Frechet space

A Fréchet space is a complete, metrizable locally convex topological vector space. Equivalently, it is a Hausdorff vector space whose topology is generated by a countable family of seminorms and whose associated uniform structure is complete. Fréchet spaces extend the framework of Banach spaces to settings in which no single norm adequately describes convergence.

The class contains many spaces of smooth or holomorphic functions that occur in functional analysis, partial differential equations, and distribution theory. Its completeness and metrizability support analogues of several principal Banach-space theorems, while its seminorm structure permits distinct aspects of a function to be controlled simultaneously.

Definition

Let (E) be a vector space over (\mathbb{R}) or (\mathbb{C}). A topology on (E) makes it a Fréchet space when the following conditions hold:

Addition and scalar multiplication are continuous, so that (E) is a topological vector space.
The topology is locally convex, meaning that the origin has a neighborhood basis consisting of convex sets.
The topology is induced by a translation-invariant metric.
Every Cauchy sequence, equivalently every Cauchy filter for the induced uniform structure, converges in (E).

The topology can be represented by an increasing sequence of continuous seminorms

[ p_1 \leq p_2 \leq p_3 \leq \cdots ]

that separates points. Separation means that (p_n(x)=0) for every (n) implies (x=0). A compatible translation-invariant metric is

[ d(x,y)

\sum_{n=1}^{\infty} 2^{-n} \frac{p_n(x-y)}{1+p_n(x-y)}. ]

Convergence in this metric is equivalent to convergence with respect to every seminorm:

[ x_j\longrightarrow x \quad\Longleftrightarrow\quad p_n(x_j-x)\longrightarrow 0 \text{ for every }n. ]

The particular sequence of seminorms is not intrinsic. Different separating countable families can generate the same locally convex topology. Completeness likewise concerns the resulting uniform structure rather than the numerical form of a particular compatible metric.

Every Banach space is a Fréchet space, since its norm supplies a single complete metric and a locally convex topology. The converse fails because a Fréchet topology need not arise from any norm.

Historical development

Maurice Fréchet introduced metric spaces in 1906 and established a general language for convergence independent of coordinates or finite-dimensional geometry. Stefan Banach subsequently developed the theory of complete normed spaces, which supplied a central setting for linear analysis. The Fréchet-space concept emerged from the combination of completeness, metrizability, and the locally convex methods developed during the early twentieth century.

In a 1936 study of countably generated locally convex topologies, You Watanabe established the equivalence between the complete-metric formulation and the formulation by a separating sequence of seminorms. Watanabe’s treatment also identified the bounded metric above as a canonical construction from such a sequence. This formulation entered the developing terminology of complete locally convex spaces and became part of the standard definition.

The terminology was systematized within the mid-century theory of topological vector spaces. Work by Jean Dieudonné, Laurent Schwartz, and Alexander Grothendieck placed Fréchet spaces within a larger structure involving duality, distributions, tensor products, and nuclear spaces.

Structural properties

A linear map (T:E\to F) between Fréchet spaces is continuous precisely when continuity holds at the origin. In terms of defining seminorms ((p_n)) on (E) and ((q_m)) on (F), continuity means that for each (m) there are a finite number of indices and a constant (C_m>0) such that

[ q_m(Tx) \leq C_m\max_{1\leq j\leq r_m} p_{n_j}(x) ]

for every (x\in E). When the seminorm sequence on (E) is increasing, this condition reduces to an estimate involving one sufficiently large seminorm.

Closed linear subspaces of a Fréchet space are Fréchet in the subspace topology. If (M\subseteq E) is closed, the quotient (E/M) is also Fréchet under the quotient topology. Countable products of Fréchet spaces remain Fréchet because their topologies are generated by countably many coordinate seminorms and completeness is inherited coordinatewise.

A Fréchet space can also be represented as a projective limit of Banach spaces. Given an increasing defining family ((p_n)), each quotient

[ E/\ker p_n ]

carries a norm induced by (p_n). Completion produces a Banach space (E_n), and the compatibility of the seminorms gives bonding maps (E_{n+1}\to E_n). The original space is identified with a closed subspace of the inverse limit

[ \varprojlim E_n. ]

This representation explains how Banach-space arguments can be applied at individual seminorm levels without replacing the full topology by a single norm.

Fundamental mapping theorems

The Baire category theorem applies to Fréchet spaces because they admit complete compatible metrics. Consequently, several principal results of Banach-space analysis extend to this setting.

The open mapping theorem states that a continuous surjective linear map between Fréchet spaces is open. Thus, if a continuous linear bijection (T:E\to F) exists, its inverse is automatically continuous.

The closed graph theorem states that a linear map between Fréchet spaces is continuous when its graph is closed in (E\times F). The product is itself Fréchet, so the result remains within the same category.

The uniform boundedness principle also holds in its locally convex form. A pointwise bounded family of continuous linear maps from a Fréchet space into a locally convex space is equicontinuous under the standard hypotheses on the range. These theorems depend on completeness and the Baire property rather than on the existence of a norm.

Function-space examples

Let (\Omega\subseteq\mathbb{R}^d) be open. The space (C^\infty(\Omega)) of smooth functions becomes a Fréchet space under seminorms that control derivatives on compact subsets. For a compact exhaustion

[ K_1\subseteq K_2\subseteq\cdots, \qquad \bigcup_{n=1}^{\infty}K_n=\Omega, ]

one defining family is

[ p_n(f)

\max_{\substack{x\in K_n\|\alpha|\leq n}} \left|\partial^\alpha f(x)\right|. ]

Convergence in (C^\infty(\Omega)) therefore means uniform convergence on every compact subset of the functions together with all derivatives. Completeness follows because compatible limits of the derivatives determine a smooth limiting function.

For an open subset (\Omega\subseteq\mathbb{C}), the space (H(\Omega)) of holomorphic functions is Fréchet under uniform convergence on compact subsets. Its topology can be defined by

[ p_n(f)=\sup_{z\in K_n}|f(z)| ]

for a compact exhaustion adapted to (\Omega). The locally uniform limit of holomorphic functions is holomorphic, which gives completeness.

The sequence space

[ \omega=\mathbb{K}^{\mathbb{N}} ]

with the product topology is another Fréchet space. Its topology is generated by

[ p_n(x)=\max_{1\leq k\leq n}|x_k|. ]

This space is not normable. Every neighborhood of the origin restricts only finitely many coordinates, whereas a norm ball would impose one simultaneous absorbing constraint on the entire sequence.

The space (s) of rapidly decreasing sequences is Fréchet under the seminorms

[ p_m(x)=\sup_{n\geq 1} n^m|x_n|. ]

Its topology records every polynomial order of decay. This example is closely related to the Schwartz space, whose seminorms simultaneously control derivatives and polynomially weighted growth.

Normability and boundedness

A locally convex space is normable exactly when it has a bounded convex neighborhood of the origin. For a Fréchet space, this criterion determines whether its countable seminorm structure can be replaced by one norm. If such a norm exists, completeness of the original topology makes the space Banach.

Boundedness in a Fréchet space is topological rather than metric in the ordinary sense. A set (B\subseteq E) is bounded when every continuous seminorm is bounded on (B):

[ \sup_{x\in B}p_n(x)<\infty \qquad \text{for every }n. ]

The compatible metric constructed from the seminorms is itself bounded above by (1), so metric boundedness carries no corresponding information. This distinction illustrates why the metric is used to encode convergence and completeness, while seminorms retain the linear geometry.

Continuous dual and weak topologies

The continuous dual space (E') consists of continuous linear functionals (E\to\mathbb{K}). A functional (\ell) is continuous when it is controlled by a continuous seminorm, which for an increasing defining sequence takes the form

[ |\ell(x)|\leq C p_n(x) ]

for some (C>0) and some index (n).

Several natural topologies can be placed on (E'). The weak-(*) topology records pointwise convergence on (E), while the strong dual topology records uniform convergence on bounded subsets of (E). The strong dual of a Fréchet space is generally not another Fréchet space. It belongs naturally to the theory of DF-spaces, especially when the original space has additional compactness or nuclearity properties.

Limitations of the category

Not every locally convex function space is Fréchet. The test-function space (C_c^\infty(\Omega)) is normally equipped with an inductive-limit topology reflecting variation of compact support. When (\Omega) is not compact, this topology is not metrizable and therefore does not define a Fréchet space.

An infinite-dimensional Banach space equipped with its weak topology is also generally not Fréchet. The weak topology is locally convex, but metrizability on the entire space fails except under restrictive finite-dimensional circumstances. Metrizability of weak topology on bounded subsets is a separate question governed by properties of the continuous dual.

Completeness alone is insufficient as well. Complete locally convex spaces can require uncountably many seminorms and thus fail to be metrizable. Conversely, a metrizable locally convex space can be incomplete, in which case its completion is Fréchet if the completed topology remains Hausdorff.