Choice of inversion algorithm
Select one of four radio-buttons to indicate which type of inversion algorithm should be used (iatype in the manual). Some information on what these choices mean can be found in the UBC-GIF tutorial on linear inversion (section 25). NOTES:  GCV is the default because the errors on data are not often well-known. However the GCV method sometimes produces models that are somewhat "over-fit". This means there is too much structure. One approach that works sometimes is to perform an inversion on a line of data using the GCV criterion, then observe the resulting Beta values that are determined along the line. Choose a suitable value that might work for most of the line and run a second inversion run using the "fixed trade-off" option where the value of Beta you chose is specified in the "trade-off" parameter box. This is not guaranteed to work, but it is one idea.