in the subsequent ones, but a priori we do not know if En is isomorphically embedded in E, i.e. if the topology induced by ⌧ind on En is identical to the topology ⌧n initially given on En. This is indeed true and it will be a consequence of the following lemma.
Lemma 1.3.3. Let X be a locally convex t.v.s., X0 a linear subspace of X equipped with the subspace topology, and U a convex neighbourhood of the origin in X0. Then there exists a convex neighbourhood V of the origin in X such that V \X0=U.
Proof.
AsX0 carries the subspace topology induced by X, there exists a neighbour- hoodW of the origin inXsuch thatU =W\X0. SinceX is a locally convex t.v.s., there exists a convex neighbourhood W0 of the origin in X such that W0✓W. LetV be the convex hull ofU[W0. Then by construction we have that V is a convex neighbourhood of the origin in X and that U ✓V which implies U =U \X0 ✓V \X0. We claim that actually V \X0 =U. Indeed, let x 2 V \X0; as x 2 V and as U and W0 are both convex, we may write x=ty+ (1 t)z with y2U, z 2W0 and t2[0,1]. If t= 1, then x=y 2U and we are done. If 0t <1, then z= (1 t) 1(x ty) belongs toX0 and so z 2 W0 \X0 ✓ W \X0 = U. This implies, by the convexity of U, that x2U. Hence,V \X0 ✓U.
Proposition 1.3.4.
Let (E,⌧ind) be an LF-space with defining sequence {(En,⌧n) :n2N}. Then
⌧ind En⌘⌧n,8n2N. Proof.
(✓) LetU be a neighbourhood of the origin in (E,⌧ind). Then, by definition of ⌧ind, there existsV convex, balanced and absorbing neighbourhood of the origin in (E,⌧ind) s.t. V ✓U and, for eachn2N,V \En is a neighbourhood of the origin in (En,⌧n). Hence, ⌧ind En✓⌧n,8n2N.
(◆) Givenn2N, let Un be a convex, balanced, absorbing neighbourhood of the origin in (En,⌧n). Since En is a linear subspace ofEn+1, we can apply Lemma 1.3.3 (for X = En+1, X0 = En and U = Un) which ensures the existence of a convex neighbourhood Un+1 of the origin in (En+1,⌧n+1) such that Un+1\En =Un. Then, by induction, we get that for any k2 N there exists a convex neighbourhood Un+k of the origin in (En+k,⌧n+k) such that Un+k\En+k 1 =Un+k 1. Hence, for anyk2N, we get Un+k\En =Un. If we consider now U := S1
k=1Un+k, then U \En = Un. Furthermore, U is a
neighbourhood of the origin in (E,⌧ind) sinceU\Emis a neighbourhood of the origin in (Em,⌧m) for all m 2N. We can then conclude that⌧n✓⌧ind En, 8n2N.
From the previous proposition we can easily deduce that any LF-space is not only a locally convex t.v.s. but also Hausdor↵. Indeed, if (E,⌧ind) is the LF-space with defining sequence {(En,⌧n) : n 2 N} and we denote by F(o) [resp. Fn(o)] the filter of neighbourhoods of the origin in (E,⌧ind) [resp. in (En,⌧n)], then:
\
V2F(o)
V = \
V2F(o)
V \ [
n2N
En
!
= [
n2N
\
V2F(o)
(V \En) = [
n2N
\
Un2Fn(o)
Un={o},
which implies that (E,⌧ind) is Hausdor↵by Corollary 2.2.4 in TVS-I.
As a particular case of Proposition1.3.1 we get that:
Proposition 1.3.5.
Let (E,⌧ind) be an LF-space with defining sequence {(En,⌧n) : n 2 N} and (F,⌧) an arbitrary locally convex t.v.s..
1. A linear mappingu fromE intoF is continuous if and only if, for each n2N, the restriction u En of u to En is continuous.
2. A linear form on E is continuous if and only if its restrictions to each En are continuous.
Note that Propositions1.3.4and1.3.5hold for any countable strict induc- tive limit of an increasing sequences of locally convex Hausdor↵ t.v.s. (even when they are not Fr´echet).
The following results is instead typical of LF-spaces as it heavily relies on the completeness of the t.v.s. of the defining sequence.
Theorem 1.3.6. Any LF-space is complete.
Proof.
Let (E,⌧ind) be an LF-space with defining sequence{(En,⌧n) :n2N}. LetF be a Cauchy filter on (E,⌧ind). Denote byFE(o) the filter of neighbourhoods of the origin in (E,⌧ind) and consider
G :={A✓E: A◆M +V for someM 2F, V 2FE(o)}. 1) G is a filter on E.
Indeed, it is clear from its definition that G does not contain the empty set and that any subset ofE containing a set inG has to belong toG. Moreover,
for anyA1, A2 2Gthere existM1, M22F,V1, V2 2FE(o) s.t. M1+V1 ✓A1 and M2+V2 ✓A2; and therefore
A1\A2 ◆(M1+V1)\(M2+V2)◆(M1\M2) + (V1\V2).
The latter proves that A1\A22G sinceF and FE(o) are both filters and so M1\M2 2F and V1\V2 2FE(o).
2)G ✓F.
In fact, for anyA2G there existM 2F and V 2FE(o) s.t.
A◆M+V M +{0}=M which implies that A2F sinceF is a filter.
3)G is a Cauchy filter on E.
Let U 2 FE(o). Then there always exists V 2 FE(o) balanced such that V +V V ✓U. AsF is a Cauchy filter on (E,⌧ind), there existsM 2F such that M M ✓V. Then
(M+V) (M+V)✓(M M) + (V V)✓V +V V ✓U which proves that G is a Cauchy filter since M+V 2G.
It is possible to show (and we do it later on) that:
9p2N: 8A2G, A\Ep6=; (1.13) This property ensures that the family
Gp :={A\Ep : A2G}
is a filter on Ep. Moreover, since G is a Cauchy filter on (E,⌧ind) and since by Proposition 1.3.4we have ⌧ind Ep=⌧p,Gp is a Cauchy filter on (Ep,⌧p).
Hence, the completeness ofEpguarantees that there existsx2Ep s.t. Gp !x.
This implies that alsoG !x and so FE(o)✓G✓F which gives F !x.
Proof. of (1.13)
Suppose that (1.13) is false, i.e. 8n2N, 9An2 G s.t. An\En=;. By the definition ofG, this means that
8n2N,9Mn2F, Vn2FE(o),s.t. (Mn+Vn)\En=;. (1.14)
Since E is a locally convex t.v.s., we may assume that each Vn is balanced and convex, and that Vn+1✓Vn for all n2N. Consider
Wn:=conv Vn[
n[1 k=1
(Vk\Ek)
! ,
then
(Wn+Mn)\En=;,8n2N.
Indeed, if there existsh2(Wn+Mn)\Enthenh2Enandh2(Wn+Mn). We may then write: h=x+ty+ (1 t)zwithx2Mn,y2Vn,z2V1\En 1and t2[0,1]. Hence,x+ty=h (1 t)z2En. But we also havex+ty2Mn+Vn, sinceVnis balanced and soty2Vn. Therefore,x+ty2(Mn+Vn)\Enwhich contradicts (1.14).
Now let us define
W :=conv [1 k=1
(Vk\Ek)
! .
As W is convex and as W \ Ek contains Vk \ Ek for all k 2 N, W is a neighbourhood of the origin in (E,⌧ind). Moreover, as (Vn)n2N is decreasing, we have that for alln2N
W =conv
n[1
k=1
(Vk\Ek)[ [1 k=n
(Vk\Ek)
!
✓conv
n[1
k=1
(Vk\Ek)[Vn
!
=Wn.
SinceF is a Cauchy filter on (E,⌧ind), there existsB 2Fsuch thatB B ✓W and soB B ✓Wn,8n2N. On the other hand we haveB\Mn6=;,8n2N, as both B and Mn belong to F. Hence, for all n2Nwe get
B (B\Mn)✓B B✓Wn, which implies
B ✓Wn+ (B\Mn)✓Wn+Mn and so
B\En✓(Wn+Mn)\En(1.14)= ;.
Therefore, we have got that B \En = ; for all n 2 N and so that B = ;, which is impossible asB 2F. Hence, (1.13) must hold true.
Example I: The space of polynomials
Letn2Nand x:= (x1, . . . , xn). Denote by R[x] the space of polynomials in the n variables x1, . . . , xn with real coefficients. A canonical algebraic basis forR[x] is given by all the monomials
x↵:=x↵11· · ·x↵nn, 8↵= (↵1, . . . ,↵n)2Nn0.
For any d 2 N0, let Rd[x] be the linear subpace of R[x] spanned by all monomials x↵ with |↵|:=Pn
i=1↵i d, i.e.
Rd[x] :={f 2R[x]|degf d}.
Since there are exactly n+dd monomials x↵ with|↵|d, we have that dim(Rd[x]) = (d+n)!
d!n! ,
and so that Rd[x] is a finite dimensional vector space. Hence, by Tychono↵
Theorem (see Corollary 3.1.4 in TVS-I) there is a unique topology ⌧ed that makesRd[x] into a Hausdor↵t.v.s. which is also complete and so Fr´echet (as it topologically isomorphic toRdim(Rd[x]) equipped with the euclidean topology).
As R[x] := S1
d=0Rd[x], we can then endow it with the inductive topol- ogy ⌧ind w.r.t. the family of F-spaces (Rd[x],⌧ed) :d2N0 ; thus (R[x],⌧ind) is a LF-space and the following properties hold (proof as Exercise 1, Sheet 3):
a) ⌧ind is the finest locally convex topology on R[x],
b) every linear map f from (R[x],⌧ind) into any t.v.s. is continuous.