I was going through the paper Quantum Risk Analysis in the section III they came up with a technique, in which they represented the analytical function into a polynomial and then used the slope of that polynomial as a parameter for the CRY gate and loaded the polynomial onto the amplitude.
My question is how did they do it? Can anyone point out to some resource regarding this. I went through the LinearAmplitudeFunction and LinearPauliRotation function of qiskit, but they are simply implementing this. Also this doesn't take into consideration what if our function is multivariate, instead of a simple $f(x) = ax+b$ where $a$ is the slope and $b$ is the offset, what if we have a function $f(x,y) = ax+by+c$ what will be the slope then?
