Inversion for elasticity tensor of focal region using machine learning algorithms

The moment tensor is an essential tool in seismology to examine the structure of seismic sources. The deformation at the focal region, using combinations of force couples arranged in a 3 x 3 matrix, is represented by the moment tensor. The moment tensor can be expressed as a linear combination of the eigenvectors of the anisotropic focal region's elasticity tensor. The eigenvalues of a vertically transversely isotropic (VTI) elasticity tensor from an occurring moment tensor of a focal region can be obtained, and the precision of this determination depends on the degree of anisotropy. Machine learning optimization involves iteratively enhancing a machine learning model's precision by reducing the error level. Choosing an algorithm that can effectively sample the search space and identify optimal solutions is necessary to optimise a function. Many algorithms are available for function optimization, but it is crucial to set a baseline to determine the practicable solutions for a given problem. This thesis defines a new objective function (misfit function). The function is proposed for obtaining the elastic parameters of an anisotropic focal region, and these parameters are calculated by using machine learning algorithms such as Grid Search, Random Search, Simulated Annealing, and Nelder-Mead.