Differentiation in Fréchet spaces

Differentiation in Fréchet spaces concerns the extension of the classical differential calculus from finite‑dimensional Euclidean spaces and Banach spaces to the broader class of Fréchet spaces. A Fréchet space is a locally convex topological vector space that is complete and metrizable; equivalently, its topology can be defined by a countable family of seminorms ${p_k}_{k\in\mathbb{N}}$.

Definitions

1. Fréchet derivative
   Let $E$ and $F$ be Fréchet spaces and let $U\subseteq E$ be an open set. A map $f:U\rightarrow F$ is said to be Fréchet differentiable at a point $x\in U$ if there exists a continuous linear map $Df(x):E\rightarrow F$ such that

   $$ \lim_{h\to 0}\frac{\bigl|f(x+h)-f(x)-Df(x)h\bigr|_F}{|h|_E}=0, $$

   where the limit is taken with respect to the metric that generates the topology of $E$. The expression $Df(x)$ is called the Fréchet derivative of $f$ at $x$.

2. Gateaux derivative
   The Gateaux derivative of $f$ at $x$ in the direction $v\in E$ is the limit (if it exists)

   $$ D_G f(x)(v)=\lim_{t\to 0}\frac{f(x+tv)-f(x)}{t}. $$

   If the Gateaux derivative exists for every direction and depends continuously on $x$ and $v$, then $f$ is Fréchet differentiable and the two notions coincide.

3. Higher‑order derivatives
   Higher‑order Fréchet derivatives are defined inductively: the second derivative $D^2f(x)$ is a continuous bilinear map $E\times E\to F$ obtained by differentiating the first derivative, and similarly for higher orders.

Fundamental Properties

• Linearity and continuity – The derivative $Df(x)$ is a continuous linear operator between Fréchet spaces.

• Chain rule – If $f:U\subseteq E\to F$ and $g:V\subseteq F\to G$ are Fréchet differentiable with $f(U)\subseteq V$, then $g\circ f$ is Fréchet differentiable and

  $$ D(g\circ f)(x)=Dg\bigl(f(x)\bigr)\circ Df(x). $$

• Local boundedness – On bounded subsets of $U$, the derivative of a continuously Fréchet‑differentiable map is uniformly bounded with respect to the seminorms that define the topology.

• Mean value inequality – If $f$ is continuously Fréchet differentiable on a convex open set, then

  $$ p\bigl(f(y)-f(x)\bigr)\leq \sup_{t\in[0,1]}p\bigl(Df(x+t(y-x))(y-x)\bigr) $$

  holds for each seminorm $p$ on $F$.

Existence Theorems

• Inverse Function Theorem (Nash–Moser) – In Banach spaces the classical inverse function theorem holds with a continuously invertible derivative. For Fréchet spaces, a direct analogue is false in general because the Open Mapping Theorem does not apply. The Nash–Moser theorem supplies a version of the inverse function theorem under additional hypotheses (e.g., a “tame” Fréchet space and a tame smooth map with a tame right inverse of its derivative).

• Implicit Function Theorem – Similar restrictions apply; a tame version exists within the Nash–Moser framework.

Calculus on Convenient Vector Spaces

An alternative approach, developed by Kriegl and Michor, replaces the Fréchet topology with the convenient setting. In this framework, smoothness of a map $f:E\to F$ between locally convex spaces is defined by requiring that $f\circ c$ be smooth for every smooth curve $c:\mathbb{R}\to E$. This definition coincides with Fréchet differentiability for Banach spaces and yields a robust differential calculus on a large class of Fréchet spaces.

Applications

• Partial differential equations – Function spaces such as $C^\infty(M)$ or Sobolev spaces of all orders are Fréchet spaces; differentiability concepts enable the formulation of nonlinear PDEs and their variational analysis.
• Infinite‑dimensional manifolds – Manifolds modeled on Fréchet spaces appear in the study of loop spaces, diffeomorphism groups, and spaces of sections of fiber bundles. Differentiability in the model spaces underlies the definition of tangent bundles and differential forms on such manifolds.
• Geometric analysis – The Nash–Moser theorem, which relies on differentiation in Fréchet spaces, was crucial for the original proof of the isometric embedding theorem of Nash.

References

1. J. M. A. S. G. M. Krieg and P. W. Michor, The Convenient Setting of Global Analysis, American Mathematical Society, 1997.
2. R. S. Hamilton, “The Inverse Function Theorem of Nash and Moser”, Bull. Amer. Math. Soc. 7 (1982), 65–222.
3. H. Cartan, Differential Calculus in Locally Convex Spaces, in Éléments de mathématique, 1970.
4. J. M. Bourbak, “Differentiable maps between Fréchet spaces”, Ann. Inst. Fourier 20 (1970), 1–33.