Bourbaki

§ 1. Duality

1. Topologies compatible with a duality

In this section, E and F denote two vector spaces put into duality by a bilinear form B (II, p. 40). We recall (II, p. 41) that we defined two linear mappings

dB:F→E∗,sB:E→F∗ d_B : F \to E^*, \quad s_B : E \to F^*

characterized by the relation

(1)

B(x,y)=⟨x,dB(y)⟩=⟨y,sB(x)⟩ B(x, y) = \langle x, d_B(y) \rangle = \langle y, s_B(x) \rangle

for x∈E, y∈Fx \in E, \ y \in F.

Definition 1

A locally convex topology T\mathscr{T} on E is said to be compatible with the duality between E and F if dBd_B is a bijection from F onto the dual of the locally convex space obtained by assigning the topology T\mathscr{T} to E.

If there exists one such topology T\mathscr{T}, the mapping dBd_B is injective, that is to say, the duality between E and F is separating in F (II, p. 41).

Proposition 1

(i) The closed convex subsets in E are the same for all the locally convex topologies on E which are compatible with the duality between E and F. (ii) *The bounded subsets of E are the same for all the locally convex topologies on E which are compatible with the duality between E and F.

Let T\mathscr{T} be a topology on E compatible with the duality between E and F, hence finer than σ(E,F)\sigma(E, F). If a convex subset of E is closed for T\mathscr{T}, it is the intersection of closed, real half-spaces (II, p. 38, cor. 1), hence it is closed for σ(E,F)\sigma(E, F). This proves (i). Assertion (ii) was proved in cor. 3 of III, p. 27.

Let FσF_\sigma denote the vector space F endowed with the weak topology σ(F,E)\sigma(F, E). Then the linear mapping sBs_B maps E onto the dual (Fσ)′(F_\sigma)' of FσF_\sigma (II, p. 43, prop. 3). Let S\mathfrak{S} be a family of bounded subsets of FσF_\sigma. By abuse of language, the inverse image under sBs_B of the S\mathfrak{S}-topology on (Fσ)′(F_\sigma)' is called the S-topology on E\mathfrak{S}\text{-topology on } E. It is defined by the family of semi-norms

pA(x)=sup⁡y∈A∣B(x,y)∣, p_A(x) = \sup_{y \in A} |B(x, y)|,

where A runs through S\mathfrak{S}. In particular, when S\mathfrak{S} is the family of finite subsets of F, the S\mathfrak{S}-topology is precisely the weak topology σ(E,F)\sigma(E, F).

Definition 2

Let E and F be two spaces in duality. The Mackey topology on E, denoted by τ(E,F)\tau(E, F) is defined as the S\mathfrak{S}-topology on E, where S\mathfrak{S} is the family of all subsets of F whose image in E∗E^* (under dBd_B) is convex, balanced and compact for σ(E∗,E)\sigma(E^*, E).

When the duality between E and F is separating in F, dBd_B is injective and the topology σ(F,E)\sigma(F, E) on F is the inverse image under dBd_B of the topology σ(E∗,E)\sigma(E^*, E) on E∗E^*. In this case, S\mathfrak{S} consists of all those subsets of F which are convex, balanced and compact for σ(F,E)\sigma(F, E).

In general, if F1=dB(F)⊂E∗F_1 = d_B(F) \subset E^*, and if we denote by (x,y1)↦B1(x,y1)(x, y_1) \mapsto B_1(x, y_1) the restriction of the canonical bilinear form (x,x∗)↦⟨x,x∗⟩(x, x^*) \mapsto \langle x, x^* \rangle to E×F1E \times F_1, then E and F1F_1 are put in duality by B1B_1, and this duality is separating in F1F_1, since by definition we have B(x,y)=B1(x,dB(y))B(x, y) = B_1(x, d_B(y)), def. 2 shows that τ(E,F)=τ(E,F1)\tau(E, F) = \tau(E, F_1).

Remark 1

— Let A be a compact convex subset of a Hausdorff locally convex space G, and let A~\tilde{A} be the closed convex balanced envelope of A. When the field K is R\mathbf{R}, the set A~\tilde{A} is the closed convex envelope of A∪(−A)A \cup (-A); when K is C\mathbf{C}, the set A~\tilde{A} is contained in the closed convex envelope of 2A∪(−2A)∪(2iA)∪(−2iA)2A \cup (-2A) \cup (2iA) \cup (-2iA). Consequently (II, p. 14, prop. 15), A~\tilde{A} is compact.

We deduce, in particular, that when the duality between E and F is separating in F, the Mackey topology τ(E,F)\tau(E, F) is also the S′\mathfrak{S}'-topology, where S′\mathfrak{S}' is the set of all convex subsets of F which are compact for σ(F,E)\sigma(F, E). In an analogous way we define the Mackey topology τ(F,E)\tau(F, E) on F.

Theorem 1 (Mackey). — Let E and F be two spaces in duality; suppose that the duality is separating in F. In order that a locally convex topology T\mathscr{T} on E be compatible with the duality between E and F, it is necessary and sufficient that T\mathscr{T} be finer than the topology σ(E,F)\sigma(E, F) and coarser than the Mackey topology τ(E,F)\tau(E, F).

Identify F with its image in E∗E^* under dBd_B. Let S0\mathfrak{S}_0 denote the set of all subsets of F which are convex, balanced and compact for σ(F,E)\sigma(F, E). By definition, τ(E,F)\tau(E, F) is the S0\mathfrak{S}_0-topology on E, hence is finer than σ(E,F)\sigma(E, F).

Lemma 1

The subspace F of E* consists of all linear forms on E which are continuous for τ(E,F)\tau(E, F).

Every element of F is a continuous mapping for σ(E,F)\sigma(E, F), hence for τ(E,F)\tau(E, F).

Conversely, let f∈E∗f \in E^* be continuous for τ(E,F)\tau(E, F). There exists a neighbourhood U of 0 in E (for τ(E,F)\tau(E, F)), such that ∣f∣≤1|f| \leq 1 on U; we can assume that there exists a set A∈S0A \in \mathscr{S}_0 such that U=A∘U = A^\circ. In other words, ff belongs to the bipolar A∘∘A^{\circ\circ} of A for the duality between E∗E^* and E. But the topology σ(F,E)\sigma(F, E) on F is induced by σ(E∗,E)\sigma(E^*, E); consequently A is convex, balanced and compact for σ(E∗,E)\sigma(E^*, E), and the theorem of bipolars (II, p. 44, th. 1) implies the equality A=A∘∘A = A^{\circ\circ}. Therefore we have that f∈Ff \in F, from which the lemma follows.

Lemma 2

Let T\mathscr{T} be a locally convex topology on E such that every linear form on E which is continuous for T\mathscr{T} belongs to F. Then T\mathscr{T} is coarser than τ(E,F)\tau(E, F).

Let U\mathscr{U} be the set of convex, balanced neighbourhoods of 0 for T\mathscr{T}. Let S\mathscr{S} be the set of polars in F of elements of U\mathscr{U}. By cor. 2 of III, p. 17, we have S⊂S0\mathscr{S} \subset \mathscr{S}_0, and by cor. 1 of prop. 7 of III, p. 19, T\mathscr{T} is identical with the S′\mathscr{S}'-topology, where S′\mathscr{S}' is the set of polars of sets of U\mathscr{U} in the dual E′E' of E. But E′⊂FE' \subset F, by hypothesis, hence every set of S′\mathscr{S}' is contained in a set of S\mathscr{S}; and the lemma follows.

Let T\mathscr{T} be a topology on E compatible with the duality between E and F. Then T\mathscr{T} is coarser than τ(E,F)\tau(E, F) by lemma 2, and evidently T\mathscr{T} is finer than σ(E,F)\sigma(E, F). Conversely, F is the dual of E for the topology τ(E,F)\tau(E, F) (lemma 1) and for the topology σ(E,F)\sigma(E, F) (II, p. 43, prop. 3), hence also for every topology intermediate between τ(E,F)\tau(E, F) and σ(E,F)\sigma(E, F).

Corollary

Let p be a semi-norm on E. The following conditions are equivalent :

(i) p is continuous for the topology τ(E,F)\tau(E, F); (ii) every linear form f on E, such that ∣f∣≤p|f| \leq p, comes from an element of F.

(i) ⇒\Rightarrow (ii) : if p is continuous for τ(E,F)\tau(E, F), every linear form f on E such that ∣f∣≤p|f| \leq p is continuous for τ(E,F)\tau(E, F), hence comes from an element of F by lemma 1.

(ii) ⇒\Rightarrow (i) : let T\mathscr{T} be the topology on E defined by the semi-norm p. If condition (ii) is satisfied, the linear forms on E which are continuous for T\mathscr{T} belong to F. By lemma 2 T\mathscr{T} is coarser than τ(E,F)\tau(E, F), hence p is continuous for τ(E,F)\tau(E, F).

Remark 2

Let K be a convex subset of F which is compact for the weak topology σ(F,E)\sigma(F, E) and μ\mu a positive measure on K. Put

p(x)=∫K∣B(x,y)∣ dμ(y) p(x) = \int_K |B(x, y)| \, d\mu(y)

for all x∈Ex \in E. It is immediate that p is a semi-norm. Moreover, for every x∈Ex \in E, the relation « ∣B(x,y)∣≤1|B(x, y)| \leq 1 for all y∈Ky \in K » implies that p(x)≤μ(K)p(x) \leq \mu(K). This proves that the semi-norm p on E is continuous for the Mackey topology τ(E,F)\tau(E, F). \

Example

Let G be a locally convex space and G' its dual. On G', the weak topology σ(G′,G)\sigma(G', G) and the topology of convex compact convergence (III, p. 14) are compatible with the duality between G' and G. In general, the strong topology and the topology of compact convergence on G' are not compatible with the duality between G' and G. Recall however that when G is Hausdorff and quasi-complete, the topology of compact convergence on G' coincides with that of convex compact convergence (III, p. 8), hence is compatible with the duality between G' and G.

Definition 3

Let E and F be two vector spaces in duality, and T\mathscr{T} the family of subsets of F which are bounded for σ(F,E)\sigma(F, E). Then the T\mathscr{T}-topology on F is denoted by β(E,F)\beta(E, F).

Similarly, we define the topology β(F,E)\beta(F, E) on F. It can be seen easily that the topology β(E,F)\beta(E, F) is identical with β(E,F/E∘)\beta(E, F/E^\circ), and we can reduce to the case when the duality between E and F is separating in F.

Remarks. — 3) Let EσE_\sigma denote the space E endowed with the topology σ(E,F)\sigma(E, F). The barrels (III, p. 24) in EσE_\sigma are the subsets of E which are convex, balanced closed and absorbent for σ(E,F)\sigma(E, F). These are none other than the polars of the subsets of F which are convex, balanced and bounded for σ(F,E)\sigma(F, E). Consequently, the family of all barrels in EσE_\sigma is a fundamental system of neighbourhoods of 0 for the topology β(E,F)\beta(E, F) in E. In other words, a semi-norm on E is continuous for β(E,F)\beta(E, F) if and only if it is lower semi-continuous for σ(E,F)\sigma(E, F) (cf. III, p. 24, prop. 1).

4) Let T\mathscr{T} be a topology on E compatible with the duality between E and F. By prop. 1, (ii) of IV, p. 1, the topology β(F,E)\beta(F, E) on F is none other than the strong topology on F, when F is identified with the dual of E (with the topology T\mathscr{T}).

5) The topology β(E,F)\beta(E, F) on E is finer than τ(E,F)\tau(E, F). It is not, in general compatible with the duality between E and F (cf. however § 2). In particular, a subset of E which is bounded for σ(E,F)\sigma(E, F) is not necessarily bounded for β(E,F)\beta(E, F).

2. Mackey topology and weakened topology on a locally convex space

Let E be a locally convex space and E' its dual. We put E and E' in duality by means of the canonical bilinear form (x,x′)↦⟨x,x′⟩(x, x') \mapsto \langle x, x' \rangle on E×E′E \times E'. This duality is separating in E'. We shall consider three topologies on E compatible with the duality between E and E' :

a) the given topology on E, which we shall call the initial topology, whenever any confusion is likely to arise;

b) the topology σ(E,E′)\sigma(E, E'), called the weakened topology on E;

c) the topology τ(E,E′)\tau(E, E'), called the Mackey topology on E.

The initial topology is finer than the weakened topology and coarser than the Mackey topology; moreover, these three topologies can be distinct (IV, p. 49, exerc. 8).

By prop. 1 of IV, p. 1, these three topologies have the same closed convex sets, the same barrels, the same bounded sets and the same adapted bornologies. In particular :

Proposition 2

Let E be a locally convex space, and let A be a convex subset of E (for example, a vector subspace of E). The closure of A is the same for the initial topology and for the weakened topology of E.

Remark 1

For a family (xi)i∈I(x_i)_{i \in I} of elements of E to be total (resp. topologically independent) for the initial topology, it is necessary and sufficient that it is so for the weakened topology; this follows from prop. 2. Hence we can apply the criteria of II, p. 43.

Remark 2

Let T1\mathscr{T}_1 and T2\mathscr{T}_2 be two locally convex topologies on E, compatible with the duality between E and E′E', T1\mathscr{T}_1 being finer than T2\mathscr{T}_2. Then every neighbourhood of 0 for T1\mathscr{T}_1, which is convex and closed for T1\mathscr{T}_1 is closed for T2\mathscr{T}_2 by prop. 1 of IV, p. 1. Consequently (GT, II, § 3, No. 3, corollary) every subset of E which is complete for T2\mathscr{T}_2 is so for T1\mathscr{T}_1 also.

In particular, every subset of E which is complete for the weakened topology is complete for the initial topology, and every subset of E complete for the initial topology is so for the Mackey topology. If E is quasi-complete for the weakened topology, it is so for every topology compatible with the duality between E and E′E'. If it is quasi-complete for the initial topology, it is so for the Mackey topology.

Remark 3

Suppose E is Hausdorff (for the initial topology). Let A be a subset of E which is closed and bounded for σ(E,E′)\sigma(E, E'), hence also for every topology compatible with the duality between E and E′E'. Since A is precompact for σ(E,E′)\sigma(E, E') (III, p. 3, Remark 5), assuming that A is compact for σ(E,E′)\sigma(E, E') is equivalent to A being complete for σ(E,E′)\sigma(E, E').

Therefore, on account of remark 2, we see that :

Proposition 3

Suppose E is Hausdorff, and E′E' its dual. Every subset of E which is precompact for the initial topology and compact for σ(E,E′)\sigma(E, E'), is compact for the initial topology.

Remark 4

The topology β(E,E′)\beta(E, E') (IV, p. 4, def. 3) is finer than the Mackey topology. If β(E,E′)\beta(E, E') is distinct from τ(E,E′)\tau(E, E'), it is not compatible with the duality between E and E′E'. The space E is barrelled if and only if the initial topology is equal to β(E,E′)\beta(E, E') (III, p. 24).

Proposition 4

Let E be a locally convex space. The Mackey topology on E is identical with the initial topology in each of the following cases :

a) E is barrelled; b) E is bornological; c) E is metrizable.

We note first that the Mackey topology on E is identical with the initial topology if and only if every convex subset of E′E' which is compact for σ(E′,E)\sigma(E', E), is equi-continuous. This is certainly the case if E is barrelled (III, p. 24, corollary).

Suppose E is bornological; let V be a convex and balanced neighbourhood of 0 in E for the topology τ(E,E′)\tau(E, E'). Let B be a subset of E, bounded for the initial topology. Since B is bounded for the Mackey topology, V absorbs B, and since E is bornological, V is a neighbourhood of 0 for the initial topology.

In case c), the space E is bornological (III, p. 12, prop. 2).

3. Transpose of a continuous linear mapping

In this section, E1E_1 and E2E_2 denote two locally convex spaces, with respective duals E1′E'_1 and E2′E'_2.

Let uu be a linear mapping from E1E_1 into E2E_2. For uu to be continuous when E1E_1 and E2E_2 are assigned the weakened topologies, it is necessary and sufficient that f∘uf \circ u belongs to E1′E'_1 for all f∈E2′f \in E'_2; this is the case if uu is continuous. Then the linear mapping f↦f∘uf \mapsto f \circ u from E2′E'_2 into E1′E'_1 is called the transpose of uu and is denoted by tu^t u.

Proposition 5

Let uu be a continuous linear mapping from E1E_1 into E2E_2.

(i) If E1E_1 and E2E_2 are Hausdorff then uu is injective if and only if the image of tu^t u is dense in E1′E'_1 for the weak topology σ(E1′:E1)\sigma(E'_1 : E_1).

(ii) For tu^t u to be injective it is necessary and sufficient that the image of uu is dense in E2E_2.

A vector subspace of E2E_2 is dense for the initial topology if and only if it is dense for the weakened topology (IV, p. 4, prop. 2). Prop. 5 then follows from II, p. 47, cor. 2.

Proposition 6

Let uu be a linear mapping from E1E_1 into E2E_2 which is continuous for the weakened topologies. For i=1,2i = 1, 2, let Si\mathfrak{S}_i be a family of bounded subsets of EiE_i. In order that tu^t u is a continuous mapping from (E2′)S2(E'_2)_{\mathfrak{S}_2} into (E1′)S1(E'_1)_{\mathfrak{S}_1}, it is necessary and sufficient that, for every set A∈S1A \in \mathfrak{S}_1, there exist sets A1,...,AnA_1, ..., A_n in S2\mathfrak{S}_2 and a real number λ>0\lambda > 0 such that λ⋅u(A)\lambda \cdot u(A) is contained in the closed convex balanced envelope of A1∪...∪AnA_1 \cup ... \cup A_n\footnote{In other words, u(S1)u(\mathfrak{S}_1) is contained in the smallest adapted bornology containing S2\mathfrak{S}_2 (III, p. 3).}.

This is an immediate consequence of prop. 2 of III, p. 15.

Corollary

Let uu be a continuous linear mapping from E1E_1 into E2E_2. Then tu^t u is continuous when the duals Ei′E'_i are assigned the following topologies:* a) the weak topologies σ(Ei′,Ei)\sigma(E'_i, E_i); b) the strong topologies β(Ei′,Ei)\beta(E'_i, E_i); c) the Mackey topologies τ(Ei′,Ei)\tau(E'_i, E_i); d) the topologies of precompact convergence. Moreover, if E2E_2 is Hausdorff, tu^t u is continuous when the duals Ei′E'_i are assigned: e) the topologies of compact convergence (resp. compact convex).

The only point which requires a proof is the case c), when the topologies of E1E_1 and E2E_2 are not necessarily Hausdorff. Then for every linear form f∈E1′∗f \in E'_1*, f∘tuf \circ ^t u is a linear form on E2′E'_2; hence there is a linear mapping v:E1′∗→E2′∗v : E'_1* \to E'_2* which is continuous for the topologies σ(E1′∗,E1′)\sigma(E'_1*, E'_1) and σ(E2′∗,E2′)\sigma(E'_2*, E'_2) and is such that dB2∘u=v∘dB1d_{B_2} \circ u = v \circ d_{B_1}, where dBid_{B_i} is the canonical mapping from EiE_i into Ei′∗E'_i* (i=1,2i = 1, 2). Consequently, if AA is a subset of E1E_1 such that dB1(A)d_{B_1}(A) is convex, balanced and compact for σ(E1′∗,E1′)\sigma(E'_1*, E'_1) then dB2(u(A))=v(dB1(A))d_{B_2}(u(A)) = v(d_{B_1}(A)) is convex, balanced and compact for σ(E2′∗,E2′)\sigma(E'_2*, E'_2) since the topologies σ(E1′∗,E1′)\sigma(E'_1*, E'_1) and σ(E2′∗,E2′)\sigma(E'_2*, E'_2) are Hausdorff.

Proposition 7

— Let u:E1→E2u : E_1 \to E_2 be a linear mapping. We assume that uu is continuous for the weakened topologies of E1E_1 and E2E_2.

(i) The mapping uu is continuous if E1E_1 and E2E_2 are assigned their Mackey topologies.

(ii) If E1E_1 is bornological or barrelled, then uu is continuous for the initial topologies of E1E_1 and E2E_2.

(iii) In order that uu be continuous for the initial topologies of E1E_1 and E2E_2, it is necessary and sufficient that the image under ′u'u of every equicontinuous subset of E2′E_2' be equicontinuous in E1′E_1'.

The hypothesis implies that ′u'u is continuous for the weak topologies σ(E2′,E2)\sigma(E_2', E_2) and σ(E1′,E1)\sigma(E_1', E_1) (II, p. 46, corollary) hence the image under ′u'u of a convex, balanced and compact subset for σ(E2′,E2)\sigma(E_2', E_2) is convex, balanced and compact for σ(E1′,E1)\sigma(E_1', E_1), the topologies σ(E2′,E2)\sigma(E_2', E_2) and σ(E1′,E1)\sigma(E_1', E_1) being Hausdorff. Therefore, assertion (i) follows from GT, X, § 1, No. 4, prop. 3, b). Assertion (ii) is a consequence of (i) : for, if E1E_1 is bornological or barrelled, its initial topology is the Mackey topology, and the Mackey topology of E2E_2 is finer than the initial topology of E2E_2. Finally, the initial topology of EiE_i is that of uniform convergence on equicontinuous subsets of Ei′E_i' (III, p. 19, cor. 1 of prop. 7). This proves (iii).

Corollary

— Suppose E1E_1 is a normed space. Let uu be a linear mapping from E1E_1 into E2E_2. The following properties are equivalent :

a) uu is continuous;

b) uu is continuous for the weakened topologies;

c) the image of the unit ball in E1E_1 under uu is bounded in E2E_2;

d) for every sequence (xn)(x_n) of points of E1E_1 tending to 0 for the initial topology, the sequence (u(xn))(u(x_n)) is bounded for the weakened topology of E2E_2.

Since E1E_1 is bornological the equivalence of a) and b) follows from prop. 7; that of a) and c) is immediate. The equivalence of a) and d) follows from prop. 1 of IV, p. 1, and from prop. 1 of III, p. 11.

Proposition 8

— (i) Let EE be a normed space, with dual E′E'. For every x∈Ex \in E, we have

∥x∥=sup⁡x′∈E′,∥x′∥≤1∣⟨x,x′⟩∣. \|x\| = \sup_{x' \in E', \|x'\| \leq 1} |\langle x, x' \rangle|.

(ii) Let E1E_1 and E2E_2 be two normed spaces and uu a continuous linear mapping from E1E_1 into E2E_2. We have

∥u∥=∥tu∥. \|u\| = \|t u\|.

Let x∈Ex \in E. For every x′∈E′x' \in E' such that ∥x′∥≤1\|x'\| \leq 1, we have

∣⟨x,x′⟩∣≤∥x∥⋅∥x′∥≤∥x∥. |\langle x, x' \rangle| \leq \|x\| \cdot \|x'\| \leq \|x\|.

By Hahn-Banach theorem (II, p. 23, cor. 2), there exists an element x′x' in E′E' such that ∥x′∥≤1\|x'\| \leq 1 and ⟨x,x′⟩=∥x∥\langle x, x' \rangle = \|x\|. This proves (i).

Let us now prove (ii). By formula (3) and the definition of the transpose, we have

∥tu∥=sup⁡∥y′∥≤1∥tu(y′)∥=sup⁡∥y′∥≤1,∥x∥≤1∣⟨x,tu(y′)⟩∣=sup⁡∥x∥≤1,∥y′∥≤1∣⟨u(x),y′⟩∣=sup⁡∥x∥≤1∥u(x)∥=∥u∥. \|^{t}u\| = \sup_{\|y'\| \leq 1} \|^{t}u(y')\| = \sup_{\|y'\| \leq 1, \|x\| \leq 1} |\langle x, ^{t}u(y') \rangle| = \sup_{\|x\| \leq 1, \|y'\| \leq 1} |\langle u(x), y' \rangle| = \sup_{\|x\| \leq 1} \|u(x)\| = \|u\|.

Remark 1

Formula (3) is a particular case of (4), corresponding to the linear mapping λ↦λx\lambda \mapsto \lambda x from K into E.

Remark 2

Put B(x,y′)=⟨u(x),y′⟩=⟨x,tu(y′)⟩B(x, y') = \langle u(x), y' \rangle = \langle x, ^{t}u(y') \rangle for x∈E1, y′∈E2′x \in E_1,\ y' \in E'_2. The above proof shows that B is a continuous bilinear form on E1×E2′E_1 \times E'_2, with norm (GT, X, § 3, No. 2) equal to ∥u∥\|u\|.

Corollary

— *Let E˙\dot{E} be a normed space satisfying the first axiom of countability. There exists a countable subset D of E′−{0}E' - \{0\} such that we have

∥x∥=sup⁡ξ∈D∣⟨x,ξ⟩∣/∥ξ∥ \|x\| = \sup_{\xi \in D} |\langle x, \xi \rangle| / \|\xi\|

for all x∈Ex \in E.

Let B' be the unit ball of the dual E′E' of E with the weak topology σ(E′,E)\sigma(E', E) assigned to it. Then B' is a compact metrizable space (III, p. 19, cor. 2); hence there exists a countable dense subset D' in B'. Put D=D′∩(E′−{0})D = D' \cap (E' - \{0\}). Let x∈Ex \in E; the mapping x′↦⟨x,x′⟩x' \mapsto \langle x, x' \rangle from B' into K is continuous, therefore

sup⁡x′∈B′∣⟨x,x′⟩∣=sup⁡ξ∈D′∣⟨x,ξ⟩∣≤sup⁡ξ∈D∣⟨x,ξ⟩∣/∥ξ∥≤∥x∥. \sup_{x' \in B'} |\langle x, x' \rangle| = \sup_{\xi \in D'} |\langle x, \xi \rangle| \leq \sup_{\xi \in D} |\langle x, \xi \rangle| / \|\xi\| \leq \|x\|.

Formula (5) now follows from (3).

4. Dual of a quotient space and of a subspace

Throughout this section, E denotes a locally convex space, M a vector subspace of E, and M∘M^\circ the orthogonal of M in the dual E′E' of E. Let pp be the canonical mapping from E onto E/ME/M; then tp^{t}p is injective, with image M∘M^\circ, hence defines a vector space isomorphism (not topological)

π:(E/M)′→M∘. \pi : (E/M)' \to M^\circ .

Similarly, let ii be the canonical injection from M into E. Then ti^{t}i is surjective (II, p. 24, prop. 2); its kernel is equal to M∘M^\circ, and we get a vector space isomorphism (not topological)

i:E′/M∘→M′. i : E'/M^\circ \to M'.

Proposition 9

— (i) For a subset A of (E/M)′(E/M)' to be equicontinuous, it is necessary and sufficient that π(A)\pi(A) is an equicontinuous subset of E′E'.

(ii) Let S\mathfrak{S} be a set of bounded subsets of EE, and S1\mathfrak{S}_1 the set of the images of subsets A∈SA \in \mathfrak{S} in E/ME/M. Then π\pi is an isomorphism from (E/M)S1′(E/M)'_{\mathfrak{S}_1} onto M∘M^\circ, where M∘M^\circ is assigned the topology induced by that of ES′E'_{\mathfrak{S}}.

(iii) Suppose EE is a normed space then π\pi is an isometry from the normed space (E/M)′(E/M)' onto the normed subspace M∘M^\circ of E′E'.

Let AA be a subset of (E/M)′(E/M)' and B=tp(A)⊂E′B = {}^t p(A) \subset E'. Put

q(ξ)=sup⁡ξ′∈A∣⟨ξ,ξ′⟩∣ q(\xi) = \sup_{\xi' \in A} |\langle \xi, \xi' \rangle|
for all ξ∈E/M\xi \in E/M. In order that AA be equicontinuous, it is necessary and sufficient that the mapping qq from E/ME/M into R‾+\overline{\mathbf{R}}_+ is a continuous semi-norm. This implies that q∘pq \circ p is a continuous semi-norm on EE (II, p. 27, prop. 5, (ii)). Since we have
(q∘p)(x)=sup⁡x′∈B∣⟨x,x′⟩∣ (q \circ p)(x) = \sup_{x' \in B} |\langle x, x' \rangle|
for all x∈Ex \in E, this in turn implies that BB is equicontinuous in E′E', and (i) follows.

Let A∈SA \in \mathfrak{S} and let ff be a continuous linear form on E/ME/M. For every λ∈R+\lambda \in \mathbf{R}_+, we have ∣f∣≤λ|f| \leq \lambda on p(A)p(A) if and only if ∣tp(f)∣≤λ|{}^t p(f)| \leq \lambda on AA; hence (ii).

Finally we prove (iii). Let y′y' be in (E/M)′(E/M)'. An element in E/ME/M has norm <1< 1 if and only if it is the image under pp of an element of norm <1< 1 in EE. Hence

∥y′∥=sup⁡y∈E/M,∥y∥<1∣⟨y,y′⟩∣=sup⁡x∈E,∥x∥<1∣⟨p(x),y′⟩∣=sup⁡x∈E,∥x∥<1∣⟨x,tp(y′)⟩∣=∥tp(y′)∥, \begin{align*} \|y'\| &= \sup_{y \in E/M, \|y\| < 1} |\langle y, y' \rangle| = \sup_{x \in E, \|x\| < 1} |\langle p(x), y' \rangle| \\ &= \sup_{x \in E, \|x\| < 1} |\langle x, {}^t p(y') \rangle| = \|{}^t p(y')\|, \end{align*}
and tp{}^t p induces an isometry from (E/M)′(E/M)' onto M∘M^\circ.

Proposition 10

— (i) For a subset AA of M′M' to be equicontinuous, it is necessary and sufficient that it is the image under ti{}^t i of an equicontinuous subset of E′E'.

(ii) Suppose MM is closed in EE. Let S\mathfrak{S} be a covering of EE consisting of bounded subsets and let S1\mathfrak{S}_1 be the set of subsets of MM of the form M∩AM \cap A for AA in S\mathfrak{S}. The bijective linear mapping ii from ES′/M∘E'_{\mathfrak{S}}/M^\circ onto MS1′M'_{\mathfrak{S}_1} is continuous. It is a homeomorphism if S\mathfrak{S} is a directed set for the relation ⊂\subset and consists of closed convex and compact sets for σ(E,E′)\sigma(E, E').

(iii) If EE is assumed to be normed, then ii is an isometry from E′/M∘E'/M^\circ onto M′M'.

The image under ti{}^t i of an equicontinuous subset of E′E' is an equicontinuous subset of M′M' (IV, p. 7, prop. 7). Conversely, let AA be an equicontinuous subset of M′M'. The topology of MM is defined by the set of restrictions to MM of the continuous semi-norms on EE. Hence there exists a continuous semi-norm pp on EE such that ∣f(x)∣≤p(x)|f(x)| \leq p(x) for all f∈Af \in A and for all x∈Mx \in M. Let BB be the set of all linear forms gg on EE such that ∣g∣≤p|g| \leq p and whose restriction to MM belongs to AA. The set BB is equicontinuous in E′E'; by Hahn-Banach theorem (II, p. 23, cor. 1), we have ti(B)=A{}^t i(B) = A, hence (i) follows.

We now prove (ii). By prop. 6 of IV, p. 6, the linear mapping ti{}^t i from ES′E'_{\mathfrak{S}} into MS1′M'_{\mathfrak{S}_1} is continuous, and defines, by passing to the quotient, a continuous linear mapping ι\iota from ES′/M∘E'_\mathscr{S}/M^\circ onto MS1′M'_{\mathscr{S}_1}. Let T\mathscr{T} be the topology on M′M' obtained by transferring that of ES′/M∘E'_\mathscr{S}/M^\circ by ι\iota; this is finer than the S1\mathscr{S}_1-topology.

Suppose now that S\mathscr{S} is a directed set for ⊂\subset and consists of closed, convex, balanced and compact sets for σ(E,E′)\sigma(E, E'). To show that ι\iota is a homeomorphism, i.e. that T\mathscr{T} is coarser than the S1\mathscr{S}_1-topology on M′M', it is enough to prove that T\mathscr{T} is compatible with the duality between M′M' and MM and that every equicontinuous set in MM (considered as the dual of MM with T\mathscr{T}) is contained in the homothetic of a set belonging to S1\mathscr{S}_1. Since T\mathscr{T} is finer than the S1\mathscr{S}_1-topology and S1\mathscr{S}_1 is a covering of MM, the linear form y′↦⟨y,y′⟩y' \mapsto \langle y, y' \rangle on M′M' is continuous for T\mathscr{T} for every y∈My \in M. Let ff be a linear form on M′M' which is continuous for T\mathscr{T}; then f∘ιf \circ \iota is a continuous linear form on ES′E'_\mathscr{S}. The S\mathscr{S}-topology on E′E' is coarser than the Mackey topology τ(E′,E)\tau(E', E); for, the mapping dB:E→E′∗d_B : E \to {E'}^* is continuous for the topologies σ(E,E′)\sigma(E, E') and σ(E′∗,E′)\sigma({E'}^*, E'), and since the latter is Hausdorff, the image under dBd_B of a set which is compact for σ(E,E′)\sigma(E, E') is compact for σ(E′∗,E′)\sigma({E'}^*, E'). By lemma 1 of IV, p. 3, there exists x0∈Ex_0 \in E such that f(ι(x′))=⟨x0,x′⟩f(\iota(x')) = \langle x_0, x' \rangle for all x′∈E′x' \in E'. In particular, ⟨x0,x′⟩=0\langle x_0, x' \rangle = 0 for all x′∈M∘x' \in M^\circ, and since MM is closed in EE, we have x0∈Mx_0 \in M (II, p. 45, cor. 2); and finally, f(y′)=⟨x0,y′⟩f(y') = \langle x_0, y' \rangle for all y′∈M′y' \in M'. This proves that T\mathscr{T} is compatible with the duality between MM and M′M'.

Now let AA be a subset of MM equicontinuous for the topology T\mathscr{T} on M′M'. By the definition of T\mathscr{T}, and in view of the hypothesis that S\mathscr{S} is directed, this means that there exists a set B∈SB \in \mathscr{S} containing 0 and such that the upper bound λ\lambda of the numbers ∣⟨y,x′⟩∣|\langle y, x' \rangle| for y∈Ay \in A and x′∈B∘x' \in B^\circ, is finite (III, p. 19, prop. 7). Since BB is closed in EE, the theorem of bipolars (II, p. 44, th. 1) shows that we have A⊂λ(B∩M)A \subset \lambda(B \cap M); this completes the proof of (ii).

We shall now prove (iii). Let y′∈M′y' \in M'. We shall prove the formula

∥y′∥=inf⁡ι(x′)=y′∥x′∥. \|y'\| = \inf_{\iota(x') = y'} \|x'\|.

By prop. 8, (ii) of IV, p. 7, we have ∥ι∥=∥i∥\|\iota\| = \|i\|, and so ∥ι∥≤1\|\iota\| \leq 1, and

∥y′∥≤inf⁡ι(x′)=y′∥x′∥. \|y'\| \leq \inf_{\iota(x') = y'} \|x'\|.

By Hahn-Banach theorem (II, p. 23, cor. 3), there exists a linear form x0′x'_0 on EE which extends y′y' and is of the same norm; hence we get the inequality opposite to (7), since ι(x0′)=y′\iota(x'_0) = y'.

Remark

— We know (II, p. 48, prop. 7, (ii)) that ι\iota is a topological vector space isomorphism from Es′/M∘E'_s/M^\circ onto Ms′M'_s (weak duals). For the topology of compact convex convergence, prop. 10 shows that ι\iota is an isomorphism from Ecc′/M∘E'_{cc}/M^\circ onto Mcc′M'_{cc} when EE is Hausdorff and MM closed in EE. For the strong topologies, ι\iota is a continuous mapping from Eb′/M∘E'_b/M^\circ onto Mb′M'_b; it is an isomorphism if EE is a Banach space \* or if EE is semi-reflexive and MM is closed in EE (IV, p. 15) *, but this is not always so if EE is a Fréchet space (IV, p. 58, exerc. 5, c)).

Proposition 11

— (i) The weakened topology on E/ME/M is the quotient of that on EE; the weakened topology on MM is induced by that of EE.

(ii) The Mackey topology on E/M is the quotient of that on E; the Mackey topology on M is finer than the topology induced by τ(E, E').

Assertion (i) follows from prop. 7 of II, p. 48.

The canonical injection i:M→Ei : M \to E is continuous for the weakened topologies, hence for the Mackey topologies τ(M,M′)\tau(M, M') and τ(E,E′)\tau(E, E') (IV, p. 7, prop. 7). Similarly, the canonical projection p:E→E/Mp : E \to E/M is continuous for the Mackey topologies. We see immediately that the quotient topology on E/ME/M obtained from τ(E,E′)\tau(E, E') is compatible with the duality between E/ME/M and (E/M)′(E/M)', hence is coarser than the Mackey topology on E/ME/M, by Mackey's theorem (IV, p. 2, th. 1). This proves (ii).

5. Dual of a direct sum and of a product

For every i∈Ii \in I, let (Ei,Fi)(E_i, F_i) be a pair of vector spaces, set in duality by a bilinear form BiB_i. We put E=∏i∈IEiE = \prod_{i \in I} E_i and F=⨁i∈IFiF = \bigoplus_{i \in I} F_i, and we identify each FiF_i with a subspace of FF. We put EE and FF in duality by means of the bilinear form

B(x,y)=∑i∈IBi(xi,yi)forx=(xi)andy=(yi) B(x, y) = \sum_{i \in I} B_i(x_i, y_i) \quad \text{for} \quad x = (x_i) \quad \text{and} \quad y = (y_i)

(the family (Bi(xi,yi))i∈I(B_i(x_i, y_i))_{i \in I} has finite support).

We recall (II, p. 50, prop. 8) that the weak topology σ(E,F)\sigma(E, F) is the product of the weak topologies σ(Ei,Fi)\sigma(E_i, F_i).

Lemma 3

— (i) *For every i∈Ii \in I, let Si\mathfrak{S}_i be a family of subsets of FiF_i, which is bounded for σ(Fi,Ei)\sigma(F_i, E_i); put S=⋃i∈ISi\mathfrak{S} = \bigcup_{i \in I} \mathfrak{S}_i. Then the S\mathfrak{S}-topology on EE is the product of the Si\mathfrak{S}_i-topologies on the EiE_i.

(ii) For every i∈Ii \in I, let Ji\mathfrak{J}_i be an adapted bornology on the space EiE_i endowed with the weak topology σ(Ei,Fi)\sigma(E_i, F_i), none equal to {∅}\{ \varnothing \}. Let J\mathfrak{J} be the family of subsets AA of E=∏i∈IEiE = \prod_{i \in I} E_i such that pr⁡i(A)∈Ji\operatorname{pr}_i(A) \in \mathfrak{J}_i for all i∈Ii \in I. Then the J\mathfrak{J}-topology on FF is the direct sum of the Ji\mathfrak{J}_i-topologies on the FiF_i.

Let T\mathscr{T} be the product of the Si\mathfrak{S}_i-topologies. The sets of the form

A=∏i∈JAi∘×∏i∈I−JEi A = \prod_{i \in J} A_i^\circ \times \prod_{i \in I - J} E_i

where J⊂IJ \subset I is finite and Ai∈SiA_i \in \mathfrak{S}_i for all i∈Ji \in J, form a fundamental system of neighbourhoods of 0 for T\mathscr{T}. We have A=(⋃i∈JAi)∘A = (\bigcup_{i \in J} A_i)^\circ, hence T\mathscr{T} is identical with the S\mathfrak{S}-topology. This proves (i).

We assign the J\mathfrak{J}-topology to FF and the Ji\mathfrak{J}_i-topology to each FiF_i. For every subset AA of EE, we have Fi∩A∘=pr⁡i(A)∘F_i \cap A^\circ = \operatorname{pr}_i(A)^\circ, hence the injection from FiF_i into FF is continuous. Let qq be a semi-norm on FF; we assume that the restriction qiq_i of qq to FiF_i is continuous for all i∈Ii \in I. Then we can find non-empty subsets Ai∈JiA_i \in \mathfrak{J}_i such that we have

qi(yi)≤sup⁡xi∈Ai∣Bi(xi,yi)∣(yi∈Fi). q_i(y_i) \leq \sup_{x_i \in A_i} |B_i(x_i, y_i)| \quad (y_i \in F_i) .

Put A=∏i∈IAiA = \prod_{i \in I} A_i; then A∈JA \in \mathfrak{J}. For y=(yi)i∈Iy = (y_i)_{i \in I} in FF, we have

q(y)≤∑i∈Iqi(yi)≤∑i∈Isup⁡xi∈Ai∣Bi(xi,yi)∣=sup⁡x∈A∣B(x,y)∣, q(y) \leq \sum_{i \in I} q_i(y_i) \leq \sum_{i \in I} \sup_{x_i \in A_i} |B_i(x_i, y_i)| = \sup_{x \in A} |B(x, y)|,
where the last equality follows from (8) since the family (yi)i∈I(y_i)_{i \in I} has finite support and the AiA_i are non-empty and can be assumed balanced (GT, IV, § 5, No. 7, cor. 2 to prop. 12). This inequality proves that qq is continuous on FF, and hence (ii).

Proposition 12

The topology β(F,E)\beta(F, E) is the direct sum of the topologies β(Fi,Ei)\beta(F_i, E_i). The topology β(E,F)\beta(E, F) is the product of the topologies β(Ei,Fi)\beta(E_i, F_i).

We shall apply lemma 3 taking for Si\mathfrak{S}_i the family of all subsets of FiF_i which are bounded for σ(Fi,Ei)\sigma(F_i, E_i) and for Ji\mathfrak{J}_i the family of all subsets of EiE_i which are bounded for σ(Ei,Fi)\sigma(E_i, F_i).

By cor. 2 of III, p. 4, J\mathfrak{J} is the family of all subsets of EiE_i which are bounded for the product topology of the σ(Ei,Fi)\sigma(E_i, F_i), which is identical with σ(E,F)\sigma(E, F). Hence our assertion on β(F,E)\beta(F, E) follows.

We endow F=⨁i∈IFiF = \bigoplus_{i \in I} F_i with the topology T\mathscr{T} which is the direct sum of the topologies σ(Fi,Ei)\sigma(F_i, E_i). Then the dual of FF consists of the linear forms y↦B(x,y)y \mapsto B(x, y) where xx runs through EE (II, p. 30, prop. 6). By prop. 1 of IV, p. 1, the topologies T\mathscr{T} and σ(F,E)\sigma(F, E) have the same bounded sets. Assume first that that the topologies σ(Fi,Ei)\sigma(F_i, E_i) are Hausdorff. By prop. 5 of III, p. 5, these sets are contained in a subset of the form ∑i∈JBi\sum_{i \in J} B_i with J⊂IJ \subset I finite and BiB_i bounded in FiF_i (for σ(Fi,Ei)\sigma(F_i, E_i)) for all i∈Ji \in J. Since ∑i∈JBi\sum_{i \in J} B_i is contained in the convex envelope of ⋃i∈JnBi\bigcup_{i \in J} nB_i, where n=Card(J)n = \mathrm{Card}(J), we can apply lemma 3, to prove the assertion on β(E,F)\beta(E, F) in this case.

For the general case, let NiN_i be the intersection of all neighbourhoods of 0 for σ(Fi,Ei)\sigma(F_i, E_i), and let N=∑i∈INiN = \sum_{i \in I} N_i; then F/NF/N is the topological direct sum of the Fi/NiF_i/N_i (II, p. 31, prop. 8); we deduce from this that every subset of FF which is bounded for T\mathscr{T} is contained in a set of the form N+∑i∈JBiN + \sum_{i \in J} B_i with J⊂IJ \subset I finite and BiB_i bounded in FiF_i for all i∈Ji \in J (III, p. 2, Remark 3); since the polar of this set in EE is the same as that of ∑i∈JBi\sum_{i \in J} B_i, the result follows as above.

Proposition 13

The Mackey topology τ(F,E)\tau(F, E) is the direct sum of the Mackey topologies τ(Fi,Ei)\tau(F_i, E_i). The topology τ(E,F)\tau(E, F) is the product of the topologies τ(Ei,Fi)\tau(E_i, F_i).

The assertion on τ(F,E)\tau(F, E) follows from lemma 3 (ii) and the following property : for a closed, convex and balanced subset of F∗=∏i∈IFi∗F^* = \prod_{i \in I} F_i^* to be compact for σ(F∗,F)\sigma(F^*, F), it is necessary and sufficient that its projection on each Fi∗F_i^* is compact for σ(Fi∗,Fi)\sigma(F_i^*, F_i).

To prove the assertion on τ(E,F)\tau(E, F), assume first that the topologies σ(Fi,Ei)\sigma(F_i, E_i) are Hausdorff, it is enough (lemma 3 (i)) to prove that every subset AA of FF which is convex, balanced and compact for σ(F,E)\sigma(F, E) is contained in a set of the form ∑i∈JAi\sum_{i \in J} A_i where J⊂IJ \subset I is finite and where AiA_i is convex, balanced and compact for σ(Fi,Ei)\sigma(F_i, E_i). But such a subset is bounded for σ(F,E)\sigma(F, E). By the proof of prop. 12, there exists a finite subset JJ of II such that A⊂∑i∈JFiA \subset \sum_{i \in J} F_i, and it is enough to take for AiA_i the projection of AA on FiF_i.

In the general case, with the same notations as in the proof of prop. 12, we have τ(Ei,Fi)=τ(Ei,Fi/Ni)\tau(E_i, F_i) = \tau(E_i, F_i/N_i) and τ(E,F)=τ(E,F/N)\tau(E, F) = \tau(E, F/N) (IV, p. 2) and since F/NF/N is the topological direct sum of the Fi/NiF_i/N_i, we have reduced to the preceding case.

Q.E.D.

For the remainder of this paragraph, we assume that (Ei)i∈I(E_i)_{i \in I} is a family of locally convex spaces. Let SS denote the topological direct sum of the EiE_i and PP, their product. We define a linear mapping θ:S′→∏i∈IEi′\theta : S' \to \prod_{i \in I} E_i', said to be canonical, by

θ(x′)=(x′∣Ei)i∈I(x′∈S′) \theta(x') = (x'|E_i)_{i \in I} \quad (x' \in S')

(where S′S' denotes the dual of SS, and Ei′E_i' that of EiE_i).

Proposition 14

(i) *The mapping θ\theta is an isomorphism from the strong (resp. weak) dual of S=⨁i∈IEiS = \bigoplus_{i \in I} E_i onto the product of the strong (resp. weak) duals of the EiE_i:

(ii) For a subset AA of S′S' to be equicontinuous, it is necessary and sufficient that the projection of θ(A)\theta(A) onto Ei′E_i' be equicontinuous for all i∈Ii \in I.

(iii) The Mackey topology τ(S,S′)\tau(S, S') is the direct sum of the Mackey topologies τ(Ei,Ei′)\tau(E_i, E_i').

(iv) The topology β(S,S′)\beta(S, S') is the direct sum of the topologies β(Ei,Ei′)\beta(E_i, E_i').

That θ\theta is bijective follows immediately from the definition of a topological direct sum (II, p. 30, prop. 6). Assertion (i) then follows from prop. 12 of IV, p. 12, for the strong topologies, and from prop. 8 of II, p. 50, for the weak topologies. Similarly (iii) follows from prop. 13 (IV, p. 12) and (iv) from prop. 12 (IV, p. 12).

To prove (ii), let AA be a subset of S′S'. Put

q(x)=sup⁡x′∈A∣⟨x,x′⟩∣forx∈S; q(x) = \sup_{x' \in A} |\langle x, x' \rangle| \quad \text{for} \quad x \in S ;

let qiq_i denote the restriction of qq to EiE_i, whence

qi(xi)=sup⁡xi′∈Ai∣⟨xi,xi′⟩∣forxi∈Ei, q_i(x_i) = \sup_{x_i' \in A_i} |\langle x_i, x_i' \rangle| \quad \text{for} \quad x_i \in E_i ,

where AiA_i denotes the projection of θ(A)\theta(A) on Ei′E_i'. For AA to be equicontinuous, it is necessary and sufficient that qq is finite (that is, that each qiq_i is finite) and continuous. In view of the characterization of continuous semi-norms on a topological direct sum (II, p. 27, prop. 5), this is the same as assuming that each qiq_i is continuous, or in fact, that each set AiA_i is equicontinuous.

Let ϕ\phi be the linear mapping, said to be canonical, from ⨁i∈IEi′\bigoplus_{i \in I} E_i' into the dual P′P' of P=∏i∈IEiP = \prod_{i \in I} E_i, defined by the formula

⟨x,ϕ(x′)⟩=∑i∈I⟨xi,xi′⟩ \langle x, \phi(x') \rangle = \sum_{i \in I} \langle x_i, x'_i \rangle

for x=(xi)x = (x_i) in PP and x′=(xi′)x' = (x'_i) in ⨁i∈IEi′\bigoplus_{i \in I} E'_i.

Proposition 15

— (i) *The map ϕ\phi is an isomorphism from the topological direct sum of the strong duals of the EiE_i onto the strong dual of P=∏i∈IEiP = \prod_{i \in I} E_i.

(ii) For a subset AA of P′P' to be equicontinuous, it is necessary and sufficient that it is contained in a finite sum ∑i∈Jϕ(Ai)\sum_{i \in J} \phi(A_i), where J⊂IJ \subset I is finite and where AiA_i is equicontinuous in Ei′E'_i for all i∈Ji \in J.

(iii) The Mackey topology τ(P,P′)\tau(P, P') is the product of the topologies τ(Ei,Ei′)\tau(E_i, E'_i).

(iv) The topology β(P,P′)\beta(P, P') is the product of the topologies β(Ei,Ei′)\beta(E_i, E'_i).

It is immediate that ϕ\phi is injective. A fundamental system of neighbourhoods of 0 in PP consists of sets of the form V=∏i∈JVi×∏i∈I−JEiV = \prod_{i \in J} V_i \times \prod_{i \in I - J} E_i, where J⊂IJ \subset I is finite and ViV_i is a neighbourhood of 0 in EiE_i for ii in JJ. The polar of VV in P′P' is equal to ∑i∈Jϕ(Vi0)\sum_{i \in J} \phi(V_i^0).

This proves the surjectivity of ϕ\phi and also assertion (ii).

Assertions (i) and (iv) follow from prop. 12 (IV, p. 12) and (iii) from prop. 13 (IV, p. 12).

Corollary

— *Every product of barrelled spaces is barrelled.

A locally convex space EE is barrelled if and only if the initial topology is identical with β(E,E′)\beta(E, E') (IV, p. 4, Remark 3). Hence it is enough to apply prop. 15 (iv).

Exercises

See the exercises for § 1.

Transcribed from the printed volume. Not translated.