The inversion of geophysical data is often very challenging and deal with nonlinear, multidimensional and multimodal optimization problems. For such problems, it is useful to investigate relative ambiguity of the model parameters, understand if all of them can be determined and select the appropriate optimization algorithm. However, we need the information about number, position, distribution and size of local maxima-and-minima, that is, full knowledge of the objective function topography. The conventional way to get the topography of the objective function is through Residual Function Map (RFM). To construct this map, two parameters are made variable and the others ones are kept fixed. However, the RFM mapping has serious limitations for models with more than two parameters since it is required the knowledge of the exact value of parameters to be fixed, which is possible only with synthetic models, and can mask local maxima-and-minima of the function. To overcome conventional methodology limitations used to construct RFM we adopted an alternative way to obtain topography of the multidimensional objective function.
Palestrante: Oleg Bokhonok