"Goal Driven Optimization" sensitivity in ANSYS

[eyeguy]

Hello,

I'm doing a little project involving Design of Experiments in ANSYS (DesignXplorer). I've noticed that I can calculate a supposed "global" sensitivity to each design factor based on the "Goal Driven Optimization" technique in the program. It does not, however, provide any reference regarding the method, nor any insight into the mathematical technique used to arrive at the values. I'm a bit wary of making any decisions based off of a black box result, so was hoping to read up on it. There's very little mention of Goal Driven Optimization online, however.

So, does anyone know how the sensitivities in GDO are calculated? Also, does anyone know where I can read more about the methods used to achieve these results?

Thanks very much.

 

[eyeguy]

Ok, so just as I posted this I finally found mention that it is somehow based on the Spearman rank correlation coefficient, although it still does not really explain how this is extended n-dimensionally. Any further insight or references would still be very welcome.