Automatic Determination of Stellar Atmospheric Parameters Using Neural Networks and Instance-Based Machine Learning

In this article we show how machine learning methods can beeffectively applied to the problem of automatically predictingstellar atmospheric parameters from spectral information, a veryimportant problem in stellar astronomy. We apply feedforwardneural networks, Kohonen's self-organizing maps andlocally-weighted regression to predict the stellar atmosphericparameters effective temperature, surface gravity and metallicityfrom spectral indices. Our experimental results show that thethree methods are capable of predicting the parameters with verygood accuracy. Locally weighted regression gives slightly betterresults than the other methods using the original dataset asinput, while self-organizing maps outperform the other methods when significant amounts of noise are added. We also implemented a heterogeneous ensemble of predictors, combining the results given by the three algorithms. This ensemble yields better results than any of the three algorithms alone, using both the original and the noisy data.     

Prediction of Stellar Atmospheric Parameters using Instance-Based Machine Learning and Genetic Algorithms

In this article we present a method for the automated prediction of stellar atmospheric parameters from spectral indices. This method uses a genetic algorithm (GA) for the selection of relevant spectral indices and prototypical stars and predicts their properties, using the k-nearest neighbors method (KNN). We have applied the method to predict the effective temperature, surface gravity, metallicity, luminosity class and spectral class of stars from spectral indices. Our experimental results show that the feature selection performed by the genetic algorithm reduces the running time of KNN up to 92%, and the predictive accuracy error up to 35%. This revised version was published online in July 2006 with corrections to the Cover Date.     

Observation of gravity waves in a boreal forest

In this paper we report the results of the analysis of two 60-min wave events that occurred in a boreal aspen forest during the 1994 BOREAS (Boreal Ecosystems-Atmosphere Study) field experiment. High frequency wind and temperature data were provided by three 3-D sonic anemometer/thermometers and fourteen fine-wire thermocouples positioned within and above the forest. Wave phase speeds, estimated from information revealed by spectral analysis and linear plane wave equations, are 2.2 and 1.3 m s^-1 for the two events. The wavelengths are 130 m and 65 m respectively and are much larger than the vertical wave displacements. There is strong evidence from the present analysis and from the literature supporting our postulate that these waves are generated by shear instability. We propose that wind shear near the top of the stand is often large enough to reduce the gradient Richardson number below the critical value of 0.25 and thus is able to trigger the instability. When external conditions are favorable, the instability will grow into waves.     

Role of Wavelets in the Physical and Statistical Modelling of Complex Geological Processes

— Today, wavelets are recognized to have a wide range of useful properties that allow them to treat effectively multifacet problems, such as data compression, scale-localization analysis, feature extraction, statistics, numerical simulation, visualization, and communication. Second-generation wavelets represent a recent improvement of the wavelet algorithm, allowing for greater flexibility in the spatial domain and other computational advantages. We will show how these wavelets can be employed to extract large-scale coherent structures from (1) three-dimensional turbulent flows and (2) high Rayleigh number thermal convection. We will discuss the concept of modelling via decomposition into coherent and incoherent fields, taking into account the effect of the incoherent field via statistical modelling. We will discuss wavelet properties and how they can be utilized and integrated in handling large-scale problems in earthquake physics and other nonlinear phenomena in the geosciences.     

ISA model: A constitutive model for soils with yield surface in the intergranular strain space

In this article, a new constitutive model for soils is proposed. It is formulated by means of plasticity, but in contrast to the precedent works, it presents a yield function describing a surface within the intergranular strain space. This latter is a state variable providing information of the recent strain history. An expression for the plastic strain rate has been proposed to guarantee the stress rate continuity. Under the application of medium or large strain amplitudes, the constitutive equation becomes independent of the intergranular strain and delivers a mathematical structure similar to some Karlsruhe hypoplastic models. Some simulations of monotonic and cyclic triaxial test are provided to evaluate and analyze the model performance. Copyright © 2015 John Wiley & Sons, Ltd.     

On the simulation of multidimensional cyclic loading with intergranular strain

A sample of soil is subjected to multidimensional cyclic loading when two or three principal components of the stress or strain tensor are simultaneously controlled to perform a repetitive path. These paths are very useful to evaluate the performance of models simulating cyclic loading. In this article, an extension of an existing constitutive model is proposed to capture the behavior of the soil under this type of loading. The reference model is based on the intergranular strain anisotropy concept and therefore incorporates an elastic locus in terms of a strain amplitude. In order to evaluate the model performance, a modified triaxial apparatus able to perform multidimensional cyclic loading has been used to conduct some experiments with a fine sand. Simulations of the extended model with multidimensional loading paths are carefully analyzed. Considering that many cycles are simulated (\(N>30\)), some additional simulations have been performed to quantify and analyze the artificial accumulation generated by the (hypo-)elastic component of the model. At the end, a simple boundary value problem with a cyclic loading as boundary condition is simulated to analyze the model response.     

Multiperiodic semiregular variable stars in the ASAS data base: A pilot study

Based on a search for multi‐periodic variability among the semi‐regular red variable stars in the database of the All Sky Automated Survey (ASAS), a sample of 72 typical examples is presented. Their period analysis was performed using the Discrete Fourier Transform. In 41 stars we identified two significant periods each, simultaneously present, while the remaining 31 cases revealed even three such periods per star. They occur in a range roughly between 50 and 3000 days. Inter‐relationships between these periods were analyzed using the “double period diagram” which compares adjacent periods, and the so‐called “Petersen diagram”, the period ratio vs. the shorter period. In both diagrams we could identify six sequences of accumulation of the period values. For five of these sequences (containing 97 % of all data points) we found an almost perfect coincidence with those of previous studies which were based on very different samples of semiregular red variables. Therefore, existence and locations of these sequences in the diagrams seem to be universal features, which appear in any data set of semi‐regularly variable red giants of the AGB; we conclude that they are caused by different pulsation modes as the typical and consistent properties of similar stellar AGB configurations. Stellar pulsations can be considered as the principal cause of the observed periodic variability of these stars, and not binary, rotation of a spotted surface or other possible reasons suggested in the literature. (© 2014 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)