I donβt think this is any faster than directly computing f\left( -\frac{b}{2a}\right) or completing the square.
This part isnβt really necessary. If the coefficient of x^2 is negative, the method still works; it simply returns the maximum instead of the minimum. Essentially, the point of interest is the vertex, and computing the derivative will return the x-coordinate of the vertex in all cases.