Recovery of the parameters of a focal zone by the data of earthquakes

We present an approach for solving the inverse kinematic problem of seismic with internal sources, based on the method of multidimensional data approximation on irregular grids. The times of arrival of elastic waves to the seismic stations are considered as known. The hodographs from earthquake to the stations are approximated for further determining the velocities of longitudinal and transverse waves using the eikonal equation. The ratio of these velocities determines the Poisson’s ratio, and the other elastic parameters of the medium can be found in units of the density. The results of implementation of the approach, based on the real data, are presented.

A divide-and-conquer algorithm for seismic data approximation by the Laguerre series

An algorithm of the Laguerre transform for approximating functions on large intervals is proposed. The idea of the considered approach is that the calculation of improper integrals of rapidly oscillating functions is replaced by a solution of an initial boundary value problem for the one-dimensional transport equation. It allows one to successfully avoid the problems associated with the stable implementation of the Laguerre transform. A divide-and-conquer algorithm based on shift operations made it possible to significantly reduce the computational cost of the proposed method. Numerical experiments have shown that the methods are economical in the number of operations, stable, and have satisfactory accuracy for seismic data approximation.

A Technique for Approximate Communication in Network-on-Chips for Image Classification

Approximation is an effective technique for reducing power consumption and latency of on-chip communication in many computing applications. However, existing approximation techniques either achieve modest improvements in these metrics or require retraining after approximation, such when convolutional neural networks (CNNs) are employed. Since classifying many images introduces intensive on-chip communication, reductions in both network latency and power consumption are highly desired. In this paper, we propose an approximate communication technique (ACT) to improve the efficiency of on-chip communications for image classification applications. The proposed technique exploits the error-tolerance of the image classification process to reduce power consumption and latency of on-chip communications, resulting in better overall performance for image classification computation. This is achieved by incorporating novel quality control and data approximation mechanisms that reduce the packet size. In particular, the proposed quality control mechanisms identify the error-resilient variables and automatically adjust the error thresholds of the variables based on the image classification accuracy. The proposed data approximation mechanisms significantly reduce packet size when the variables are transmitted. The proposed technique reduces the number of flits in each data packet as well as the on-chip communication, while maintaining an excellent image classification accuracy. The cycle-accurate simulation results show that ACT achieves 23% in network latency reduction and 24% in dynamic power reduction compared to the existing approximate communication technique with less than 0.99% classification accuracy loss.

A Technique for Approximate Communication in Network-on-Chips for Image Classification

Approximation is an effective technique for reducing power consumption and latency of on-chip communication in many computing applications. However, existing approximation techniques either achieve modest improvements in these metrics or require retraining after approximation, such when convolutional neural networks (CNNs) are employed. Since classifying many images introduces intensive on-chip communication, reductions in both network latency and power consumption are highly desired. In this paper, we propose an approximate communication technique (ACT) to improve the efficiency of on-chip communications for image classification applications. The proposed technique exploits the error-tolerance of the image classification process to reduce power consumption and latency of on-chip communications, resulting in better overall performance for image classification computation. This is achieved by incorporating novel quality control and data approximation mechanisms that reduce the packet size. In particular, the proposed quality control mechanisms identify the error-resilient variables and automatically adjust the error thresholds of the variables based on the image classification accuracy. The proposed data approximation mechanisms significantly reduce packet size when the variables are transmitted. The proposed technique reduces the number of flits in each data packet as well as the on-chip communication, while maintaining an excellent image classification accuracy. The cycle-accurate simulation results show that ACT achieves 23% in network latency reduction and 24% in dynamic power reduction compared to the existing approximate communication technique with less than 0.99% classification accuracy loss.

Polyconvex hyperelastic modeling of rubberlike materials

In this work a number of selected, isotropic, invariant-based hyperelastic models are analyzed. The considered constitutive relations of hyperelasticity include the model by Gent (G) and its extension, the so-called generalized Gent model (GG), the exponential-power law model (Exp-PL) and the power law model (PL). The material parameters of the models under study have been identified for eight different experimental data sets. As it has been demonstrated, the much celebrated Gent’s model does not always allow to obtain an acceptable quality of the experimental data approximation. Furthermore, it is observed that the best curve fitting quality is usually achieved when the experimentally derived conditions that were proposed by Rivlin and Saunders are fulfilled. However, it is shown that the conditions by Rivlin and Saunders are in a contradiction with the mathematical requirements of stored energy polyconvexity. A polyconvex stored energy function is assumed in order to ensure the existence of solutions to a properly defined boundary value problem and to avoid non-physical material response. It is found that in the case of the analyzed hyperelastic models the application of polyconvexity conditions leads to only a slight decrease in the curve fitting quality. When the energy polyconvexity is assumed, the best experimental data approximation is usually obtained for the PL model. Among the non-polyconvex hyperelastic models, the best curve fitting results are most frequently achieved for the GG model. However, it is shown that both the G and the GG models are problematic due to the presence of the locking effect.