Interpolation method with random start

This method tries to improve the parameter space exploration by changing the index of the first layer. A random integer, between 0 and $n-1$ (if $n$ is the number of layers), specifies the index of the starting calculation. The computation of the velocity profile is done in the same way. A new degree of freedom must be added and the parameter space has a more complex shape. In the preceding method, there is a one-to-one relationship between the velocity profile and the set of generated random deviates. With this new parameter, one velocity profile corresponds to an infinity of possible sets of random deviates. In figure B.4, this method clearly provides a more uniform prior distribution than the preceding approaches but at the cost of a supplementary parameter.