Movatterモバイル変換


[0]ホーム

URL:


Jump to content
WikipediaThe Free Encyclopedia
Search

Complete topological vector space

From Wikipedia, the free encyclopedia
Structure in functional analysis

Infunctional analysis and related areas ofmathematics, acomplete topological vector space is atopological vector space (TVS) with the property that whenever points get progressively closer to each other, then there exists some pointx{\displaystyle x} towards which they all get closer. The notion of "points that get progressively closer" is made rigorous byCauchy nets orCauchy filters, which are generalizations ofCauchy sequences, while "pointx{\displaystyle x} towards which they all get closer" means that this Cauchynet or filterconverges tox.{\displaystyle x.} The notion of completeness for TVSs uses the theory ofuniform spaces as a framework to generalize the notion ofcompleteness for metric spaces. But unlike metric-completeness, TVS-completeness does not depend on any metric and is defined forall TVSs, including those that are notmetrizable orHausdorff.

Completeness is an extremely important property for a topological vector space to possess. The notions of completeness fornormed spaces andmetrizable TVSs, which are commonly defined in terms ofcompleteness of a particular norm or metric, can both be reduced down to this notion of TVS-completeness – a notion that is independent of any particular norm or metric. Ametrizable topological vector spaceX{\displaystyle X} with atranslation invariant metric[note 1]d{\displaystyle d} is complete as a TVS if and only if(X,d){\displaystyle (X,d)} is acomplete metric space, which by definition means that everyd{\displaystyle d}-Cauchy sequence converges to some point inX.{\displaystyle X.} Prominent examples of complete TVSs that are alsometrizable include allF-spaces and consequently also allFréchet spaces,Banach spaces, andHilbert spaces. Prominent examples of complete TVS that are (typically)not metrizable include strictLF-spaces such as thespace of test functionsCc(U){\displaystyle C_{c}^{\infty }(U)} with it canonical LF-topology, thestrong dual space of any non-normableFréchet space, as well as many otherpolar topologies oncontinuous dual space or othertopologies on spaces of linear maps.

Explicitly, atopological vector spaces (TVS) iscomplete if everynet, or equivalently, everyfilter, that isCauchy with respect to the space'scanonicaluniformity necessarily converges to some point. Said differently, a TVS is complete if its canonical uniformity is acomplete uniformity. Thecanonical uniformity on a TVS(X,τ){\displaystyle (X,\tau )} is the unique[note 2] translation-invariantuniformity that induces onX{\displaystyle X} the topologyτ.{\displaystyle \tau .} This notion of "TVS-completeness" dependsonly on vector subtraction and the topology of the TVS; consequently, it can be applied to all TVSs, including those whose topologies can not be defined in termsmetrics orpseudometrics. Afirst-countable TVS is complete if and only if every Cauchy sequence (or equivalently, everyelementary Cauchy filter) converges to some point.

Every topological vector spaceX,{\displaystyle X,} even if it is notmetrizable or notHausdorff, has acompletion, which by definition is a complete TVSC{\displaystyle C} into whichX{\displaystyle X} can beTVS-embedded as adensevector subspace. Moreover, every Hausdorff TVS has aHausdorff completion, which is necessarily uniqueup toTVS-isomorphism. However, as discussed below, all TVSs have infinitely many non-Hausdorff completions that arenot TVS-isomorphic to one another.

Definitions

[edit]
Main articles:Net (mathematics) andFilters in topology

This section summarizes the definition of a completetopological vector space (TVS) in terms of bothnets andprefilters. Information about convergence of nets and filters, such as definitions and properties, can be found in the article aboutfilters in topology.

Every topological vector space (TVS) is a commutativetopological group with identity under addition and the canonical uniformity of a TVS is definedentirely in terms of subtraction (and thus addition); scalar multiplication is not involved and no additional structure is needed.

Canonical uniformity

[edit]

Thediagonal ofX{\displaystyle X} is the set[1]ΔX =def {(x,x):xX}{\displaystyle \Delta _{X}~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{(x,x):x\in X\}} and for anyNX,{\displaystyle N\subseteq X,} thecanonical entourage/vicinity aroundN{\displaystyle N} is the setΔX(N)  =def {(x,y)X×X : xyN}=yX[(y+N)×{y}]=ΔX+(N×{0}){\displaystyle {\begin{alignedat}{4}\Delta _{X}(N)~&~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{(x,y)\in X\times X~:~x-y\in N\}\\&=\bigcup _{y\in X}[(y+N)\times \{y\}]\\&=\Delta _{X}+(N\times \{0\})\end{alignedat}}}where if0N{\displaystyle 0\in N} thenΔX(N){\displaystyle \Delta _{X}(N)} contains the diagonalΔX({0})=ΔX.{\displaystyle \Delta _{X}(\{0\})=\Delta _{X}.}

IfN{\displaystyle N} is asymmetric set (that is, ifN=N{\displaystyle -N=N}), thenΔX(N){\displaystyle \Delta _{X}(N)} issymmetric, which by definition means thatΔX(N)=(ΔX(N))op{\displaystyle \Delta _{X}(N)=\left(\Delta _{X}(N)\right)^{\operatorname {op} }} holds where(ΔX(N))op =def {(y,x):(x,y)ΔX(N)},{\displaystyle \left(\Delta _{X}(N)\right)^{\operatorname {op} }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{(y,x):(x,y)\in \Delta _{X}(N)\right\},} and in addition, this symmetric set'scomposition with itself is:ΔX(N)ΔX(N)  =def {(x,z)X×X :  there exists yX such that x,zy+N}=yX[(y+N)×(y+N)]=ΔX+(N×N).{\displaystyle {\begin{alignedat}{4}\Delta _{X}(N)\circ \Delta _{X}(N)~&~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{(x,z)\in X\times X~:~{\text{ there exists }}y\in X{\text{ such that }}x,z\in y+N\right\}\\&=\bigcup _{y\in X}[(y+N)\times (y+N)]\\&=\Delta _{X}+(N\times N).\end{alignedat}}}

IfL{\displaystyle {\mathcal {L}}} is any neighborhood basis at the origin in(X,τ){\displaystyle (X,\tau )} then thefamily of subsets ofX×X:{\displaystyle X\times X:}BL =def {ΔX(N):NL}{\displaystyle {\mathcal {B}}_{\mathcal {L}}~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{\Delta _{X}(N):N\in {\mathcal {L}}\right\}}is aprefilter onX×X.{\displaystyle X\times X.} IfNτ(0){\displaystyle {\mathcal {N}}_{\tau }(0)} is theneighborhood filter at the origin in(X,τ){\displaystyle (X,\tau )} thenBNτ(0){\displaystyle {\mathcal {B}}_{{\mathcal {N}}_{\tau }(0)}} forms abase of entourages for auniform structure onX{\displaystyle X} that is consideredcanonical.[2] Explicitly, by definition,thecanonical uniformity onX{\displaystyle X} induced by(X,τ){\displaystyle (X,\tau )}[2] is thefilterUτ{\displaystyle {\mathcal {U}}_{\tau }} onX×X{\displaystyle X\times X}generated by the above prefilter:Uτ =def BNτ(0) =def {SX×X : there exists NNτ(0) such that ΔX(N)S}{\displaystyle {\mathcal {U}}_{\tau }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~{\mathcal {B}}_{{\mathcal {N}}_{\tau }(0)}^{\uparrow }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{S\subseteq X\times X~:~{\text{there exists }}N\in {\mathcal {N}}_{\tau }(0){\text{ such that }}\Delta _{X}(N)\subseteq S\right\}}whereBNτ(0){\displaystyle {\mathcal {B}}_{{\mathcal {N}}_{\tau }(0)}^{\uparrow }} denotes theupward closure ofBNτ(0){\displaystyle {\mathcal {B}}_{{\mathcal {N}}_{\tau }(0)}} inX×X.{\displaystyle X\times X.} The same canonical uniformity would result by using a neighborhood basis of the origin rather the filter of all neighborhoods of the origin. IfL{\displaystyle {\mathcal {L}}} is any neighborhood basis at the origin in(X,τ){\displaystyle (X,\tau )} then the filter onX×X{\displaystyle X\times X} generated by the prefilterBL{\displaystyle {\mathcal {B}}_{\mathcal {L}}} is equal to the canonical uniformityUτ{\displaystyle {\mathcal {U}}_{\tau }} induced by(X,τ).{\displaystyle (X,\tau ).}

Cauchy net

[edit]
See also:Net (mathematics) andConvex series

The general theory ofuniform spaces has its own definition of a "Cauchy prefilter" and "Cauchy net". For the canonical uniformity onX,{\displaystyle X,} these definitions reduce down to those given below.

Supposex=(xi)iI{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i\in I}} is a net inX{\displaystyle X} andy=(yj)jJ{\displaystyle y_{\bullet }=\left(y_{j}\right)_{j\in J}} is a net inY.{\displaystyle Y.} The productI×J{\displaystyle I\times J} becomes adirected set by declaring(i,j)(i2,j2){\displaystyle (i,j)\leq \left(i_{2},j_{2}\right)} if and only ifii2{\displaystyle i\leq i_{2}} andjj2.{\displaystyle j\leq j_{2}.} Thenx×y =def (xi,yj)(i,j)I×J{\displaystyle x_{\bullet }\times y_{\bullet }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(x_{i},y_{j}\right)_{(i,j)\in I\times J}}denotes the (Cartesian)product net, where in particularx×x =def (xi,xj)(i,j)I×I.{\textstyle x_{\bullet }\times x_{\bullet }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(x_{i},x_{j}\right)_{(i,j)\in I\times I}.} IfX=Y{\displaystyle X=Y} then the image of this net under the vector addition mapX×XX{\displaystyle X\times X\to X} denotes thesum of these two nets:[3]x+y =def (xi+yj)(i,j)I×J{\displaystyle x_{\bullet }+y_{\bullet }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(x_{i}+y_{j}\right)_{(i,j)\in I\times J}}and similarly theirdifference is defined to be the image of the product net under the vector subtraction map(x,y)xy{\displaystyle (x,y)\mapsto x-y}:xy =def (xiyj)(i,j)I×J.{\displaystyle x_{\bullet }-y_{\bullet }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(x_{i}-y_{j}\right)_{(i,j)\in I\times J}.} In particular, the notationxx=(xi)iI(xi)iI{\displaystyle x_{\bullet }-x_{\bullet }=\left(x_{i}\right)_{i\in I}-\left(x_{i}\right)_{i\in I}} denotes theI2{\displaystyle I^{2}}-indexed net(xixj)(i,j)I×I{\displaystyle \left(x_{i}-x_{j}\right)_{(i,j)\in I\times I}} and not theI{\displaystyle I}-indexed net(xixi)iI=(0)iI{\displaystyle \left(x_{i}-x_{i}\right)_{i\in I}=(0)_{i\in I}} since using the latter as the definition would make the notation useless.

Anetx=(xi)iI{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i\in I}} in a TVSX{\displaystyle X} is called aCauchy net[4] ifxx =def (xixj)(i,j)I×I0 in X.{\displaystyle x_{\bullet }-x_{\bullet }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(x_{i}-x_{j}\right)_{(i,j)\in I\times I}\to 0\quad {\text{ in }}X.} Explicitly, this means that for every neighborhoodN{\displaystyle N} of0{\displaystyle 0} inX,{\displaystyle X,} there exists some indexi0I{\displaystyle i_{0}\in I} such thatxixjN{\displaystyle x_{i}-x_{j}\in N} for all indicesi,jI{\displaystyle i,j\in I} that satisfyii0{\displaystyle i\geq i_{0}} andji0.{\displaystyle j\geq i_{0}.} It suffices to check any of these defining conditions for any givenneighborhood basis of0{\displaystyle 0} inX.{\displaystyle X.} ACauchy sequence is a sequence that is also a Cauchy net.

Ifxx{\displaystyle x_{\bullet }\to x} thenx×x(x,x){\displaystyle x_{\bullet }\times x_{\bullet }\to (x,x)} inX×X{\displaystyle X\times X} and so the continuity of the vector subtraction mapS:X×XX,{\displaystyle S:X\times X\to X,} which is defined byS(x,y) =def xy,{\displaystyle S(x,y)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~x-y,} guarantees thatS(x×x)S(x,x){\displaystyle S\left(x_{\bullet }\times x_{\bullet }\right)\to S(x,x)} inX,{\displaystyle X,} whereS(x×x)=(xixj)(i,j)I×I=xx{\displaystyle S\left(x_{\bullet }\times x_{\bullet }\right)=\left(x_{i}-x_{j}\right)_{(i,j)\in I\times I}=x_{\bullet }-x_{\bullet }} andS(x,x)=xx=0.{\displaystyle S(x,x)=x-x=0.}This proves that every convergent net is a Cauchy net. By definition, a space is calledcomplete if the converse is also always true. That is,X{\displaystyle X} is complete if and only if the following holds:

wheneverx{\displaystyle x_{\bullet }} is a net inX,{\displaystyle X,} thenx{\displaystyle x_{\bullet }} converges (to some point) inX{\displaystyle X} if and only ifxx0{\displaystyle x_{\bullet }-x_{\bullet }\to 0} inX.{\displaystyle X.}

A similar characterization of completeness holds if filters and prefilters are used instead of nets.

A seriesi=1xi{\displaystyle \sum _{i=1}^{\infty }x_{i}} is called aCauchy series (respectively, aconvergent series) if the sequence ofpartial sums(i=1nxi)n=1{\displaystyle \left(\sum _{i=1}^{n}x_{i}\right)_{n=1}^{\infty }} is aCauchy sequence (respectively, aconvergent sequence).[5] Every convergent series is necessarily a Cauchy series. In a complete TVS, every Cauchy series is necessarily a convergent series.

Cauchy filter and Cauchy prefilter

[edit]

See also:Filters in topology

AprefilterB{\displaystyle {\mathcal {B}}} on atopological vector spaceX{\displaystyle X} is called aCauchy prefilter[6] if it satisfies any of the following equivalent conditions:

  1. BB0{\displaystyle {\mathcal {B}}-{\mathcal {B}}\to 0} inX.{\displaystyle X.}
  2. {BB:BB}0{\displaystyle \{B-B:B\in {\mathcal {B}}\}\to 0} inX.{\displaystyle X.}
  3. For every neighborhoodN{\displaystyle N} of the origin inX,{\displaystyle X,}B{\displaystyle {\mathcal {B}}} contains someN{\displaystyle N}-small set (that is, there exists someBB{\displaystyle B\in {\mathcal {B}}} such thatBBN{\displaystyle B-B\subseteq N}).[6]
  4. For every neighborhoodN{\displaystyle N} of the origin inX,{\displaystyle X,} there exists somexX{\displaystyle x\in X} and someBB{\displaystyle B\in {\mathcal {B}}} such thatBx+N.{\displaystyle B\subseteq x+N.}[6]
  5. Every neighborhood of the origin inX{\displaystyle X} contains some subset of the formx+B{\displaystyle x+B} wherexX{\displaystyle x\in X} andBB.{\displaystyle B\in {\mathcal {B}}.}

It suffices to check any of the above conditions for any givenneighborhood basis of0{\displaystyle 0} inX.{\displaystyle X.} ACauchy filter is a Cauchy prefilter that is also afilter onX.{\displaystyle X.}

IfB{\displaystyle {\mathcal {B}}} is a prefilter on a topological vector spaceX{\displaystyle X} and ifxX,{\displaystyle x\in X,} thenBx{\displaystyle {\mathcal {B}}\to x} inX{\displaystyle X} if and only ifxclB{\displaystyle x\in \operatorname {cl} {\mathcal {B}}} andB{\displaystyle {\mathcal {B}}} is Cauchy.[3]

Complete subset

[edit]
Main article:Complete uniform space

For anySX,{\displaystyle S\subseteq X,} a prefilterC{\displaystyle {\mathcal {C}}}onS{\displaystyle S} is necessarily a subset of(S){\displaystyle \wp (S)}; that is,C(S).{\displaystyle {\mathcal {C}}\subseteq \wp (S).}

A subsetS{\displaystyle S} of a TVS(X,τ){\displaystyle (X,\tau )} is called acomplete subset if it satisfies any of the following equivalent conditions:

  1. Every Cauchy prefilterC(S){\displaystyle {\mathcal {C}}\subseteq \wp (S)} onS{\displaystyle S}converges to at least one point ofS.{\displaystyle S.}
  2. Every Cauchy net inS{\displaystyle S}converges to at least one point ofS.{\displaystyle S.}
  3. S{\displaystyle S} is acomplete uniform space (under the point-set topology definition of "complete uniform space") whenS{\displaystyle S} is endowed with the uniformity induced on it by the canonical uniformity ofX.{\displaystyle X.}

The subsetS{\displaystyle S} is called asequentially complete subset if every Cauchy sequence inS{\displaystyle S} (or equivalently, every elementary Cauchy filter/prefilter onS{\displaystyle S}) converges to at least one point ofS.{\displaystyle S.}

Importantly,convergence to points outside ofS{\displaystyle S} does not prevent a set from being complete: IfX{\displaystyle X} is not Hausdorff and if every Cauchy prefilter onS{\displaystyle S} converges to some point ofS,{\displaystyle S,} thenS{\displaystyle S} will be complete even if some or all Cauchy prefilters onS{\displaystyle S}also converge to points(s) inXS.{\displaystyle X\setminus S.} In short, there is no requirement that these Cauchy prefilters onS{\displaystyle S} convergeonly to points inS.{\displaystyle S.} The same can be said of the convergence of Cauchy nets inS.{\displaystyle S.}

As a consequence, if a TVSX{\displaystyle X} isnot Hausdorff then every subset of the closure of{0}{\displaystyle \{0\}} inX{\displaystyle X} is complete because it is compact and every compact set is necessarily complete. In particular, ifSclX{0}{\displaystyle \varnothing \neq S\subseteq \operatorname {cl} _{X}\{0\}} is a proper subset, such asS={0}{\displaystyle S=\{0\}} for example, thenS{\displaystyle S} would be complete even thoughevery Cauchy net inS{\displaystyle S} (and also every Cauchy prefilter onS{\displaystyle S}) converges toevery point inclX{0},{\displaystyle \operatorname {cl} _{X}\{0\},} including those points inclX{0}{\displaystyle \operatorname {cl} _{X}\{0\}} that do not belong toS.{\displaystyle S.} This example also shows that complete subsets (and indeed, even compact subsets) of a non-Hausdorff TVS may fail to be closed. For example, ifSclX{0}{\displaystyle \varnothing \neq S\subseteq \operatorname {cl} _{X}\{0\}} thenS=clX{0}{\displaystyle S=\operatorname {cl} _{X}\{0\}} if and only ifS{\displaystyle S} is closed inX.{\displaystyle X.}

Complete topological vector space

[edit]

Atopological vector spaceX{\displaystyle X} is called acomplete topological vector space if any of the following equivalent conditions are satisfied:

  1. X{\displaystyle X} is acomplete uniform space when it is endowed with its canonical uniformity.
  2. X{\displaystyle X} is a complete subset of itself.
  3. There exists a neighborhood of the origin inX{\displaystyle X} that is also a complete subset ofX.{\displaystyle X.}[6]
    • This implies that everylocally compact TVS is complete (even if the TVS is not Hausdorff).
  4. Every Cauchy prefilterC(X){\displaystyle {\mathcal {C}}\subseteq \wp (X)} onX{\displaystyle X}converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}
  5. Every Cauchyfilter onX{\displaystyle X} converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}
  6. Every Cauchy net inX{\displaystyle X}converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}

where if in additionX{\displaystyle X} ispseudometrizable or metrizable (for example, anormed space) then this list can be extended to include:

  1. X{\displaystyle X} is sequentially complete.

A topological vector spaceX{\displaystyle X} issequentially complete if any of the following equivalent conditions are satisfied:

  1. X{\displaystyle X} is a sequentially complete subset of itself.
  2. Every Cauchy sequence inX{\displaystyle X} converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}
  3. Every elementary Cauchy prefilter onX{\displaystyle X} converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}
  4. Every elementary Cauchy filter onX{\displaystyle X} converges inX{\displaystyle X} to at least one point ofX.{\displaystyle X.}

Uniqueness of the canonical uniformity

[edit]
See also:Topological group

The existence of the canonical uniformity was demonstrated above by defining it. The theorem below establishes that the canonical uniformity of any TVS(X,τ){\displaystyle (X,\tau )} is the only uniformity onX{\displaystyle X} that is both (1) translation invariant, and (2) generates onX{\displaystyle X} the topologyτ.{\displaystyle \tau .}

Theorem[7] (Existence and uniqueness of the canonical uniformity)The topology of any TVS can be derived from a unique translation-invariant uniformity. IfN(0){\displaystyle {\mathcal {N}}(0)} is anyneighborhood base of the origin, then the family{Δ(N):NN(0)}{\displaystyle \left\{\Delta (N):N\in {\mathcal {N}}(0)\right\}} is a base for this uniformity.

This section is dedicated to explaining the precise meanings of the terms involved in this uniqueness statement.

Uniform spaces and translation-invariant uniformities

[edit]
Main article:Uniform space

For any subsetsΦ,ΨX×X,{\displaystyle \Phi ,\Psi \subseteq X\times X,} let[1]Φop =def {(y,x) : (x,y)Φ}{\displaystyle \Phi ^{\operatorname {op} }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{(y,x)~:~(x,y)\in \Phi \}}and letΦΨ  =def {(x,z): there exists yX such that (x,y)Ψ and (y,z)Φ}= yX{(x,z) : (x,y)Ψ and (y,z)Φ}{\displaystyle {\begin{alignedat}{4}\Phi \circ \Psi ~&~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{(x,z):{\text{ there exists }}y\in X{\text{ such that }}(x,y)\in \Psi {\text{ and }}(y,z)\in \Phi \right\}\\&=~\bigcup _{y\in X}\{(x,z)~:~(x,y)\in \Psi {\text{ and }}(y,z)\in \Phi \}\end{alignedat}}}A non-empty familyB(X×X){\displaystyle {\mathcal {B}}\subseteq \wp (X\times X)} is called abase of entourages or afundamental system of entourages ifB{\displaystyle {\mathcal {B}}} is aprefilter onX×X{\displaystyle X\times X} satisfying all of the following conditions:

  1. Every set inB{\displaystyle {\mathcal {B}}} contains the diagonal ofX{\displaystyle X} as a subset; that is,ΔX =def {(x,x):xX}Φ{\displaystyle \Delta _{X}~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{(x,x):x\in X\}\subseteq \Phi } for everyΦB.{\displaystyle \Phi \in {\mathcal {B}}.} Said differently, the prefilterB{\displaystyle {\mathcal {B}}} isfixed onΔX.{\displaystyle \Delta _{X}.}
  2. For everyΩB{\displaystyle \Omega \in {\mathcal {B}}} there exists someΦB{\displaystyle \Phi \in {\mathcal {B}}} such thatΦΦΩ.{\displaystyle \Phi \circ \Phi \subseteq \Omega .}
  3. For everyΩB{\displaystyle \Omega \in {\mathcal {B}}} there exists someΦB{\displaystyle \Phi \in {\mathcal {B}}} such thatΦΩop =def {(y,x):(x,y)Ω}.{\displaystyle \Phi \subseteq \Omega ^{\operatorname {op} }~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{(y,x):(x,y)\in \Omega \}.}

Auniformity oruniform structure onX{\displaystyle X} is afilterU{\displaystyle {\mathcal {U}}} onX×X{\displaystyle X\times X} that is generated by some base of entouragesB,{\displaystyle {\mathcal {B}},} in which case we say thatB{\displaystyle {\mathcal {B}}} is abase of entouragesforU.{\displaystyle {\mathcal {U}}.}

For a commutative additive groupX,{\displaystyle X,} atranslation-invariant fundamental system of entourages[7] is a fundamental system of entouragesB{\displaystyle {\mathcal {B}}} such that for everyΦB,{\displaystyle \Phi \in {\mathcal {B}},}(x,y)Φ{\displaystyle (x,y)\in \Phi } if and only if(x+z,y+z)Φ{\displaystyle (x+z,y+z)\in \Phi } for allx,y,zX.{\displaystyle x,y,z\in X.} A uniformityB{\displaystyle {\mathcal {B}}} is called atranslation-invariant uniformity[7] if it has a base of entourages that is translation-invariant. The canonical uniformity on any TVS is translation-invariant.[7]

The binary operator{\displaystyle \;\circ \;} satisfies all of the following:

Symmetric entourages

Call a subsetΦX×X{\displaystyle \Phi \subseteq X\times X}symmetric ifΦ=Φop,{\displaystyle \Phi =\Phi ^{\operatorname {op} },} which is equivalent toΦopΦ.{\displaystyle \Phi ^{\operatorname {op} }\subseteq \Phi .} This equivalence follows from the identity(Φop)op=Φ{\displaystyle \left(\Phi ^{\operatorname {op} }\right)^{\operatorname {op} }=\Phi } and the fact that ifΨX×X,{\displaystyle \Psi \subseteq X\times X,} thenΦΨ{\displaystyle \Phi \subseteq \Psi } if and only ifΦopΨop.{\displaystyle \Phi ^{\operatorname {op} }\subseteq \Psi ^{\operatorname {op} }.} For example, the setΦopΦ{\displaystyle \Phi ^{\operatorname {op} }\cap \Phi } is always symmetric for everyΦX×X.{\displaystyle \Phi \subseteq X\times X.} And because(ΦΨ)op=ΦopΨop,{\displaystyle (\Phi \cap \Psi )^{\operatorname {op} }=\Phi ^{\operatorname {op} }\cap \Psi ^{\operatorname {op} },} ifΦ{\displaystyle \Phi } andΨ{\displaystyle \Psi } are symmetric then so isΦΨ.{\displaystyle \Phi \cap \Psi .}

Topology generated by a uniformity

[edit]
See also:Topological space § Neighborhood definition, andCharacterizations of the category of topological spaces

Relatives

LetΦX×X{\displaystyle \Phi \subseteq X\times X} be arbitrary and letPr1,Pr2:X×XX{\displaystyle \operatorname {Pr} _{1},\operatorname {Pr} _{2}:X\times X\to X} be the canonical projections onto the first and second coordinates, respectively.

For anySX,{\displaystyle S\subseteq X,} defineSΦ =def {yX:Φ(S×{x})} = Pr2(Φ(S×X)){\displaystyle S\cdot \Phi ~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{y\in X:\Phi \cap (S\times \{x\})\neq \varnothing \}~=~\operatorname {Pr} _{2}(\Phi \cap (S\times X))}ΦS =def {xX:Φ({x}×S)} = Pr1(Φ(X×S))=S(Φop){\displaystyle \Phi \cdot S~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{x\in X:\Phi \cap (\{x\}\times S)\neq \varnothing \}~=~\operatorname {Pr} _{1}(\Phi \cap (X\times S))=S\cdot \left(\Phi ^{\operatorname {op} }\right)}whereΦS{\displaystyle \Phi \cdot S} (respectively,SΦ{\displaystyle S\cdot \Phi }) is called the set ofleft (respectively,right)Φ{\displaystyle \Phi }-relatives of (points in)S.{\displaystyle S.} Denote the special case whereS={p}{\displaystyle S=\{p\}} is a singleton set for somepX{\displaystyle p\in X} by:pΦ =def {p}Φ = {yX:(p,y)Φ}{\displaystyle p\cdot \Phi ~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{p\}\cdot \Phi ~=~\{y\in X:(p,y)\in \Phi \}}Φp =def Φ{p} = {xX:(x,p)Φ} = p(Φop){\displaystyle \Phi \cdot p~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\Phi \cdot \{p\}~=~\{x\in X:(x,p)\in \Phi \}~=~p\cdot \left(\Phi ^{\operatorname {op} }\right)}IfΦ,ΨX×X{\displaystyle \Phi ,\Psi \subseteq X\times X} then(ΦΨ)S=Φ(ΨS).{\textstyle (\Phi \circ \Psi )\cdot S=\Phi \cdot (\Psi \cdot S).} Moreover,{\displaystyle \,\cdot \,}right distributes over both unions and intersections, meaning that ifR,SX{\displaystyle R,S\subseteq X} then(RS)Φ = (RΦ)(SΦ){\displaystyle (R\cup S)\cdot \Phi ~=~(R\cdot \Phi )\cup (S\cdot \Phi )} and(RS)Φ  (RΦ)(SΦ).{\displaystyle (R\cap S)\cdot \Phi ~\subseteq ~(R\cdot \Phi )\cap (S\cdot \Phi ).}

Neighborhoods and open sets

Two pointsx{\displaystyle x} andy{\displaystyle y} areΦ{\displaystyle \Phi }-close if(x,y)Φ{\displaystyle (x,y)\in \Phi } and a subsetSX{\displaystyle S\subseteq X} is calledΦ{\displaystyle \Phi }-small ifS×SΦ.{\displaystyle S\times S\subseteq \Phi .}

LetB(X×X){\displaystyle {\mathcal {B}}\subseteq \wp (X\times X)} be a base of entourages onX.{\displaystyle X.} Theneighborhood prefilter at a pointpX{\displaystyle p\in X} and, respectively, on a subsetSX{\displaystyle S\subseteq X} are thefamilies of sets:Bp =def B{p}={Φp:ΦB} and BS =def {ΦS:ΦB}{\displaystyle {\mathcal {B}}\cdot p~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~{\mathcal {B}}\cdot \{p\}=\{\Phi \cdot p:\Phi \in {\mathcal {B}}\}\qquad {\text{ and }}\qquad {\mathcal {B}}\cdot S~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{\Phi \cdot S:\Phi \in {\mathcal {B}}\}}and the filters onX{\displaystyle X} that each generates is known as theneighborhood filter ofp{\displaystyle p} (respectively, ofS{\displaystyle S}). Assign to everyxX{\displaystyle x\in X} the neighborhood prefilterBx =def {Φx:ΦB}{\displaystyle {\mathcal {B}}\cdot x~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{\Phi \cdot x:\Phi \in {\mathcal {B}}\}}and use theneighborhood definition of "open set" to obtain atopology onX{\displaystyle X} called thetopology induced byB{\displaystyle {\mathcal {B}}} or theinduced topology. Explicitly, a subsetUX{\displaystyle U\subseteq X} is open in this topology if and only if for everyuU{\displaystyle u\in U} there exists someNBu{\displaystyle N\in {\mathcal {B}}\cdot u} such thatNU;{\displaystyle N\subseteq U;} that is,U{\displaystyle U} is open if and only if for everyuU{\displaystyle u\in U} there exists someΦB{\displaystyle \Phi \in {\mathcal {B}}} such thatΦu =def {xX:(x,u)Φ}U.{\displaystyle \Phi \cdot u~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{x\in X:(x,u)\in \Phi \}\subseteq U.}

The closure of a subsetSX{\displaystyle S\subseteq X} in this topology is:clXS=ΦB(ΦS)=ΦB(SΦ).{\displaystyle \operatorname {cl} _{X}S=\bigcap _{\Phi \in {\mathcal {B}}}(\Phi \cdot S)=\bigcap _{\Phi \in {\mathcal {B}}}(S\cdot \Phi ).}

Cauchy prefilters and complete uniformities

A prefilterF(X){\displaystyle {\mathcal {F}}\subseteq \wp (X)} on a uniform spaceX{\displaystyle X} with uniformityU{\displaystyle {\mathcal {U}}} is called aCauchy prefilter if for every entourageNU,{\displaystyle N\in {\mathcal {U}},} there exists someFF{\displaystyle F\in {\mathcal {F}}} such thatF×FN.{\displaystyle F\times F\subseteq N.}

A uniform space(X,U){\displaystyle (X,{\mathcal {U}})} is called acomplete uniform space (respectively, asequentially complete uniform space) if every Cauchy prefilter (respectively, every elementary Cauchy prefilter) onX{\displaystyle X} converges to at least one point ofX{\displaystyle X} whenX{\displaystyle X} is endowed with the topology induced byU.{\displaystyle {\mathcal {U}}.}

Case of a topological vector space

If(X,τ){\displaystyle (X,\tau )} is atopological vector space then for anySX{\displaystyle S\subseteq X} andxX,{\displaystyle x\in X,}ΔX(N)S=S+N and ΔX(N)x=x+N,{\displaystyle \Delta _{X}(N)\cdot S=S+N\qquad {\text{ and }}\qquad \Delta _{X}(N)\cdot x=x+N,}and the topology induced onX{\displaystyle X} by the canonical uniformity is the same as the topology thatX{\displaystyle X} started with (that is, it isτ{\displaystyle \tau }).

Uniform continuity

[edit]

LetX{\displaystyle X} andY{\displaystyle Y} be TVSs,DX,{\displaystyle D\subseteq X,} andf:DY{\displaystyle f:D\to Y} be a map. Thenf:DY{\displaystyle f:D\to Y} isuniformly continuous if for every neighborhoodU{\displaystyle U} of the origin inX,{\displaystyle X,} there exists a neighborhoodV{\displaystyle V} of the origin inY{\displaystyle Y} such that for allx,yD,{\displaystyle x,y\in D,} ifyxU{\displaystyle y-x\in U} thenf(y)f(x)V.{\displaystyle f(y)-f(x)\in V.}

Suppose thatf:DY{\displaystyle f:D\to Y} is uniformly continuous. Ifx=(xi)iI{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i\in I}} is a Cauchy net inD{\displaystyle D} thenfx=(f(xi))iI{\displaystyle f\circ x_{\bullet }=\left(f\left(x_{i}\right)\right)_{i\in I}} is a Cauchy net inY.{\displaystyle Y.} IfB{\displaystyle {\mathcal {B}}} is a Cauchy prefilter inD{\displaystyle D} (meaning thatB{\displaystyle {\mathcal {B}}} is a family of subsets ofD{\displaystyle D} that is Cauchy inX{\displaystyle X}) thenf(B){\displaystyle f\left({\mathcal {B}}\right)} is a Cauchy prefilter inY.{\displaystyle Y.} However, ifB{\displaystyle {\mathcal {B}}} is a Cauchy filter onD{\displaystyle D} then althoughf(B){\displaystyle f\left({\mathcal {B}}\right)} will be a Cauchyprefilter, it will be a Cauchy filter inY{\displaystyle Y} if and only iff:DY{\displaystyle f:D\to Y} is surjective.

TVS completeness vs completeness of (pseudo)metrics

[edit]

Preliminaries: Complete pseudometric spaces

[edit]
Main articles:Complete metric space,Pseudometric space, andCauchy sequence

We review the basic notions related to the general theory of complete pseudometric spaces. Recall that everymetric is apseudometric and that a pseudometricp{\displaystyle p} is a metric if and only ifp(x,y)=0{\displaystyle p(x,y)=0} impliesx=y.{\displaystyle x=y.} Thus everymetric space is apseudometric space and a pseudometric space(X,p){\displaystyle (X,p)} is a metric space if and only ifp{\displaystyle p} is a metric.

IfS{\displaystyle S} is a subset of apseudometric space(X,d){\displaystyle (X,d)} then thediameter ofS{\displaystyle S} is defined to bediam(S) =def sup{d(s,t):s,tS}.{\displaystyle \operatorname {diam} (S)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\sup _{}\{d(s,t):s,t\in S\}.}

A prefilterB{\displaystyle {\mathcal {B}}} on a pseudometric space(X,d){\displaystyle (X,d)} is called ad{\displaystyle d}-Cauchy prefilter or simply aCauchy prefilter if for eachrealr>0,{\displaystyle r>0,} there is someBB{\displaystyle B\in {\mathcal {B}}} such that the diameter ofB{\displaystyle B} is less thanr.{\displaystyle r.}

Suppose(X,d){\displaystyle (X,d)} is a pseudometric space. Anetx=(xi)iI{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i\in I}} inX{\displaystyle X} is called ad{\displaystyle d}-Cauchy net or simply aCauchy net ifTails(x){\displaystyle \operatorname {Tails} \left(x_{\bullet }\right)} is a Cauchy prefilter, which happens if and only if

for everyr>0{\displaystyle r>0} there is someiI{\displaystyle i\in I} such that ifj,kI{\displaystyle j,k\in I} withji{\displaystyle j\geq i} andki{\displaystyle k\geq i} thend(xj,xk)<r{\displaystyle d\left(x_{j},x_{k}\right)<r}

or equivalently, if and only if(d(xj,xk))(i,j)I×I0{\displaystyle \left(d\left(x_{j},x_{k}\right)\right)_{(i,j)\in I\times I}\to 0} inR.{\displaystyle \mathbb {R} .} This is analogous to the following characterization of the converge ofx{\displaystyle x_{\bullet }} to a point: ifxX,{\displaystyle x\in X,} thenxx{\displaystyle x_{\bullet }\to x} in(X,d){\displaystyle (X,d)} if and only if(xi,x)iI0{\displaystyle \left(x_{i},x\right)_{i\in I}\to 0} inR.{\displaystyle \mathbb {R} .}

ACauchy sequence is a sequence that is also a Cauchy net.[note 3]

Every pseudometricp{\displaystyle p} on a setX{\displaystyle X} induces the usual canonical topology onX,{\displaystyle X,} which we'll denote byτp{\displaystyle \tau _{p}}; it also induces a canonicaluniformity onX,{\displaystyle X,} which we'll denote byUp.{\displaystyle {\mathcal {U}}_{p}.} The topology onX{\displaystyle X} induced by the uniformityUp{\displaystyle {\mathcal {U}}_{p}} is equal toτp.{\displaystyle \tau _{p}.} A netx=(xi)iI{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i\in I}} inX{\displaystyle X} is Cauchy with respect top{\displaystyle p} if and only if it is Cauchy with respect to the uniformityUp.{\displaystyle {\mathcal {U}}_{p}.} The pseudometric space(X,p){\displaystyle (X,p)} is acomplete (resp. a sequentially complete) pseudometric space if and only if(X,Up){\displaystyle \left(X,{\mathcal {U}}_{p}\right)} is acomplete (resp. a sequentially complete) uniform space. Moreover, the pseudometric space(X,p){\displaystyle (X,p)} (resp. the uniform space(X,Up){\displaystyle \left(X,{\mathcal {U}}_{p}\right)}) is complete if and only if it is sequentially complete.

A pseudometric space(X,d){\displaystyle (X,d)} (for example, ametric space) is calledcomplete andd{\displaystyle d} is called acomplete pseudometric if any of the following equivalent conditions hold:

  1. Every Cauchy prefilter onX{\displaystyle X} converges to at least one point ofX.{\displaystyle X.}
  2. The above statement but with the word "prefilter" replaced by "filter."
  3. Every Cauchy net inX{\displaystyle X} converges to at least one point ofX.{\displaystyle X.}
  4. Every Cauchy sequence inX{\displaystyle X} converges to at least one point ofX.{\displaystyle X.}
    • Thus to prove that(X,d){\displaystyle (X,d)} is complete, it suffices to only consider Cauchy sequences inX{\displaystyle X} (and it is not necessary to consider the more general Cauchy nets).
  5. The canonical uniformity onX{\displaystyle X} induced by the pseudometricd{\displaystyle d} is a complete uniformity.

And if additiond{\displaystyle d} is a metric then we may add to this list:

  1. Every decreasing sequence of closed balls whose diameters shrink to0{\displaystyle 0} has non-empty intersection.[8]

Complete pseudometrics and complete TVSs

[edit]

EveryF-space, and thus also everyFréchet space,Banach space, andHilbert space is a complete TVS. Note that everyF-space is aBaire space but there are normed spaces that are Baire but not Banach.[9]

A pseudometricd{\displaystyle d} on a vector spaceX{\displaystyle X} is said to be atranslation invariant pseudometric ifd(x,y)=d(x+z,y+z){\displaystyle d(x,y)=d(x+z,y+z)} for all vectorsx,y,zX.{\displaystyle x,y,z\in X.}

Suppose(X,τ){\displaystyle (X,\tau )} ispseudometrizable TVS (for example, a metrizable TVS) and thatp{\displaystyle p} isany pseudometric onX{\displaystyle X} such that the topology onX{\displaystyle X} induced byp{\displaystyle p} is equal toτ.{\displaystyle \tau .} Ifp{\displaystyle p} is translation-invariant, then(X,τ){\displaystyle (X,\tau )} is a complete TVS if and only if(X,p){\displaystyle (X,p)} is a complete pseudometric space.[10] Ifp{\displaystyle p} isnot translation-invariant, then may be possible for(X,τ){\displaystyle (X,\tau )} to be a complete TVS but(X,p){\displaystyle (X,p)} tonot be a complete pseudometric space[10] (see this footnote[note 4] for an example).[10]

Theorem[11][12] (Klee)Letd{\displaystyle d} beany[note 5] metric on a vector spaceX{\displaystyle X} such that the topologyτ{\displaystyle \tau } induced byd{\displaystyle d} onX{\displaystyle X} makes(X,τ){\displaystyle (X,\tau )} into a topological vector space. If(X,d){\displaystyle (X,d)} is a complete metric space then(X,τ){\displaystyle (X,\tau )} is a complete-TVS.

Complete norms and equivalent norms

[edit]

Two norms on a vector space are calledequivalent if and only if they induce the same topology.[13] Ifp{\displaystyle p} andq{\displaystyle q} are two equivalent norms on a vector spaceX{\displaystyle X} then thenormed space(X,p){\displaystyle (X,p)} is aBanach space if and only if(X,q){\displaystyle (X,q)} is a Banach space. See this footnote for an example of a continuous norm on a Banach space that isnot equivalent to that Banach space's given norm.[note 6][13] All norms on a finite-dimensional vector space are equivalent and every finite-dimensional normed space is a Banach space.[14] Every Banach space is a complete TVS. A normed space is a Banach space (that is, its canonical norm-induced metric is complete) if and only if it is complete as a topological vector space.

Completions

[edit]
See also:Completion (uniform space) andCompletion (metric space)

Acompletion[15] of a TVSX{\displaystyle X} is a complete TVS that contains a dense vector subspace that is TVS-isomorphic toX.{\displaystyle X.} In other words, it is a complete TVSC{\displaystyle C} into whichX{\displaystyle X} can beTVS-embedded as adensevector subspace. Every TVS-embedding is auniform embedding.

Every topological vector space has a completion. Moreover, every Hausdorff TVS has aHausdorff completion, which is necessarily uniqueup toTVS-isomorphism. However, all TVSs, even those that are Hausdorff, (already) complete, and/or metrizable have infinitely many non-Hausdorff completions that arenot TVS-isomorphic to one another.

Examples of completions

[edit]

For example, the vector space consisting of scalar-valuedsimple functionsf{\displaystyle f} for which|f|p<{\displaystyle |f|_{p}<\infty } (where this seminorm is defined in the usual way in terms ofLebesgue integration) becomes aseminormed space when endowed with this seminorm, which in turn makes it into both apseudometric space and a non-Hausdorff non-complete TVS; any completion of this space is a non-Hausdorff complete seminormed space that whenquotiented by the closure of its origin (so as toobtain a Hausdorff TVS) results in (a spacelinearlyisometrically-isomorphic to) the usual complete HausdorffLp{\displaystyle L^{p}}-space (endowed with the usual completep{\displaystyle \|\cdot \|_{p}}norm).

As another example demonstrating the usefulness of completions, the completions oftopological tensor products, such asprojective tensor products orinjective tensor products, of the Banach space1(S){\displaystyle \ell ^{1}(S)} with a complete Hausdorff locally convex TVSY{\displaystyle Y} results in a complete TVS that is TVS-isomorphic to a "generalized"1(S;Y){\displaystyle \ell ^{1}(S;Y)}-space consistingY{\displaystyle Y}-valued functions onS{\displaystyle S} (where this "generalized" TVS is defined analogously to original space1(S){\displaystyle \ell ^{1}(S)} of scalar-valued functions onS{\displaystyle S}). Similarly, the completion of the injective tensor product of thespace of scalar-valuedCk{\displaystyle C^{k}}-test functions with such a TVSY{\displaystyle Y} is TVS-isomorphic to the analogously defined TVS ofY{\displaystyle Y}-valuedCk{\displaystyle C^{k}} test functions.

Non-uniqueness of all completions

[edit]

As the example below shows, regardless of whether or not a space is Hausdorff or already complete, everytopological vector space (TVS) has infinitely many non-isomorphic completions.[16]

However, every Hausdorff TVS has aHausdorff completion that is unique up to TVS-isomorphism.[16] But nevertheless, every Hausdorff TVS still has infinitely many non-isomorphic non-Hausdorff completions.

Example (Non-uniqueness of completions):[15] LetC{\displaystyle C} denote any complete TVS and letI{\displaystyle I} denote any TVS endowed with theindiscrete topology, which recall makesI{\displaystyle I} into a complete TVS. Since bothI{\displaystyle I} andC{\displaystyle C} are complete TVSs, so is their productI×C.{\displaystyle I\times C.} IfU{\displaystyle U} andV{\displaystyle V} are non-empty open subsets ofI{\displaystyle I} andC,{\displaystyle C,} respectively, thenU=I{\displaystyle U=I} and(U×V)({0}×C)={0}×V,{\displaystyle (U\times V)\cap (\{0\}\times C)=\{0\}\times V\neq \varnothing ,} which shows that{0}×C{\displaystyle \{0\}\times C} is a dense subspace ofI×C.{\displaystyle I\times C.} Thus by definition of "completion,"I×C{\displaystyle I\times C} is a completion of{0}×C{\displaystyle \{0\}\times C} (it doesn't matter that{0}×C{\displaystyle \{0\}\times C} is already complete). So by identifying{0}×C{\displaystyle \{0\}\times C} withC,{\displaystyle C,} ifXC{\displaystyle X\subseteq C} is a dense vector subspace ofC,{\displaystyle C,} thenX{\displaystyle X} has bothC{\displaystyle C} andI×C{\displaystyle I\times C} as completions.

Hausdorff completions

[edit]

Every Hausdorff TVS has aHausdorff completion that is unique up to TVS-isomorphism.[16] But nevertheless, as shown above, every Hausdorff TVS still has infinitely many non-isomorphic non-Hausdorff completions.

Properties of Hausdorff completions[17]Suppose thatX{\displaystyle X} andC{\displaystyle C} are Hausdorff TVSs withC{\displaystyle C} complete. Suppose thatE:XC{\displaystyle E:X\to C} is a TVS-embedding onto a dense vector subspace ofC.{\displaystyle C.} Then

Universal property: for every continuous linear mapf:XZ{\displaystyle f:X\to Z} into a complete Hausdorff TVSZ,{\displaystyle Z,} there exists a unique continuous linear mapF:CZ{\displaystyle F:C\to Z} such thatf=FE.{\displaystyle f=F\circ E.}

IfE2:XC2{\displaystyle E_{2}:X\to C_{2}} is a TVS embedding onto a dense vector subspace of a complete Hausdorff TVSC2{\displaystyle C_{2}} having the above universal property, then there exists a unique (bijective) TVS-isomorphismI:CC2{\displaystyle I:C\to C_{2}} such thatE2=IE.{\displaystyle E_{2}=I\circ E.}

Corollary[17]SupposeC{\displaystyle C} is a complete Hausdorff TVS andX{\displaystyle X} is a dense vector subspace ofC.{\displaystyle C.} Then every continuous linear mapf:XZ{\displaystyle f:X\to Z} into a complete Hausdorff TVSZ{\displaystyle Z} has a unique continuous linear extension to a mapCZ.{\displaystyle C\to Z.}

Existence of Hausdorff completions

See also:Filters in topology

A Cauchy filterB{\displaystyle {\mathcal {B}}} on a TVSX{\displaystyle X} is called aminimal Cauchy filter[17] if there doesnot exist a Cauchy filter onX{\displaystyle X} that is strictly coarser thanB{\displaystyle {\mathcal {B}}} (that is, "strictly coarser thanB{\displaystyle {\mathcal {B}}}" means contained as a proper subset ofB{\displaystyle {\mathcal {B}}}).

IfB{\displaystyle {\mathcal {B}}} is a Cauchy filter onX{\displaystyle X} then the filter generated by the following prefilter:{B+N : BB and N is a neighborhood of 0 in X}{\displaystyle \left\{B+N~:~B\in {\mathcal {B}}{\text{ and }}N{\text{ is a neighborhood of }}0{\text{ in }}X\right\}}is the unique minimal Cauchy filter onX{\displaystyle X} that is contained as a subset ofB.{\displaystyle {\mathcal {B}}.}[17] In particular, for anyxX,{\displaystyle x\in X,} the neighborhood filter atx{\displaystyle x} is a minimal Cauchy filter.

LetM{\displaystyle \mathbb {M} } be the set of all minimal Cauchy filters onX{\displaystyle X} and letE:XM{\displaystyle E:X\rightarrow \mathbb {M} } be the map defined by sendingxX{\displaystyle x\in X} to the neighborhood filter ofx{\displaystyle x} inX.{\displaystyle X.} EndowM{\displaystyle \mathbb {M} } with the following vector space structure: GivenB,CM{\displaystyle {\mathcal {B}},{\mathcal {C}}\in \mathbb {M} } and a scalars,{\displaystyle s,} letB+C{\displaystyle {\mathcal {B}}+{\mathcal {C}}} (resp.sB{\displaystyle s{\mathcal {B}}}) denote the unique minimal Cauchy filter contained in the filter generated by{B+C:BB,CC}{\displaystyle \left\{B+C:B\in {\mathcal {B}},C\in {\mathcal {C}}\right\}} (resp.{sB:BB}{\displaystyle \{sB:B\in {\mathcal {B}}\}}).

For everybalanced neighborhoodN{\displaystyle N} of the origin inX,{\displaystyle X,} letU(N) =def {BM :  there exist BB and a neighborhood V of the origin in X such that B+VN}{\displaystyle \mathbb {U} (N)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{{\mathcal {B}}\in \mathbb {M} ~:~{\text{ there exist }}B\in {\mathcal {B}}{\text{ and a neighborhood }}V{\text{ of the origin in }}X{\text{ such that }}B+V\subseteq N\right\}}

IfX{\displaystyle X} is Hausdorff then the collection of all setsU(N),{\displaystyle \mathbb {U} (N),} asN{\displaystyle N} ranges over all balanced neighborhoods of the origin inX,{\displaystyle X,} forms a vector topology onM{\displaystyle \mathbb {M} } makingM{\displaystyle \mathbb {M} } into a complete Hausdorff TVS. Moreover, the mapE:XM{\displaystyle E:X\rightarrow \mathbb {M} } is a TVS-embedding onto a dense vector subspace ofM.{\displaystyle \mathbb {M} .}[17]

IfX{\displaystyle X} is ametrizable TVS then a Hausdorff completion ofX{\displaystyle X} can be constructed using equivalence classes of Cauchy sequences instead of minimal Cauchy filters.

Non-Hausdorff completions

[edit]

This subsection details how every non-Hausdorff TVSX{\displaystyle X} can be TVS-embedded onto a dense vector subspace of a complete TVS. The proof that every Hausdorff TVS has a Hausdorff completion is widely available and so this fact will be used (without proof) to show that every non-Hausdorff TVS also has a completion. These details are sometimes useful for extending results from Hausdorff TVSs to non-Hausdorff TVSs.

LetI=cl{0}{\displaystyle I=\operatorname {cl} \{0\}} denote the closure of the origin inX,{\displaystyle X,} whereI{\displaystyle I} is endowed with its subspace topology induced byX{\displaystyle X} (so thatI{\displaystyle I} has theindiscrete topology). SinceI{\displaystyle I} has the trivial topology, it is easily shown that every vector subspace ofX{\displaystyle X} that is an algebraic complement ofI{\displaystyle I} inX{\displaystyle X} is necessarily atopological complement ofI{\displaystyle I} inX.{\displaystyle X.}[18][19] LetH{\displaystyle H} denote any topological complement ofI{\displaystyle I} inX,{\displaystyle X,} which is necessarily a Hausdorff TVS (since it is TVS-isomorphic to the quotient TVSX/I{\displaystyle X/I}[note 7]). SinceX{\displaystyle X} is thetopological direct sum ofI{\displaystyle I} andH{\displaystyle H} (which means thatX=IH{\displaystyle X=I\oplus H} in the category of TVSs), the canonical mapI×HIH=X given by (x,y)x+y{\displaystyle I\times H\to I\oplus H=X\quad {\text{ given by }}\quad (x,y)\mapsto x+y}is a TVS-isomorphism.[19] LetA : X=IH  I×H{\displaystyle A~:~X=I\oplus H~\to ~I\times H} denote the inverse of this canonical map. (As a side note, it follows that every open and every closed subsetU{\displaystyle U} ofX{\displaystyle X} satisfiesU=I+U.{\displaystyle U=I+U.}[proof 1])

The Hausdorff TVSH{\displaystyle H} can be TVS-embedded, say via the mapInH:HC,{\displaystyle \operatorname {In} _{H}:H\to C,} onto a dense vector subspace of its completionC.{\displaystyle C.} SinceI{\displaystyle I} andC{\displaystyle C} are complete, so is their productI×C.{\displaystyle I\times C.} LetIdI:II{\displaystyle \operatorname {Id} _{I}:I\to I} denote the identity map and observe that the product mapIdI×InH:I×HI×C{\displaystyle \operatorname {Id} _{I}\times \operatorname {In} _{H}:I\times H\to I\times C} is a TVS-embedding whose image is dense inI×C.{\displaystyle I\times C.} Define the map[note 8]B:X=IHI×C by B =def (IdI×InH)A{\displaystyle B:X=I\oplus H\to I\times C\quad {\text{ by }}\quad B~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(\operatorname {Id} _{I}\times \operatorname {In} _{H}\right)\circ A}which is a TVS-embedding ofX=IH{\displaystyle X=I\oplus H} onto a dense vector subspace of the complete TVSI×C.{\displaystyle I\times C.} Moreover, observe that the closure of the origin inI×C{\displaystyle I\times C} is equal toI×{0},{\displaystyle I\times \{0\},} and thatI×{0}{\displaystyle I\times \{0\}} and{0}×C{\displaystyle \{0\}\times C} are topological complements inI×C.{\displaystyle I\times C.}

To summarize,[19] given any algebraic (and thus topological) complementH{\displaystyle H} ofI =def cl{0}{\displaystyle I~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\operatorname {cl} \{0\}} inX{\displaystyle X} and given any completionC{\displaystyle C} of the Hausdorff TVSH{\displaystyle H} such thatHC,{\displaystyle H\subseteq C,} then the natural inclusion[20]InH:X=IHIC{\displaystyle \operatorname {In} _{H}:X=I\oplus H\to I\oplus C}is a well-defined TVS-embedding ofX{\displaystyle X} onto a dense vector subspace of the complete TVSIC{\displaystyle I\oplus C} where moreover,X=IHICI×C.{\displaystyle X=I\oplus H\subseteq I\oplus C\cong I\times C.}

Topology of a completion

[edit]

Theorem[7][21] (Topology of a completion)LetC{\displaystyle C} be a complete TVS and letX{\displaystyle X} be a dense vector subspace ofX.{\displaystyle X.} IfNX(0){\displaystyle {\mathcal {N}}_{X}(0)} is anyneighborhood base of the origin inX{\displaystyle X} then the setNX(0) =def {clCN : NNX(0)}{\displaystyle {\mathcal {N}}_{X}(0)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{\operatorname {cl} _{C}N~:~N\in {\mathcal {N}}_{X}(0)\right\}}is a neighborhood of the origin in the completionC{\displaystyle C} ofX.{\displaystyle X.}

IfX{\displaystyle X} is locally convex andP{\displaystyle {\mathcal {P}}} is a family of continuous seminorms onX{\displaystyle X} that generate the topology ofX,{\displaystyle X,} then the family of all continuous extensions toC{\displaystyle C} of all members ofP{\displaystyle {\mathcal {P}}} is a generating family of seminorms forC.{\displaystyle C.}

Said differently, ifC{\displaystyle C} is a completion of a TVSX{\displaystyle X} withXC{\displaystyle X\subseteq C} and ifN{\displaystyle {\mathcal {N}}} is aneighborhood base of the origin inX,{\displaystyle X,} then the family of sets{clCN : NN}{\displaystyle \left\{\operatorname {cl} _{C}N~:~N\in {\mathcal {N}}\right\}}is a neighborhood basis at the origin inC.{\displaystyle C.}[3]

Theorem[22] (Completions of quotients)LetM{\displaystyle M} be ametrizable topological vector space and letN{\displaystyle N} be a closed vector subspace ofM.{\displaystyle M.} Suppose thatC{\displaystyle C} is a completion ofM.{\displaystyle M.} Then the completion ofM/N{\displaystyle M/N} is TVS-isomorphic toC/clCN.{\displaystyle C/\operatorname {cl} _{C}N.} If in additionM{\displaystyle M} is a normed space, then this TVS-isomorphism is also an isometry.

Grothendieck's Completeness Theorem

See also:Filters in topology

LetE{\displaystyle {\mathcal {E}}} denote theequicontinuous compactology on the continuous dual spaceX,{\displaystyle X^{\prime },} which by definition consists of allequicontinuousweak-* closed and weak-*boundedabsolutely convex subsets ofX{\displaystyle X^{\prime }}[23] (which are necessarily weak-* compact subsets ofX{\displaystyle X^{\prime }}). Assume that everyEE{\displaystyle E^{\prime }\in {\mathcal {E}}} is endowed with theweak-* topology. AfilterB{\displaystyle {\mathcal {B}}} onX{\displaystyle X^{\prime }} is said toconverge continuously toxX{\displaystyle x^{\prime }\in X^{\prime }} if there exists someEEB{\displaystyle E^{\prime }\in {\mathcal {E}}\cap {\mathcal {B}}} containingx{\displaystyle x^{\prime }} (that is,xE{\displaystyle x^{\prime }\in E^{\prime }}) such that the trace ofB{\displaystyle {\mathcal {B}}} onE,{\displaystyle E^{\prime },} which is the familyB|E =def {BE:BB},{\displaystyle {\mathcal {B}}{\big \vert }_{E^{\prime }}~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left\{B\cap E^{\prime }:B\in {\mathcal {B}}\right\},}converges tox{\displaystyle x^{\prime }} inE{\displaystyle E^{\prime }} (that is, ifB|Ex{\displaystyle {\mathcal {B}}{\big \vert }_{E^{\prime }}\to x^{\prime }} in the given weak-* topology).[24] The filterB{\displaystyle {\mathcal {B}}} converges continuously tox{\displaystyle x^{\prime }} if and only ifBx{\displaystyle {\mathcal {B}}-x^{\prime }} converges continuously to the origin, which happens if and only if for everyxX,{\displaystyle x\in X,} the filterB,x+Nx,x{\displaystyle \langle {\mathcal {B}},x+{\mathcal {N}}\rangle \to \langle x^{\prime },x\rangle } in the scalar field (which isR{\displaystyle \mathbb {R} } orC{\displaystyle \mathbb {C} }) whereN{\displaystyle {\mathcal {N}}} denotes any neighborhood basis at the origin inX,{\displaystyle X,},{\displaystyle \langle \cdot ,\cdot \rangle } denotes theduality pairing, andB,x+N{\displaystyle \langle {\mathcal {B}},x+{\mathcal {N}}\rangle } denotes the filter generated by{B,x+N : BB,NN}.{\displaystyle \{\langle B,x+N\rangle ~:~B\in {\mathcal {B}},N\in {\mathcal {N}}\}.}[24] A mapf:XT{\displaystyle f:X^{\prime }\to T} into a topological space (such asR{\displaystyle \mathbb {R} } orC{\displaystyle \mathbb {C} }) is said to beγ{\displaystyle \gamma }-continuous if whenever a filterB{\displaystyle {\mathcal {B}}} onX{\displaystyle X^{\prime }}converges continuously toxX,{\displaystyle x^{\prime }\in X^{\prime },} thenf(B)f(x).{\displaystyle f({\mathcal {B}})\to f\left(x^{\prime }\right).}[24]

Grothendieck's Completeness Theorem[24]IfX{\displaystyle X} is a Hausdorff topological vector space then its completion is linearly isomorphic to the set of allγ{\displaystyle \gamma }-continuous linear functions onX.{\displaystyle X^{\prime }.}

Properties preserved by completions

[edit]

If a TVSX{\displaystyle X} has any of the following properties then so does its completion:

Completions of Hilbert spaces

Every inner product space(H,,){\displaystyle \left(H,\langle \cdot ,\cdot \rangle \right)} has a completion(H¯,,H¯){\displaystyle \left({\overline {H}},\langle \cdot ,\cdot \rangle _{\overline {H}}\right)} that is a Hilbert space, where the inner product,H¯{\displaystyle \langle \cdot ,\cdot \rangle _{\overline {H}}} is the unique continuous extension toH¯{\displaystyle {\overline {H}}} of the original inner product,.{\displaystyle \langle \cdot ,\cdot \rangle .} The norm induced by(H¯,,H¯){\displaystyle \left({\overline {H}},\langle \cdot ,\cdot \rangle _{\overline {H}}\right)} is also the unique continuous extension toH¯{\displaystyle {\overline {H}}} of the norm induced by,.{\displaystyle \langle \cdot ,\cdot \rangle .}[25][21]

Other preserved properties

IfX{\displaystyle X} is aHausdorff TVS, then the continuous dual space ofX{\displaystyle X} is identical to the continuous dual space of the completion ofX.{\displaystyle X.}[30] The completion of a locally convexbornological space is abarrelled space.[27] IfX{\displaystyle X} andY{\displaystyle Y} areDF-spaces then theprojective tensor product, as well as its completion, of these spaces is a DF-space.[31]

The completion of theprojective tensor product of two nuclear spaces is nuclear.[26] The completion of a nuclear space is TVS-isomorphic with a projective limit ofHilbert spaces.[26]

IfX=YZ{\displaystyle X=Y\oplus Z} (meaning that the addition mapY×ZX{\displaystyle Y\times Z\to X} is a TVS-isomorphism) has a Hausdorff completionC{\displaystyle C} then(clCY)+(clCZ)=C.{\displaystyle \left(\operatorname {cl} _{C}Y\right)+\left(\operatorname {cl} _{C}Z\right)=C.} If in additionX{\displaystyle X} is aninner product space andY{\displaystyle Y} andZ{\displaystyle Z} areorthogonal complements of each other inX{\displaystyle X} (that is,Y,Z={0}{\displaystyle \langle Y,Z\rangle =\{0\}}), thenclCY{\displaystyle \operatorname {cl} _{C}Y} andclCZ{\displaystyle \operatorname {cl} _{C}Z} are orthogonal complements in theHilbert spaceC.{\displaystyle C.}

Properties of maps preserved by extensions to a completion

[edit]

Iff:XY{\displaystyle f:X\to Y} is anuclear linear operator between two locally convex spaces and ifC{\displaystyle C} be a completion ofX{\displaystyle X} thenf{\displaystyle f} has a unique continuous linear extension to a nuclear linear operatorF:CY.{\displaystyle F:C\to Y.}[26]

LetX{\displaystyle X} andY{\displaystyle Y} be two Hausdorff TVSs withY{\displaystyle Y} complete. LetC{\displaystyle C} be a completion ofX.{\displaystyle X.} LetL(X;Y){\displaystyle L(X;Y)} denote the vector space of continuous linear operators and letI:L(X;Y)L(C;Y){\displaystyle I:L(X;Y)\to L(C;Y)} denote the map that sends everyfL(X;Y){\displaystyle f\in L(X;Y)} to its unique continuous linear extension onC.{\displaystyle C.} ThenI:L(X;Y)L(C;Y){\displaystyle I:L(X;Y)\to L(C;Y)} is a (surjective) vector space isomorphism. Moreover,I:L(X;Y)L(C;Y){\displaystyle I:L(X;Y)\to L(C;Y)} maps families ofequicontinuous subsets onto each other. Suppose thatL(X;Y){\displaystyle L(X;Y)} is endowed with aG{\displaystyle {\mathcal {G}}}-topology and thatH{\displaystyle {\mathcal {H}}} denotes the closures inC{\displaystyle C} of sets inG.{\displaystyle {\mathcal {G}}.} Then the mapI:LG(X;Y)LH(C;Y){\displaystyle I:L_{\mathcal {G}}(X;Y)\to L_{\mathcal {H}}(C;Y)} is also a TVS-isomorphism.[26]

Examples and sufficient conditions for a complete TVS

[edit]

Theorem[11]Letd{\displaystyle d} beany (not assumed to be translation-invariant) metric on a vector spaceX{\displaystyle X} such that the topologyτ{\displaystyle \tau } induced byd{\displaystyle d} onX{\displaystyle X} makes(X,τ){\displaystyle (X,\tau )} into a topological vector space. If(X,d){\displaystyle (X,d)} is a complete metric space then(X,τ){\displaystyle (X,\tau )} is a complete-TVS.

Properties

[edit]

Complete TVSs

[edit]

Every TVS has acompletion and every Hausdorff TVS has a Hausdorff completion.[36] Every complete TVS isquasi-complete space andsequentially complete.[37] However, the converses of the above implications are generally false.[37] There exists asequentially complete locally convex TVS that is notquasi-complete.[29]

If a TVS has a complete neighborhood of the origin then it is complete.[38] Every completepseudometrizable TVS is abarrelled space and aBaire space (and thus non-meager).[39] The dimension of a complete metrizable TVS is either finite or uncountable.[19]

Cauchy nets and prefilters

[edit]

Anyneighborhood basis of any point in a TVS is a Cauchy prefilter.

Every convergent net (respectively, prefilter) in a TVS is necessarily a Cauchy net (respectively, a Cauchy prefilter).[6] Any prefilter that is subordinate to (that is, finer than) a Cauchy prefilter is necessarily also a Cauchy prefilter[6] and any prefilter finer than a Cauchy prefilter is also a Cauchy prefilter. The filter associated with a sequence in a TVS is Cauchy if and only if the sequence is a Cauchy sequence. Every convergent prefilter is a Cauchy prefilter.

IfX{\displaystyle X} is a TVS and ifxX{\displaystyle x\in X} is a cluster point of a Cauchy net (respectively, Cauchy prefilter), then that Cauchy net (respectively, that Cauchy prefilter) converges tox{\displaystyle x} inX.{\displaystyle X.}[3] If a Cauchy filter in a TVS has anaccumulation pointx{\displaystyle x} then it converges tox.{\displaystyle x.}

Uniformly continuous maps send Cauchy nets to Cauchy nets.[3] A Cauchy sequence in a Hausdorff TVSX,{\displaystyle X,} when considered as a set, is not necessarilyrelatively compact (that is, its closure inX{\displaystyle X} is not necessarily compact[note 9]) although it is precompact (that is, its closure in the completion ofX{\displaystyle X} is compact).

Every Cauchy sequence is abounded subset but this is not necessarily true of Cauchy net. For example, letN{\displaystyle \mathbb {N} } have it usual order, let{\displaystyle \,\leq \,} denote anypreorder on the non-indiscrete TVSX{\displaystyle X} (that is,X{\displaystyle X} does not have thetrivial topology; it is also assumed thatXN={\displaystyle X\cap \mathbb {N} =\varnothing }) and extend these two preorders to the unionI =def XN{\displaystyle I~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~X\cup \mathbb {N} } by declaring thatxn{\displaystyle x\leq n} holds for everyxX{\displaystyle x\in X} andnN.{\displaystyle n\in \mathbb {N} .} Letf:IX{\displaystyle f:I\to X} be defined byf(i)=i{\displaystyle f(i)=i} ifiX{\displaystyle i\in X} andf(i)=0{\displaystyle f(i)=0} otherwise (that is, ifiN{\displaystyle i\in \mathbb {N} }), which is a net inX{\displaystyle X} since the preordered set(I,){\displaystyle (I,\leq )} isdirected (this preorder onI{\displaystyle I} is alsopartial order (respectively, atotal order) if this is true of(X,){\displaystyle (X,\leq )}). This netf{\displaystyle f} is a Cauchy net inX{\displaystyle X} because it converges to the origin, but the set{f(i):iI}=X{\displaystyle \{f(i):i\in I\}=X} is not a bounded subset ofX{\displaystyle X} (becauseX{\displaystyle X} does not have the trivial topology).

Suppose thatX=(Xi)iI{\displaystyle X_{\bullet }=\left(X_{i}\right)_{i\in I}} is a family of TVSs and thatX{\displaystyle X} denotes the product of these TVSs. Suppose that for every indexi,{\displaystyle i,}Bi{\displaystyle {\mathcal {B}}_{i}} is a prefilter onXi.{\displaystyle X_{i}.} Then the product of this family of prefilters is a Cauchy filter onX{\displaystyle X} if and only if eachBi{\displaystyle {\mathcal {B}}_{i}} is a Cauchy filter onXi.{\displaystyle X_{i}.}[17]

Maps

[edit]

Iff:XY{\displaystyle f:X\to Y} is an injectivetopological homomorphism from a complete TVS into a Hausdorff TVS then the image off{\displaystyle f} (that is,f(X){\displaystyle f(X)}) is a closed subspace ofY.{\displaystyle Y.}[34] Iff:XY{\displaystyle f:X\to Y} is atopological homomorphism from a completemetrizable TVS into a Hausdorff TVS then the range off{\displaystyle f} is a closed subspace ofY.{\displaystyle Y.}[34] Iff:XY{\displaystyle f:X\to Y} is auniformly continuous map between two Hausdorff TVSs then the image underf{\displaystyle f} of a totally bounded subset ofX{\displaystyle X} is a totally bounded subset ofY.{\displaystyle Y.}[40]

Uniformly continuous extensions

Suppose thatf:DY{\displaystyle f:D\to Y} is a uniformly continuous map from a dense subsetD{\displaystyle D} of a TVSX{\displaystyle X} into a complete Hausdorff TVSY.{\displaystyle Y.} Thenf{\displaystyle f} has a unique uniformly continuous extension to all ofX.{\displaystyle X.}[3] If in additionf{\displaystyle f} is a homomorphism then its unique uniformly continuous extension is also a homomorphism.[3] This remains true if "TVS" is replaced by "commutative topological group."[3] The mapf{\displaystyle f} is not required to be a linear map and thatD{\displaystyle D} is not required to be a vector subspace ofX.{\displaystyle X.}

Uniformly continuous linear extensions

Supposef:XY{\displaystyle f:X\to Y} be a continuous linear operator between two Hausdorff TVSs. IfM{\displaystyle M} is a dense vector subspace ofX{\displaystyle X} and if the restrictionf|M:MY{\displaystyle f{\big \vert }_{M}:M\to Y} toM{\displaystyle M} is atopological homomorphism thenf:XY{\displaystyle f:X\to Y} is also a topological homomorphism.[41] So ifC{\displaystyle C} andD{\displaystyle D} are Hausdorff completions ofX{\displaystyle X} andY,{\displaystyle Y,} respectively, and iff:XY{\displaystyle f:X\to Y} is a topological homomorphism, thenf{\displaystyle f}'s unique continuous linear extensionF:CD{\displaystyle F:C\to D} is a topological homomorphism. (Note that it's possible forf:XY{\displaystyle f:X\to Y} to be surjective but forF:CD{\displaystyle F:C\to D} tonot be injective.)[41]

SupposeX{\displaystyle X} andY{\displaystyle Y} are Hausdorff TVSs,M{\displaystyle M} is a dense vector subspace ofX,{\displaystyle X,} andN{\displaystyle N} is a dense vector subspaces ofY.{\displaystyle Y.} IfM{\displaystyle M} are andN{\displaystyle N} are topologically isomorphic additive subgroups via a topological homomorphismf{\displaystyle f} then the same is true ofX{\displaystyle X} andY{\displaystyle Y} via the unique uniformly continuous extension off{\displaystyle f} (which is also a homeomorphism).[42]

Subsets

[edit]

Complete subsets

Every complete subset of a TVS issequentially complete. A complete subset of a Hausdorff TVSX{\displaystyle X} is a closed subset ofX.{\displaystyle X.}[3][38]

Every compact subset of a TVS is complete (even if the TVS is not Hausdorff or not complete).[3][38] Closed subsets of a complete TVS are complete; however, if a TVSX{\displaystyle X} is not complete thenX{\displaystyle X} is a closed subset ofX{\displaystyle X} that is not complete. The empty set is complete subset of every TVS. IfC{\displaystyle C} is a complete subset of a TVS (the TVS is not necessarily Hausdorff or complete) then any subset ofC{\displaystyle C} that is closed inC{\displaystyle C} is complete.[38]

Topological complements

IfX{\displaystyle X} is a non-normableFréchet space on which there exists a continuous norm thenX{\displaystyle X} contains a closed vector subspace that has notopological complement.[29] IfX{\displaystyle X} is a complete TVS andM{\displaystyle M} is a closed vector subspace ofX{\displaystyle X} such thatX/M{\displaystyle X/M} is not complete, thenH{\displaystyle H} doesnot have atopological complement inX.{\displaystyle X.}[29]

Subsets of completions

LetM{\displaystyle M} be aseparable locally convexmetrizable topological vector space and letC{\displaystyle C} be its completion. IfS{\displaystyle S} is a bounded subset ofC{\displaystyle C} then there exists a bounded subsetR{\displaystyle R} ofX{\displaystyle X} such thatSclCR.{\displaystyle S\subseteq \operatorname {cl} _{C}R.}[29]

Relation to compact subsets

A subset of a TVS (not assumed to be Hausdorff or complete) iscompact if and only if it is complete andtotally bounded.[43][proof 2] Thus a closed andtotally bounded subset of a complete TVS is compact.[44][3]

In a Hausdorff locally convex TVS, the convex hull of aprecompact set is again precompact.[45] Consequently, in a complete locally convex Hausdorff TVS, the closed convex hull of a compact subset is again compact.[46]

The convex hull of compact subset of aHilbert space isnot necessarily closed and so alsonot necessarily compact. For example, letH{\displaystyle H} be the separable Hilbert space2(N){\displaystyle \ell ^{2}(\mathbb {N} )} of square-summable sequences with the usual norm2{\displaystyle \|\cdot \|_{2}} and leten=(0,,0,1,0,){\displaystyle e_{n}=(0,\ldots ,0,1,0,\ldots )} be the standardorthonormal basis (that is1{\displaystyle 1} at thenth{\displaystyle n^{\text{th}}}-coordinate). The closed setS={0}{1nen}{\displaystyle S=\{0\}\cup \left\{{\tfrac {1}{n}}e_{n}\right\}} is compact but its convex hullcoS{\displaystyle \operatorname {co} S} isnot a closed set becauseh:=n=112n1nen{\displaystyle h:=\sum _{n=1}^{\infty }{\tfrac {1}{2^{n}}}{\tfrac {1}{n}}e_{n}} belongs to the closure ofcoS{\displaystyle \operatorname {co} S} inH{\displaystyle H} buthcoS{\displaystyle h\not \in \operatorname {co} S} (since every sequencezcoS{\displaystyle z\in \operatorname {co} S} is a finiteconvex combination of elements ofS{\displaystyle S} and so is necessarily0{\displaystyle 0} in all but finitely many coordinates, which is not true ofh{\displaystyle h}).[47] However, like in all complete Hausdorff locally convex spaces, theclosed convex hullK:=co¯S{\displaystyle K:={\overline {\operatorname {co} }}S} of this compact subset is compact.[46] The vector subspaceX:=spanS{\displaystyle X:=\operatorname {span} S} is apre-Hilbert space when endowed with the substructure that the Hilbert spaceH{\displaystyle H} induces on it butX{\displaystyle X} is not complete andhKX{\displaystyle h\not \in K\cap X} (sincehX{\displaystyle h\not \in X}). The closed convex hull ofS{\displaystyle S} inX{\displaystyle X} (here, "closed" means with respect toX,{\displaystyle X,} and not toH{\displaystyle H} as before) is equal toKX,{\displaystyle K\cap X,} which is not compact (because it is not a complete subset). This shows that in a Hausdorff locally convex space that is not complete, the closed convex hull of compact subset mightfail to be compact (although it will beprecompact/totally bounded).

Every complete totally bounded set is relatively compact.[3] IfX{\displaystyle X} is any TVS then the quotient mapq:XX/clX{0}{\displaystyle q:X\to X/\operatorname {cl} _{X}\{0\}} is aclosed map[48] and thusS+clX{0}clXS{\displaystyle S+\operatorname {cl} _{X}\{0\}\subseteq \operatorname {cl} _{X}S} A subsetS{\displaystyle S} of a TVSX{\displaystyle X} is totally bounded if and only if its image under the canonical quotient mapq:XX/clX{0}{\displaystyle q:X\to X/\operatorname {cl} _{X}\{0\}} is totally bounded.[19] ThusS{\displaystyle S} is totally bounded if and only ifS+clX{0}{\displaystyle S+\operatorname {cl} _{X}\{0\}} is totally bounded. In any TVS, the closure of a totally bounded subset is again totally bounded.[3] In a locally convex space, the convex hull and thedisked hull of a totally bounded set is totally bounded.[36] IfS{\displaystyle S} is a subset of a TVSX{\displaystyle X} such that every sequence inS{\displaystyle S} has a cluster point inS{\displaystyle S} thenS{\displaystyle S} is totally bounded.[19] A subsetS{\displaystyle S} of a Hausdorff TVSX{\displaystyle X} is totally bounded if and only if every ultrafilter onS{\displaystyle S} is Cauchy, which happens if and only if it is pre-compact (that is, its closure in the completion ofX{\displaystyle X} is compact).[40]

IfSX{\displaystyle S\subseteq X} is compact, thenclXS=S+clX{0}{\displaystyle \operatorname {cl} _{X}S=S+\operatorname {cl} _{X}\{0\}} and this set is compact. Thus the closure of a compact set is compact[note 10] (that is, all compact sets arerelatively compact).[49] Thus the closure of a compact set is compact. Every relatively compact subset of a Hausdorff TVS is totally bounded.[40]

In a complete locally convex space, the convex hull and the disked hull of a compact set are both compact.[36] More generally, ifK{\displaystyle K} is a compact subset of a locally convex space, then the convex hullcoK{\displaystyle \operatorname {co} K} (resp. the disked hullcobalK{\displaystyle \operatorname {cobal} K}) is compact if and only if it is complete.[36] Every subsetS{\displaystyle S} ofclX{0}{\displaystyle \operatorname {cl} _{X}\{0\}} is compact and thus complete.[proof 3] In particular, ifX{\displaystyle X} is not Hausdorff then there exist compact complete sets that are not closed.[3]

See also

[edit]

Notes

[edit]
  1. ^A metricD{\displaystyle D} on a vector spaceX{\displaystyle X} is said to betranslation invariant ifD(x,y)=D(x+z,y+z){\displaystyle D(x,y)=D(x+z,y+z)} for all vectorsx,y,zX.{\displaystyle x,y,z\in X.} A metric that is induced by a norm is always translation invariant.
  2. ^Completeness ofnormed spaces andmetrizable TVSs are defined in terms ofnorms andmetrics. In general, many different norms (for example,equivalent norms) and metrics may be used to determine completeness of such space. This stands in contrast to the uniqueness of this translation-invariant canonical uniformity.
  3. ^Every sequence is also a net.
  4. ^The normed space(R,||){\displaystyle (\mathbb {R} ,|\cdot |)} is a Banach space where the absolute value is a norm that induces the usual Euclidean topology onR.{\displaystyle \mathbb {R} .} Define a metricD{\displaystyle D} onR{\displaystyle \mathbb {R} } byD(x,y)=|arctan(x)arctan(y)|{\displaystyle D(x,y)=\left|\arctan(x)-\arctan(y)\right|} for allx,yR,{\displaystyle x,y\in \mathbb {R} ,} where one may show thatD{\displaystyle D} induces the usual Euclidean topology onR.{\displaystyle \mathbb {R} .} However,D{\displaystyle D} is not a complete metric since the sequencex=(xi)i=1{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i=1}^{\infty }} defined byxi=i{\displaystyle x_{i}=i} is aD{\displaystyle D}-Cauchy sequence that does not converge inR{\displaystyle \mathbb {R} } to any point ofR.{\displaystyle \mathbb {R} .} Note also that thisD{\displaystyle D}-Cauchy sequence is not a Cauchy sequence in(R,||){\displaystyle (\mathbb {R} ,|\cdot |)} (that is, it is not a Cauchy sequence with respect to the norm||{\displaystyle |\cdot |}).
  5. ^Not assumed to be translation-invariant.
  6. ^Let(C([0,1]),){\displaystyle \left(C([0,1]),\|\cdot \|_{\infty }\right)} denotes theBanach space of continuous functions with the supremum norm, letX=C([0,1]){\displaystyle X=C([0,1])} whereX{\displaystyle X} is given the topology induced by,{\displaystyle \|\cdot \|_{\infty },} and denote the restriction of theL1-norm toC([0,1]){\displaystyle C([0,1])} by1.{\displaystyle \|\cdot \|_{1}.} Then one may show that1{\displaystyle \|\cdot \|_{1}\leq \|\cdot \|_{\infty }} so that the norm1:XR{\displaystyle \|\cdot \|_{1}:X\to \mathbb {R} } is a continuous function. However,1{\displaystyle \|\cdot \|_{1}} isnot equivalent to the norm{\displaystyle \|\cdot \|_{\infty }} and so in particular,(C([0,1]),1){\displaystyle \left(C([0,1]),\|\cdot \|_{1}\right)} isnot a Banach space.
  7. ^This particular quotient mapq:XX/I{\displaystyle q:X\to X/I} is in fact also a closed map.
  8. ^Explicitly, this map is defined as follows: for eachxX,{\displaystyle x\in X,} let(i,h)=A(x){\displaystyle (i,h)=A(x)} and so thatB(x) =def (i,InHh).{\displaystyle B(x)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\left(i,\operatorname {In} _{H}h\right).} ThenB(i+h)=(i,InHh){\displaystyle B(i+h)=\left(i,\operatorname {In} _{H}h\right)} holds for alliI{\displaystyle i\in I} andhH.{\displaystyle h\in H.}
  9. ^IfX{\displaystyle X} is anormable TVS such that for every Cauchy sequencex=(xi)i=1,{\displaystyle x_{\bullet }=\left(x_{i}\right)_{i=1}^{\infty },} the closure ofS =def {x1,x2,,}{\displaystyle S~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~\{x_{1},x_{2},\ldots ,\}} inX{\displaystyle X} is compact (and thussequentially compact) then this guarantees that there always exist somexclXS{\displaystyle x\in \operatorname {cl} _{X}S} such thatxx{\displaystyle x_{\bullet }\to x} inX.{\displaystyle X.} Thus any normed space with this property is necessarily sequentially complete. Since not all normed spaces are complete, the closure of a Cauchy sequence is not necessarily compact.
  10. ^In general topology, the closure of a compact subset of a non-Hausdorff space may fail to be compact (for example, theparticular point topology on an infinite set). This result shows that this does not happen in non-Hausdorff TVSs. The proof uses the fact thatS{\displaystyle S} is compact (but possibly not closed) andclX{0}{\displaystyle \operatorname {cl} _{X}\{0\}} is both closed and compact so thatS+clX{0},{\displaystyle S+\operatorname {cl} _{X}\{0\},} which is the image of the compact setS+clX{0}{\displaystyle S+\operatorname {cl} _{X}\{0\}} under the continuous addition map+:X×XX,{\displaystyle \cdot +\cdot :X\times X\to X,} is also compact. Recall also that the sum of a compact set (that is,S{\displaystyle S}) and a closed set is closed soS+clX{0}{\displaystyle S+\operatorname {cl} _{X}\{0\}} is closed inX.{\displaystyle X.}

Proofs

  1. ^LetW{\displaystyle W} be a neighborhood of the origin inX.{\displaystyle X.} SinceA(W){\displaystyle A(W)} is a neighborhood of0{\displaystyle 0} inI×H,{\displaystyle I\times H,} there exists an open (resp. closed) neighborhoodV{\displaystyle V} of0{\displaystyle 0} inH{\displaystyle H} such thatI×VA(W){\displaystyle I\times V\subseteq A(W)} is a neighborhood of the origin. Clearly,V{\displaystyle V} is open (resp. closed) if and only ifI×V{\displaystyle I\times V} is open (resp. closed). LetU=I+V{\displaystyle U=I+V} so thatA(U)=I×VA(W){\displaystyle A(U)=I\times V\subseteq A(W)} whereU{\displaystyle U} is open (resp. closed) if and only ifV{\displaystyle V} is open (resp. closed).
  2. ^SupposeS{\displaystyle S} is compact inX{\displaystyle X} and letC{\displaystyle {\mathcal {C}}} be a Cauchy filter onS.{\displaystyle S.} LetD={clSC : CC}{\displaystyle {\mathcal {D}}=\left\{\operatorname {cl} _{S}C~:~C\in {\mathcal {C}}\right\}} so thatD{\displaystyle {\mathcal {D}}} is a Cauchy filter of closed sets. SinceD{\displaystyle {\mathcal {D}}} has the finite intersection property, there exists somesS{\displaystyle s\in S} such thatsclSC{\displaystyle s\operatorname {cl} _{S}C} for allCC{\displaystyle C\in {\mathcal {C}}} so {sclC{\displaystyle s\in \operatorname {cl} {\mathcal {C}}} (that is,s{\displaystyle s} is an accumulation point ofC{\displaystyle {\mathcal {C}}}). SinceC{\displaystyle {\mathcal {C}}} is Cauchy,Cx{\displaystyle {\mathcal {C}}\to x} inS.{\displaystyle S.} ThusS{\displaystyle S} is complete. ThatS{\displaystyle S} is also totally bounded follows immediately from the compactness ofS.{\displaystyle S.}
  3. ^Given any open cover ofS,{\displaystyle S,} pick any open setU{\displaystyle U} from that cover that contains the origin. SinceU{\displaystyle U} is a neighborhood of the origin,U{\displaystyle U} containsclX{0}{\displaystyle \operatorname {cl} _{X}\{0\}} and thus containsS.{\displaystyle S.}

Citations

[edit]
  1. ^abSchaefer & Wolff 1999, pp. 1–11.
  2. ^abEdwards 1995, p. 61.
  3. ^abcdefghijklmnopNarici & Beckenstein 2011, pp. 47–66.
  4. ^Narici & Beckenstein 2011, p. 48.
  5. ^Zălinescu 2002, pp. 1–23.
  6. ^abcdefghNarici & Beckenstein 2011, pp. 48–51.
  7. ^abcdeSchaefer & Wolff 1999, pp. 12–19.
  8. ^Narici & Beckenstein 2011, pp. 64–66.
  9. ^Wilansky 2013, p. 29.
  10. ^abcNarici & Beckenstein 2011, pp. 47–51.
  11. ^abSchaefer & Wolff 1999, p. 35.
  12. ^Klee, V. L. (1952)."Invariant metrics in groups (solution of a problem of Banach)"(PDF).Proc. Amer. Math. Soc.3 (3):484–487.doi:10.1090/s0002-9939-1952-0047250-4.
  13. ^abConrad, Keith."Equivalence of norms"(PDF).kconrad.math.uconn.edu. RetrievedSeptember 7, 2020.
  14. ^see Corollary1.4.18, p.32 inMegginson (1998).
  15. ^abNarici & Beckenstein 2011, pp. 60–61.
  16. ^abcdeNarici & Beckenstein 2011, pp. 93–113.
  17. ^abcdefgHorváth 1966, pp. 139–141.
  18. ^Wilansky 2013, p. 63.
  19. ^abcdefSchaefer & Wolff 1999, pp. 12–35.
  20. ^where for alliI{\displaystyle i\in I} andhH,{\displaystyle h\in H,}InH(i+h) =def i+h.{\displaystyle \operatorname {In} _{H}(i+h)~{\stackrel {\scriptscriptstyle {\text{def}}}{=}}~i+h.}
  21. ^abSchaefer & Wolff 1999, pp. 36–72.
  22. ^Schaefer & Wolff 1999, pp. 73−121.
  23. ^Jarchow 1981, pp. 151, 157.
  24. ^abcdJarchow 1981, pp. 175−178.
  25. ^abTrèves 2006, pp. 112–125.
  26. ^abcdeSchaefer & Wolff 1999, pp. 73–121.
  27. ^abSchaefer & Wolff 1999, pp. 68–72.
  28. ^Schaefer & Wolff 1999, pp. 122–202.
  29. ^abcdefSchaefer & Wolff 1999, pp. 190–202.
  30. ^Narici & Beckenstein 2011, pp. 225–273.
  31. ^Schaefer & Wolff 1999, pp. 199–202.
  32. ^abcdJarchow 1981, pp. 56–73.
  33. ^Narici & Beckenstein 2011, p. 57.
  34. ^abcdHorváth 1966, pp. 129–141.
  35. ^abcNarici & Beckenstein 2011, pp. 441–457.
  36. ^abcdNarici & Beckenstein 2011, pp. 67–113.
  37. ^abNarici & Beckenstein 2011, pp. 155–176.
  38. ^abcdNarici & Beckenstein 2011, pp. 115–154.
  39. ^Narici & Beckenstein 2011, pp. 371–423.
  40. ^abcHorváth 1966, pp. 145–149.
  41. ^abSchaefer & Wolff 1999, p. 116.
  42. ^Narici & Beckenstein 2011, p. 59.
  43. ^Narici & Beckenstein 2011, pp. 55–56.
  44. ^Narici & Beckenstein 2011, pp. 55–66.
  45. ^Trèves 2006, p. 67.
  46. ^abTrèves 2006, p. 145.
  47. ^Aliprantis & Border 2006, p. 185.
  48. ^Narici & Beckenstein 2011, pp. 107–112.
  49. ^Narici & Beckenstein 2011, p. 156.

Bibliography

[edit]
Spaces
Properties
Theorems
Operators
Algebras
Open problems
Applications
Advanced topics
Basic concepts
Main results
Maps
Types of sets
Set operations
Types of TVSs
Retrieved from "https://en.wikipedia.org/w/index.php?title=Complete_topological_vector_space&oldid=1321109188"
Categories:
Hidden categories:

[8]ページ先頭

©2009-2025 Movatter.jp