Namespaces
Variants
Actions

Difference between revisions of "Lannes-T-functor"

From Encyclopedia of Mathematics
Jump to: navigation, search
m (latex)
m (tex done)
 
Line 1: Line 1:
 +
 
<!--This article has been texified automatically. Since there was no Nroff source code for this article,  
 
<!--This article has been texified automatically. Since there was no Nroff source code for this article,  
 
the semi-automatic procedure described at https://encyclopediaofmath.org/wiki/User:Maximilian_Janisch/latexlist
 
the semi-automatic procedure described at https://encyclopediaofmath.org/wiki/User:Maximilian_Janisch/latexlist
Line 6: Line 7:
 
Out of 100 formulas, 98 were replaced by TEX code.-->
 
Out of 100 formulas, 98 were replaced by TEX code.-->
  
{{TEX|semi-auto}}{{TEX|part}}
+
$$
 +
\newcommand{\Sq}{\mathcal{S}\text{q}}
 +
$$
 +
{{TEX|semi-auto}}{{TEX|done}}
 
The calculation of the homotopy type of the space of continuous mappings $\operatorname{Map}( X , Y )$ is a fundamental problem of [[Homotopy|homotopy]] theory. The set of path components, $\pi_0 \; \operatorname { Map } ( X , Y ) = [ X , Y ]$ corresponds to the homotopy classes of such mappings. There are relatively few cases for which this information is explicitly known (as of 1998). A major impact of the work [[#References|[a1]]] of J. Lannes on unstable modules and the T-functor has been to expand this knowledge to include many cases in which the sources and targets are classifying spaces of finite and compact Lie groups (cf. also [[Lie group|Lie group]]).
 
The calculation of the homotopy type of the space of continuous mappings $\operatorname{Map}( X , Y )$ is a fundamental problem of [[Homotopy|homotopy]] theory. The set of path components, $\pi_0 \; \operatorname { Map } ( X , Y ) = [ X , Y ]$ corresponds to the homotopy classes of such mappings. There are relatively few cases for which this information is explicitly known (as of 1998). A major impact of the work [[#References|[a1]]] of J. Lannes on unstable modules and the T-functor has been to expand this knowledge to include many cases in which the sources and targets are classifying spaces of finite and compact Lie groups (cf. also [[Lie group|Lie group]]).
  
 
The work of N. Steenrod and others assigns in a natural way to each [[Topological space|topological space]] $X$ and each prime number $p$ an algebraic model, consisting of a [[Graded algebra|graded algebra]] $H ^ { * } ( X , \mathbf{F} _ { p } ) = R ^ { * }$ over $\mathbf{F} _ { p }$ and an algebra $\mathcal{A} _ { p }$ of natural operations, called the Steenrod algebra. Each $f : X \rightarrow Y$ induces an element $f ^ { * } \in \text { Hom}_{\text{alg} } ( H ^ { * } ( Y , {\bf F} _ { p } ) , H ^ { * } ( X , {\bf F} _ { p } ) )$ that commutes with the action of $\mathcal{A} _ { p }$. $\mathcal{A} _ { p }$ is a connected graded [[Hopf algebra|Hopf algebra]] acting on the graded algebra $R ^ { * }$.
 
The work of N. Steenrod and others assigns in a natural way to each [[Topological space|topological space]] $X$ and each prime number $p$ an algebraic model, consisting of a [[Graded algebra|graded algebra]] $H ^ { * } ( X , \mathbf{F} _ { p } ) = R ^ { * }$ over $\mathbf{F} _ { p }$ and an algebra $\mathcal{A} _ { p }$ of natural operations, called the Steenrod algebra. Each $f : X \rightarrow Y$ induces an element $f ^ { * } \in \text { Hom}_{\text{alg} } ( H ^ { * } ( Y , {\bf F} _ { p } ) , H ^ { * } ( X , {\bf F} _ { p } ) )$ that commutes with the action of $\mathcal{A} _ { p }$. $\mathcal{A} _ { p }$ is a connected graded [[Hopf algebra|Hopf algebra]] acting on the graded algebra $R ^ { * }$.
  
The hypothesis that $R ^ { * }$ is the [[Cohomology|cohomology]] of a space imposes an additional  "unstable"  condition. This is most simply stated if $p = 2$: ${\cal A} _ { 2 }$ is generated as an (non-commutative) algebra by the Steenrod operations $\{ \mathcal{S}  \operatorname {q}  ^ { i } : i \geq 0 \}$, with relations forced by its actions of the cohomology of all topological spaces. For example, $\mathcal{S} \text{q} ^ { 0 } = \operatorname{Id}$ and $\mathcal{S} \text{q} ^ { 1 } = \beta$, the modulo-$2$ Bockstein operator. The unstable condition is then that ${\cal S} \operatorname {q} ^ { i } x _ { n } = 0$ for $i &gt; n$ and $\mathcal{S} \operatorname{q} ^ { n } x _ { n } = x _ { n } ^ { 2 }$. The algebraic category $\mathcal{K}$ of unstable algebras $\{ \mathcal{R} ^ { * } \}$ over $\mathcal{A} _ { p }$ is thus an approximation to the homotopy category of topological spaces. The larger category $\mathcal U$ of unstable modules over $\mathcal{A} _ { p }$ has also proved useful.
+
The hypothesis that $R ^ { * }$ is the [[Cohomology|cohomology]] of a space imposes an additional  "unstable"  condition. This is most simply stated if $p = 2$: ${\cal A} _ { 2 }$ is generated as an (non-commutative) algebra by the Steenrod operations $\{ \Sq^{i} : i \geq 0 \}$, with relations forced by its actions of the cohomology of all topological spaces. For example, $\Sq^{0} = \operatorname{Id}$ and $\Sq^{ 1 } = \beta$, the modulo-$2$ Bockstein operator. The unstable condition is then that $\Sq^{ i } x _ { n } = 0$ for $i > n$ and $\Sq^ { n } x _ { n } = x _ { n } ^ { 2 }$. The algebraic category $\mathcal{K}$ of unstable algebras $\{ \mathcal{R} ^ { * } \}$ over $\mathcal{A} _ { p }$ is thus an approximation to the homotopy category of topological spaces. The larger category $\mathcal U$ of unstable modules over $\mathcal{A} _ { p }$ has also proved useful.
  
 
For $p &gt; 2$, the structure of $\mathcal{A} _ { p }$ and unstable actions are similar, but slightly more involved. However, in all cases, the set of relations in the Steenrod algebra and the unstable condition are derivable from the known action of $\mathcal{A} _ { p }$ on the cohomology of products of copies of $B {\bf Z} / p {\bf Z}$. In the following, explicit references to the coefficients are omitted.
 
For $p &gt; 2$, the structure of $\mathcal{A} _ { p }$ and unstable actions are similar, but slightly more involved. However, in all cases, the set of relations in the Steenrod algebra and the unstable condition are derivable from the known action of $\mathcal{A} _ { p }$ on the cohomology of products of copies of $B {\bf Z} / p {\bf Z}$. In the following, explicit references to the coefficients are omitted.
Line 67: Line 71:
 
5) If $G$ is a compact [[Lie group|Lie group]], then
 
5) If $G$ is a compact [[Lie group|Lie group]], then
  
$$T_E H^\ast BG = \Pi_{\varphi \in \operatorname{Hom}(E,G)/G} H^\ast BC_G(\operatorname{im} \varphi(E))$$
+
$$
 
+
T_E H^\ast BG = \prod_{\varphi \in \operatorname{Hom}_{\text{grp}}(E,G)/G\text{-conj}} H^\ast BC_G(\operatorname{im} \varphi(E))
<table class="eq" style="width:100%;"> <tr><td style="width:94%;text-align:center;" valign="top"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/l/l120/l120030/l120030100.png"/></td> </tr></table>
+
$$
  
 
These examples each have powerful topological consequences. For example, the first and fourth lead to new proofs of the Sullivan conjecture, originally proved by Miller and Carlsson. The last leads to a new view of the homotopy theory of classifying spaces. Most of the above is referenced in [[#References|[a2]]].
 
These examples each have powerful topological consequences. For example, the first and fourth lead to new proofs of the Sullivan conjecture, originally proved by Miller and Carlsson. The last leads to a new view of the homotopy theory of classifying spaces. Most of the above is referenced in [[#References|[a2]]].

Latest revision as of 06:16, 15 February 2024


$$ \newcommand{\Sq}{\mathcal{S}\text{q}} $$ The calculation of the homotopy type of the space of continuous mappings $\operatorname{Map}( X , Y )$ is a fundamental problem of homotopy theory. The set of path components, $\pi_0 \; \operatorname { Map } ( X , Y ) = [ X , Y ]$ corresponds to the homotopy classes of such mappings. There are relatively few cases for which this information is explicitly known (as of 1998). A major impact of the work [a1] of J. Lannes on unstable modules and the T-functor has been to expand this knowledge to include many cases in which the sources and targets are classifying spaces of finite and compact Lie groups (cf. also Lie group).

The work of N. Steenrod and others assigns in a natural way to each topological space $X$ and each prime number $p$ an algebraic model, consisting of a graded algebra $H ^ { * } ( X , \mathbf{F} _ { p } ) = R ^ { * }$ over $\mathbf{F} _ { p }$ and an algebra $\mathcal{A} _ { p }$ of natural operations, called the Steenrod algebra. Each $f : X \rightarrow Y$ induces an element $f ^ { * } \in \text { Hom}_{\text{alg} } ( H ^ { * } ( Y , {\bf F} _ { p } ) , H ^ { * } ( X , {\bf F} _ { p } ) )$ that commutes with the action of $\mathcal{A} _ { p }$. $\mathcal{A} _ { p }$ is a connected graded Hopf algebra acting on the graded algebra $R ^ { * }$.

The hypothesis that $R ^ { * }$ is the cohomology of a space imposes an additional "unstable" condition. This is most simply stated if $p = 2$: ${\cal A} _ { 2 }$ is generated as an (non-commutative) algebra by the Steenrod operations $\{ \Sq^{i} : i \geq 0 \}$, with relations forced by its actions of the cohomology of all topological spaces. For example, $\Sq^{0} = \operatorname{Id}$ and $\Sq^{ 1 } = \beta$, the modulo-$2$ Bockstein operator. The unstable condition is then that $\Sq^{ i } x _ { n } = 0$ for $i > n$ and $\Sq^ { n } x _ { n } = x _ { n } ^ { 2 }$. The algebraic category $\mathcal{K}$ of unstable algebras $\{ \mathcal{R} ^ { * } \}$ over $\mathcal{A} _ { p }$ is thus an approximation to the homotopy category of topological spaces. The larger category $\mathcal U$ of unstable modules over $\mathcal{A} _ { p }$ has also proved useful.

For $p > 2$, the structure of $\mathcal{A} _ { p }$ and unstable actions are similar, but slightly more involved. However, in all cases, the set of relations in the Steenrod algebra and the unstable condition are derivable from the known action of $\mathcal{A} _ { p }$ on the cohomology of products of copies of $B {\bf Z} / p {\bf Z}$. In the following, explicit references to the coefficients are omitted.

The relationship of $\pi _0 \operatorname { Map } ( X , Y )$ to its model $\operatorname{Hom}_{ \mathcal{K} } ( H ^ { * } ( Y , \mathbf{F} _ { p } ) , H ^ { * } ( X , \mathbf{F} _ { p } ) )$ is of particular interest. The equivalence

\begin{equation*} \operatorname { Map } ( X \times Z , Y ) \rightarrow \operatorname { Map } ( X , \operatorname { Map } ( Z , Y ) ) \end{equation*}

raises the hope that in very favourable cases the mapping

\begin{equation*} \operatorname{Hom}_{\mathcal{K}} ( H ^ { * } \operatorname { Map } ( Z , Y ) , H ^ { * } X ) \rightarrow \end{equation*}

\begin{equation*} \rightarrow \operatorname{Hom}_{\mathcal{K}} ( H ^ { * } Y , H ^ { * } X \bigotimes H ^ { * } Z ) \end{equation*}

might be an isomorphism. That suggests that in the category $\mathcal{K}$, $H ^ { * } \operatorname { Map } ( Z , Y )$ should be approximated by the left adjoint functor to tensoring on the right by $H ^ { * } Z$. This motivated J. Lannes to define the functor $T$ as follows: If $E$ is a finite-dimensional $\mathbf{F} _ { p }$-vector space, then the $T$-functor $T _ { E } : \mathcal{U} \rightarrow \mathcal{U} $ is the left adjoint in $\mathcal U$ of the functor $( ( _- ) \otimes _ {{\bf F}_p } H ^ { * } B V ) :\cal U \rightarrow U$. In the topological case, there is a natural mapping

\begin{equation*} \lambda _ { X } : T _ { E } H ^ { * } X \rightarrow H ^ { * } \operatorname { Map } ( B E , X ). \end{equation*}

For general $Z$, the adjoint to $( ( \_ ) \otimes _ { \mathbf{F}_p } H ^ { * } Z )$ accounts for only part of the starting page of a Bousfield–Kan unstable Adams spectral sequence for $\operatorname{Map}( Z , Y )$. Lannes provides the basic connection to topology by blending the algebraic properties of $T _ { E }$ and $\mathcal{K}$ with the Bousfield–Kan spectral sequence: For many interesting spaces $X$,

\begin{equation*} H ^ { * } \operatorname { Map } ( B E , X ) \approx T _ { E } H ^ { * } X. \end{equation*}

In particular,

\begin{equation*} \pi _ { 0 } \operatorname { Map } ( B E , X ) = [ B E , X ] = \operatorname { Hom } _ { \mathcal{K} } ( H ^ { * } X , H ^ { * } B E ). \end{equation*}

For $f : X \rightarrow Y$, one has the path component $\operatorname{Map}( X , Y ) _ { f }$ of functions homotopic to $f$. The analogous $T$-construct is as follows: Each $\varphi \in \operatorname{Hom}_{\mathcal{K}}( R ^ { * } , H ^ { * } B E )$ induces a $T ^ { 0 } E$-module structure on $\mathbf{F} _ { p }$ and

\begin{equation*} T _ { E , \varphi } R ^ { * } = T _ { E } R ^ { * } \bigotimes _ { T ^ { 0 } E } \mathbf{F} _ { p }. \end{equation*}

The most striking features of $T _ { E }$ are summarized below (see also [a1]). To some extent, these were presaged by work of G. Carlsson and H.T. Miller, who established that the $\{ H ^ { * } B V \}$ are injectives in $\mathcal U$.

a) $T _ { E }$ is exact.

b) $T _ { E }$ respects tensor products, i.e $T _ { E } ( M \otimes _ { \mathbf{F}_ p} N) = T _ { E } M \otimes _ { \mathbf{F}_ p} T _ { E } N$.

c) $T _ { E }$ commutes with the $p$th power operations in a suitable sense.

d) $T _ { E }$ maps $\mathcal{K}$ to $\mathcal{K}$.

In principle, $T _ { E } M ^ { * }$ can be calculated by using the exactness property and a resolution of $M ^ { * }$ by free unstable $\mathcal{A} _ { p }$-modules. In practice, other methods are often more effective; for example,

1) If $M ^ { * }$ is finite, then $T _ { E } M ^ { * } = M ^ { * }$.

2) If $R ^ { * } = H ^ { * } B V$, then

\begin{equation*} T _ { E } R ^ { * } = \prod _ { \text { Hom}_{ \text{grp} } ( E , V ) } H ^ { * } B V, \end{equation*}

for $E$ and $V$ finite-dimensional $\mathbf{F} _ { p }$-vector spaces.

3) If $\tau : R ^ { * } \rightarrow H ^ { * } B E$ in $\mathcal{K}$ is an inclusion, then $T _ { E , \tau } R ^ { * }$ is the smallest sub-Hopf algebra of $H ^ { * } B E$ that contains $\tau ( R ^ { * } )$.

4) If $X$ is a finite $E$-complex with fixed point set $X ^ { E }$ and $H ^ { *_{E}} X$ is the modulo $p$ cohomology of the Borel construction, then $T_{E, \text{id}} H _ { E } ^ { * } X = H ^ { * } B E \otimes _ { \text{F}_ p } H ^ { * } X ^ { E }$ in $\mathcal{K}$.

5) If $G$ is a compact Lie group, then

$$ T_E H^\ast BG = \prod_{\varphi \in \operatorname{Hom}_{\text{grp}}(E,G)/G\text{-conj}} H^\ast BC_G(\operatorname{im} \varphi(E)) $$

These examples each have powerful topological consequences. For example, the first and fourth lead to new proofs of the Sullivan conjecture, originally proved by Miller and Carlsson. The last leads to a new view of the homotopy theory of classifying spaces. Most of the above is referenced in [a2].

References

[a1] J. Lannes, "Sur les espaces fonctionnels dont la source est le classifiant d'un $p$-groupe abélien élémentaire" Inst. Hautes Etudes Sci. Publ. Math. , 75 (1992) pp. 135–244 (Appendix by M. Zisman)
[a2] L. Schwartz, "Unstable modules over the Steenrod algebra and Sullivan's fixed point set conjecture" , Univ. Chicago Press (1994)
How to Cite This Entry:
Lannes-T-functor. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Lannes-T-functor&oldid=55519
This article was adapted from an original article by Clarence W. Wilkerson, Jr. (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article