). It is observed that AR-NN model of order 7 having 6 nodes in the hidden layer has maximum prediction capacity. It is further observed that any increase in the orders of AR-NN leads to less accurate prediction.     

An experimental approach to strain pattern and folding in unconfined and/or partitioned transpressional deformation

Three different series of experiments were carried out with pitch (bitumen) and/or composite pitch-plasticine models to observe the spatial and temporal changes of strain pattern and/or fold styles in a tectonic zone undergoing sinistral-sense unconfined transpression. In the first series, rectangular pitch models with circular strain markers, when subjected to vertically and laterally unconstrained transpression, showed that the circular markers deformed into ellipses with long axes (maximum instantaneous stretching axis: ISA_max) oriented <45° with the plane of simple shear at the onset of deformation. This initial angle decreased with increasing angle of convergence (α = 15°, 30°, 45° and 60°) between the boundary walls. In the second series, a longitudinally pre-cut pitch block simulated a pre-existing fault zone in the crust. Lubrication along the cut was varied to simulate varying ease of fault slip. During oblique convergence, transpressional strain was partitioned between the pre-existing cut (dominantly the simple shear component) and the pitch blocks (mostly the pure shear component). Partitioning was higher for higher lubrication and lower angle of convergence in these experiments, which envisaged interesting observations on the efficiency of natural ‘partitioned transpression’ systems. In series 3, folds in a thin plasticine layer placed over the pitch block initiated parallel to the long axis of the finite strain ellipse in transpression and rotated anticlockwise with increasing strain.     

Radiatively driven winds from effective boundary layer around black holes

Matter accreting onto black holes suffers a standing or oscillating shock wave in much of the parameter space. The post-shock region is hot, puffed up and reprocesses soft photons from a Keplerian disc to produce the characteristic hard tail of the spectrum of accretion discs. The post-shock torus is also the base of the bipolar jets. We study the interaction of these jets with the hard photons emitted from the disc. We show that radiative force can accelerate outflows but the drag can limit the terminal speed. We introduce an equilibrium speed υ_eq as a function of distance, above which the flow will experience radiative deceleration.     

Torsional wave propagation in non‐homogeneous layer between non‐homogeneous half‐spaces

The study of surface wave in a layered medium has their possible application in geophysical prospecting. In the present work, dispersion equation for torsional wave in an inhomogeneous isotropic layer between inhomogeneous isotropic half‐spaces has been derived. Two cases are discussed separately for torsional wave propagation in inhomogeneous layer between homogeneous and non‐homogeneous half‐spaces, respectively. Further, two possible modes for torsional wave propagation are obtained in case of inhomogeneous layer sandwiched between non‐homogeneous half‐spaces. Closed form solutions for displacement in the layer and half‐spaces are obtained in each case. The study reveals that the layer width, layer inhomogeneity, frequency of inhomogeneity, as well as inhomogeneity in the half‐space has significant effect on the propagation of torsional surface waves. Displacement and implicit dispersion equation for torsional wave velocities are expressed in terms of Heun functions and their derivatives. Effects of inhomogeneity on torsional wave velocity are also discussed graphically by plotting the dimensionless phase velocity against dimensionless and scaled wave number for different values of inhomogeneity parameter. Copyright © 2012 John Wiley & Sons, Ltd.     

Investigation of power-law correlations within daily total ozone time series and search for a stable node through phase portrait

In the study reported in this paper we have dealt with the intrinsic complexity of the daily total ozone time series over Mumbai, India (19°23′N, 72 15°00′E) by means of detrended fluctuation analysis and phase portrait analysis. From the detrended fluctuation analysis of first order it is revealed that the said time series is characterized by persistent power-law correlation which is stronger in shorter time scale than in the longer time scale. The phase portrait analysis is based on a non-linear autonomous system generated from regression equations; infinitely many stable nodes are found along the straight line x = 273.556 and from the behaviors of the contours, it has been revealed that the daily total ozone concentration is unlikely to go beyond 315 DU (approximately) and it is almost stable.