Pretty vague question.

One assumption that “mathy types” like to make is that the slope be negative-proportional to how far the value (not slope) is from the desired target value… and then you get an exponential decay (buzzword). But there are lots of other assumptions one could make… some of which lead to PID control (buzz; very mathy stuff).

But these days you could use a neural net (buzz; so mathy they don’t usually pretend to understand what the NN “learns”) or fuzzy logic (buzz; which is ideally intuitive but has many surprisingly mathy assumptions) to make the behavior nonlinear and go to the desired result much faster… so really, there are many many possible answers. Maybe you can watch some ELI5 videos about these buzzwords and refine your question?

There are thousands of programs for Linux… but you should be warned that relatively few programs run natively on both Windows and Linux. In some cases there are ways to run “Windows programs” on Linux, but in general such successes are special cases. If you absolutely must have Windows you can run it in a virtual machine… but you will most likely be happiest with Linux if you aren’t chasing after such things.

I use Windows for work because our IT department only supports that… but I use cygwin and wsl to get a smidgen of my familiar Linux tools that I use on my personal computers.

Why downvote? This is an often overlooked trap for programmers… especially those of the “data science” variety, but certainly not restricted to that subset.

I stopped buying phones from carriers 15 years ago for this very reason.