<p>Given Banach spaces <i>E</i>, <i>F</i>, <i>G</i>, we denote by <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq1.gif" Format="GIF" Height="21" Rendition="HTML" Resolution="72" Type="Linedraw" Width="87" /> </InlineMediaObject> <EquationSource Format="TEX">\(\mathcal L^2_\text {I}(E,F;G)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mi mathvariant="script">L</mi> <mtext>I</mtext> <mn>2</mn> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo>;</mo> <mi>G</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> the space of integral bilinear mappings from <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq2.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="50" /> </InlineMediaObject> <EquationSource Format="TEX">\(E\times F\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>E</mi> <mo>×</mo> <mi>F</mi> </mrow> </math></EquationSource> </InlineEquation> into <i>G</i>, by <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq3.gif" Format="GIF" Height="21" Rendition="HTML" Resolution="72" Type="Linedraw" Width="90" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {L}}^2_\text {N}(E,F;G)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mtext>N</mtext> <mn>2</mn> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo>;</mo> <mi>G</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> the space of nuclear bilinear maps, and by <InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq4.gif" Format="GIF" Height="24" Rendition="HTML" Resolution="72" Type="Linedraw" Width="99" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {L}}^k_{\ell _1,\text {I}}(E,F;G)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mrow> <msub> <mi>ℓ</mi> <mn>1</mn> </msub> <mo>,</mo> <mtext>I</mtext> </mrow> <mi>k</mi> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo>;</mo> <mi>G</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> the space of left integral <InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq5.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\ell _1\)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>ℓ</mi> <mn>1</mn> </msub> </math></EquationSource> </InlineEquation>-factorable bilinear maps. When <i>G</i> is the scalar field, it is omitted. It is well-known that <InlineEquation ID="IEq6"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq6.gif" Format="GIF" Height="23" Rendition="HTML" Resolution="72" Type="Linedraw" Width="125" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {L}}^2_{\text{ N }}\subseteq {\mathcal {L}}^2_{\ell _1,\text {I}} \subseteq {\mathcal {L}}^2_\text {I}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mrow> <mspace width="0.333333em" /> <mtext>N</mtext> <mspace width="0.333333em" /> </mrow> <mn>2</mn> </msubsup> <mo>⊆</mo> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mrow> <msub> <mi>ℓ</mi> <mn>1</mn> </msub> <mo>,</mo> <mtext>I</mtext> </mrow> <mn>2</mn> </msubsup> <mo>⊆</mo> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mtext>I</mtext> <mn>2</mn> </msubsup> </mrow> </math></EquationSource> </InlineEquation>. We prove that, if the injective tensor product <InlineEquation ID="IEq7"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq7.gif" Format="GIF" Height="20" Rendition="HTML" Resolution="72" Type="Linedraw" Width="48" /> </InlineMediaObject> <EquationSource Format="TEX">\(E{{\widehat{\otimes }}}_\epsilon F\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>E</mi> <msub> <mover accent="true"> <mo>⊗</mo> <mo stretchy="true">^</mo> </mover> <mi>ϵ</mi> </msub> <mi>F</mi> </mrow> </math></EquationSource> </InlineEquation> contains no copy of <InlineEquation ID="IEq8"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq5.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="15" /> </InlineMediaObject> <EquationSource Format="TEX">\(\ell _1\)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>ℓ</mi> <mn>1</mn> </msub> </math></EquationSource> </InlineEquation>, then <InlineEquation ID="IEq9"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq9.gif" Format="GIF" Height="21" Rendition="HTML" Resolution="72" Type="Linedraw" Width="157" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {L}}^2_\text {I}(E,F)\equiv {\mathcal {L}}^2_\text {N}(E,F)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mtext>I</mtext> <mn>2</mn> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo stretchy="false">)</mo> </mrow> <mo>≡</mo> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mtext>N</mtext> <mn>2</mn> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation>, where the symbol <InlineEquation ID="IEq10"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq10.gif" Format="GIF" Height="9" Rendition="HTML" Resolution="72" Type="Linedraw" Width="17" /> </InlineMediaObject> <EquationSource Format="TEX">\(\equiv \)</EquationSource> <EquationSource Format="MATHML"><math> <mo>≡</mo> </math></EquationSource> </InlineEquation> means “isometrically isomorphic". It is also shown that the latter identity is equivalent to <InlineEquation ID="IEq11"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq11.gif" Format="GIF" Height="24" Rendition="HTML" Resolution="72" Type="Linedraw" Width="210" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {L}}^k_{\ell _1,\text {I}}(E,F;G)\equiv {\mathcal {L}}^2_\text {N}(E,F;G)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mrow> <msub> <mi>ℓ</mi> <mn>1</mn> </msub> <mo>,</mo> <mtext>I</mtext> </mrow> <mi>k</mi> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo>;</mo> <mi>G</mi> <mo stretchy="false">)</mo> </mrow> <mo>≡</mo> <msubsup> <mrow> <mi mathvariant="script">L</mi> </mrow> <mtext>N</mtext> <mn>2</mn> </msubsup> <mrow> <mo stretchy="false">(</mo> <mi>E</mi> <mo>,</mo> <mi>F</mi> <mo>;</mo> <mi>G</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> for every Banach space <i>G</i>. We find other necessary conditions for this identity in terms of isomorphic properties of <i>E</i> and <i>F</i>. All the above mentioned results are also valid for <i>k</i>-linear mappings <InlineEquation ID="IEq12"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq12.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="54" /> </InlineMediaObject> <EquationSource Format="TEX">\((k\ge 2)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mo stretchy="false">(</mo> <mi>k</mi> <mo>≥</mo> <mn>2</mn> <mo stretchy="false">)</mo> </mrow> </math></EquationSource> </InlineEquation>. The arguments used allow for identification of a large class of integral multilinear mappings which are automatically nuclear. As a consequence, we give a simple proof of a Grothendieck type theorem stating that a multilinear mapping on a product of <InlineEquation ID="IEq13"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq13.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="42" /> </InlineMediaObject> <EquationSource Format="TEX">\(L_1(\mu )\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>L</mi> <mn>1</mn> </msub> <mrow> <mo stretchy="false">(</mo> <mi>μ</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> spaces is Riesz-representable if and only if its composition with the canonical inclusions <InlineEquation ID="IEq14"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq14.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="115" /> </InlineMediaObject> <EquationSource Format="TEX">\(L_\infty (\mu )\rightarrow L_1(\mu )\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>L</mi> <mi>∞</mi> </msub> <mrow> <mo stretchy="false">(</mo> <mi>μ</mi> <mo stretchy="false">)</mo> </mrow> <mo stretchy="false">→</mo> <msub> <mi>L</mi> <mn>1</mn> </msub> <mrow> <mo stretchy="false">(</mo> <mi>μ</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> is nuclear. A more involved proof of this result had been given in 2024. It is proved that the spaces <InlineEquation ID="IEq15"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq15.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="79" /> </InlineMediaObject> <EquationSource Format="TEX">\(E_1,\ldots , E_k\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>E</mi> <mn>1</mn> </msub> <mo>,</mo> <mo>…</mo> <mo>,</mo> <msub> <mi>E</mi> <mi>k</mi> </msub> </mrow> </math></EquationSource> </InlineEquation> are <InlineEquation ID="IEq16"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq16.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="30" /> </InlineMediaObject> <EquationSource Format="TEX">\(\mathscr {L}_\infty \)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi mathvariant="script">L</mi> <mi>∞</mi> </msub> </math></EquationSource> </InlineEquation>-spaces if and only if the 1-dominated <i>k</i>-linear mappings from <InlineEquation ID="IEq17"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="12215_2025_1275_Article_IEq17.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="103" /> </InlineMediaObject> <EquationSource Format="TEX">\(E_1\times \cdots \times E_k\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>E</mi> <mn>1</mn> </msub> <mo>×</mo> <mo>⋯</mo> <mo>×</mo> <msub> <mi>E</mi> <mi>k</mi> </msub> </mrow> </math></EquationSource> </InlineEquation> into an arbitrary Banach space coincide with a subclass of the <i>k</i>-linear integral mappings (called S-factorable). This extends to the multilinear case a result of Stegall and Retherford, and strengthens a previous partial extension of 2002.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Spaces of integral and nuclear multilinear mappings

  • Gabriela Badea,
  • Raffaella Cilia,
  • Joaquín M. Gutiérrez

摘要

Given Banach spaces E, F, G, we denote by \(\mathcal L^2_\text {I}(E,F;G)\) L I 2 ( E , F ; G ) the space of integral bilinear mappings from \(E\times F\) E × F into G, by \({\mathcal {L}}^2_\text {N}(E,F;G)\) L N 2 ( E , F ; G ) the space of nuclear bilinear maps, and by \({\mathcal {L}}^k_{\ell _1,\text {I}}(E,F;G)\) L 1 , I k ( E , F ; G ) the space of left integral \(\ell _1\) 1 -factorable bilinear maps. When G is the scalar field, it is omitted. It is well-known that \({\mathcal {L}}^2_{\text{ N }}\subseteq {\mathcal {L}}^2_{\ell _1,\text {I}} \subseteq {\mathcal {L}}^2_\text {I}\) L N 2 L 1 , I 2 L I 2 . We prove that, if the injective tensor product \(E{{\widehat{\otimes }}}_\epsilon F\) E ^ ϵ F contains no copy of \(\ell _1\) 1 , then \({\mathcal {L}}^2_\text {I}(E,F)\equiv {\mathcal {L}}^2_\text {N}(E,F)\) L I 2 ( E , F ) L N 2 ( E , F ) , where the symbol \(\equiv \) means “isometrically isomorphic". It is also shown that the latter identity is equivalent to \({\mathcal {L}}^k_{\ell _1,\text {I}}(E,F;G)\equiv {\mathcal {L}}^2_\text {N}(E,F;G)\) L 1 , I k ( E , F ; G ) L N 2 ( E , F ; G ) for every Banach space G. We find other necessary conditions for this identity in terms of isomorphic properties of E and F. All the above mentioned results are also valid for k-linear mappings \((k\ge 2)\) ( k 2 ) . The arguments used allow for identification of a large class of integral multilinear mappings which are automatically nuclear. As a consequence, we give a simple proof of a Grothendieck type theorem stating that a multilinear mapping on a product of \(L_1(\mu )\) L 1 ( μ ) spaces is Riesz-representable if and only if its composition with the canonical inclusions \(L_\infty (\mu )\rightarrow L_1(\mu )\) L ( μ ) L 1 ( μ ) is nuclear. A more involved proof of this result had been given in 2024. It is proved that the spaces \(E_1,\ldots , E_k\) E 1 , , E k are \(\mathscr {L}_\infty \) L -spaces if and only if the 1-dominated k-linear mappings from \(E_1\times \cdots \times E_k\) E 1 × × E k into an arbitrary Banach space coincide with a subclass of the k-linear integral mappings (called S-factorable). This extends to the multilinear case a result of Stegall and Retherford, and strengthens a previous partial extension of 2002.