How to understand divergence_approx

[jatentaki]

I'm trying to understand divergence_approx. I am not sure what exactly it is trying to compute, I suppose something related to the trace of the Jacobian, that is equation 7 in your paper (linking in case the numbering differed between versions or something).

I am confused on two levels. Firstly, I don't know what the variables mean: the function arguments are called (f, y, e) but the function returns something called dzdx and finally it is called in a context where f=dy and y=y, so it would compute something related to d^2y/y^2?? Secondly, I don't understand the overall implementation, why is it looping (and seemingly computing the same thing in a loop, since this call is equivalent to that from line 10 and its parameters are not being updated between the iterations. Finally, why is the e parameter fixed?

Thank you in advance for your response

[emilemathieu]

Hello,

I am also wondering what is the while loop for? @stevenygd

Best, Emile