0
$\begingroup$

I came across this term Adjoint Algorithmic Differentiation.

It was explained to me that this is how people in the finance industry uses to measure the sensitivity of their numerical models.

I tried googling but I found no easy intuitive way to understand this Adjoint Algorithmic Differentiation nor how it helps to measure sensitivity of models.

Can anyone help?

Context:

I know basic differentiation. Basic statistics. I can write programming using languages such as php and python.

  • 0
    I think that, first, you should need to have a look at https://en.wikipedia.org/wiki/Automatic_differentiation – 2017-02-05
  • 0
    I read it and all i get is that you can perform differentiation using computers. Even if I can accept that explanation, how does it help with the measurement of sensitivity of models? – 2017-02-05
  • 0
    If you use automatic differentiation of programs, you get the derivatives with respect to any variable that is to say the gradient and sensitivity analysis follows. Have a look at http://www-sop.inria.fr/tropics/tapenade.html. You can use it for free online (the code must be in Fortran).. – 2017-02-05
  • 0
    Maybe I should back up some more. What does it mean to do sensitive analysis on a model? – 2017-02-05

0 Answers 0