Premium
Solution of inverse problems by statistical optimization methods
Author(s) -
Rastrigin L. A.
Publication year - 1965
Publication title -
reviews of geophysics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 8.087
H-Index - 156
eISSN - 1944-9208
pISSN - 8755-1209
DOI - 10.1029/rg003i001p00111
Subject(s) - inverse problem , inversion (geology) , random search , mathematical optimization , inverse , computer science , interval (graph theory) , process (computing) , algorithm , mathematics , geology , combinatorics , geometry , mathematical analysis , paleontology , structural basin , operating system
Inverse problems of seismology can be formulated as matrix inversion problems. If there are constraints on the system, such as the requirement that the parameters take positive values, and if the number of variables is large, statistical search methods become useful. A random search method is proposed which is superior to the usual gradient methods of search and which approaches the ‘ravine’ method if the interval of search is not changed by a ‘learning’ process during the search.