Can SU2_GEO FD computations be replaced with a SU2_DOT_AD ?

[NAnand-TUD]

Hello Community,

Recently, I was looking into the SU2_GEO code with my colleague and we see that sens computations in SU2_GEO are done using FD. Just wondering, in theory, SU2_DOT_AD differentiates the deformation for you and should give you volume and area sensitivities, it only needs to be linked to FFD DVs. Hence, I assume SU2_GEO computation using SU2_DOT_AD will be faster especially when a large number of DVs are in play.

In a current problem time to compute area sensitivities using SU2_GEO is comparable to that of the primal problem, which just does not sound right to me.

Am I oversimplifying the problem? Or am I missing something in my thought process?

What is your general thought on this?

-Nitish