The Fréchet derivative of a map f between Banach spaces at a point x is the bounded linear operator Df(x) that best approximates f near x, meaning f(x+h) = f(x) + Df(x)[h] + o(‖h‖) as h→0.

Principle

It formalizes the idea of total linear approximation in normed spaces: the derivative is the unique linear map whose error term is little-o of the norm increment.

Demonstration

For f: R → R, f(x)=x^2, the Fréchet derivative at x is the linear map h ↦ 2x h because (x+h)^2 = x^2 + 2x h + h^2 and h^2 = o(|h|).

Misapplication

Assuming existence of all directional (Gâteaux) derivatives guarantees a Fréchet derivative; directional differentiability alone need not imply total (Fréchet) differentiability.

Consequence

When present, the Fréchet derivative yields stable linearization, validity of the chain rule in Banach spaces, and error control for Newton-type methods.

Reversal

The opposite notion is Gâteaux (directional) derivative, which tests linear approximation only along individual directions rather than uniformly in norm.

Boundary

Defined only between normed (typically Banach) spaces and requires the linear approximation to be bounded and uniform in all directions; it excludes merely pointwise or weak derivatives.

Semantic Tension

Confused in practice with weaker notions (Gâteaux derivative, distributional derivatives or pointwise partial derivatives); Fréchet derivative demands uniform smallness of the remainder in norm.

Synthesis

The Fréchet derivative is the unique bounded linear map that provides the best uniform linear approximation to a mapping between normed spaces, ensuring strong differentiability and classical calculus rules in infinite-dimensional settings.