@vv Thank you, Direct Search is doing wonders. Though I still don't get how Minimize works with f(x), it isn't C^2 either.

Ok now I get why an expression used as a function would work. But why doesn't Minimize(g(x), {x>=-2, x<=2}) work? And more important, what package could I use to optimize without assuming the objective is C^2 everywhere?

@acer Yes, you're totally right, I just saw one can plot a function instead of an expression, and that giving "g(y)" is trying to give an expression, instead of a function. The right way to do it is then:



Should I delete the comment? It could make it confusing.

I also had some trouble with the derivative, since it's a function, one should use D, in this case D(g) is the derivative of g. But I still have a problem, if I specify a range for the inverse as in

then, the derivative is zero, but I can't seem to understand why. Do you have any idea?

Cryme

 

To whomever would want to plot the inverse, you can use such a procedure,



And use it like plot,

@Kitonum Hello,

I also have tried this method, it does work when you ask for specific values as you wrote, but I get an error when I try to plot the inverse. Am I doomed to make a procedure to generate a list to plot, or is there any other solution?

Here is the code,

Error, (in fsolve) y is in the equation, and is not solved for

Sincerely,

Cryme

@vv Yes, that does seem to do the trick and yes I'm ok with rational polynomials. Thank you!
So, I didn't look too much into it, it's basically trying a lot of dyadic rationals?