Adams resolution

In mathematics, specifically algebraic topology, there is a resolution analogous to free resolutions of spectra yielding a tool for constructing the Adams spectral sequence. Essentially, the idea is to take a connective spectrum of finite type X {\displaystyle X} and iteratively resolve with other spectra that are in the homotopy kernel of a map resolving the cohomology classes in H ( X ; Z / p ) {\displaystyle H^{*}(X;\mathbb {Z} /p)} using Eilenberg–MacLane spectra.

This construction can be generalized using a spectrum E {\displaystyle E} , such as the Brown–Peterson spectrum B P {\displaystyle BP} , or the complex cobordism spectrum M U {\displaystyle MU} , and is used in the construction of the Adams–Novikov spectral sequence[1]pg 49.

Construction

The mod p {\displaystyle p} Adams resolution ( X s , g s ) {\displaystyle (X_{s},g_{s})} for a spectrum X {\displaystyle X} is a certain "chain-complex" of spectra induced from recursively looking at the fibers of maps into generalized Eilenberg–Maclane spectra giving generators for the cohomology of resolved spectra[1]pg 43. By this, we start by considering the map

X K {\displaystyle {\begin{matrix}X\\\downarrow \\K\end{matrix}}}

where K {\displaystyle K} is an Eilenberg–Maclane spectrum representing the generators of H ( X ) {\displaystyle H^{*}(X)} , so it is of the form

K = k = 1 I k Σ k H Z / p {\displaystyle K=\bigwedge _{k=1}^{\infty }\bigwedge _{I_{k}}\Sigma ^{k}H\mathbb {Z} /p}

where I k {\displaystyle I_{k}} indexes a basis of H k ( X ) {\displaystyle H^{k}(X)} , and the map comes from the properties of Eilenberg–Maclane spectra. Then, we can take the homotopy fiber of this map (which acts as a homotopy kernel) to get a space X 1 {\displaystyle X_{1}} . Note, we now set X 0 = X {\displaystyle X_{0}=X} and K 0 = K {\displaystyle K_{0}=K} . Then, we can form a commutative diagram

X 0 X 1 K 0 {\displaystyle {\begin{matrix}X_{0}&\leftarrow &X_{1}\\\downarrow &&\\K_{0}\end{matrix}}}

where the horizontal map is the fiber map. Recursively iterating through this construction yields a commutative diagram

X 0 X 1 X 2 K 0 K 1 K 2 {\displaystyle {\begin{matrix}X_{0}&\leftarrow &X_{1}&\leftarrow &X_{2}&\leftarrow \cdots \\\downarrow &&\downarrow &&\downarrow \\K_{0}&&K_{1}&&K_{2}\end{matrix}}}

giving the collection ( X s , g s ) {\displaystyle (X_{s},g_{s})} . This means

X s = Hofiber ( f s 1 : X s 1 K s 1 ) {\displaystyle X_{s}={\text{Hofiber}}(f_{s-1}:X_{s-1}\to K_{s-1})}

is the homotopy fiber of f s 1 {\displaystyle f_{s-1}} and g s : X s X s 1 {\displaystyle g_{s}:X_{s}\to X_{s-1}} comes from the universal properties of the homotopy fiber.

Resolution of cohomology of a spectrum

Now, we can use the Adams resolution to construct a free A p {\displaystyle {\mathcal {A}}_{p}} -resolution of the cohomology H ( X ) {\displaystyle H^{*}(X)} of a spectrum X {\displaystyle X} . From the Adams resolution, there are short exact sequences

0 H ( X s ) H ( K s ) H ( Σ X s + 1 ) 0 {\displaystyle 0\leftarrow H^{*}(X_{s})\leftarrow H^{*}(K_{s})\leftarrow H^{*}(\Sigma X_{s+1})\leftarrow 0}

which can be strung together to form a long exact sequence

0 H ( X ) H ( K 0 ) H ( Σ K 1 ) H ( Σ 2 K 2 ) {\displaystyle 0\leftarrow H^{*}(X)\leftarrow H^{*}(K_{0})\leftarrow H^{*}(\Sigma K_{1})\leftarrow H^{*}(\Sigma ^{2}K_{2})\leftarrow \cdots }

giving a free resolution of H ( X ) {\displaystyle H^{*}(X)} as an A p {\displaystyle {\mathcal {A}}_{p}} -module.

E*-Adams resolution

Because there are technical difficulties with studying the cohomology ring E ( E ) {\displaystyle E^{*}(E)} in general[2]pg 280, we restrict to the case of considering the homology coalgebra E ( E ) {\displaystyle E_{*}(E)} (of co-operations). Note for the case E = H F p {\displaystyle E=H\mathbb {F} _{p}} , H F p ( H F p ) = A {\displaystyle H\mathbb {F} _{p*}(H\mathbb {F} _{p})={\mathcal {A}}_{*}} is the dual Steenrod algebra. Since E ( X ) {\displaystyle E_{*}(X)} is an E ( E ) {\displaystyle E_{*}(E)} -comodule, we can form the bigraded group

Ext E ( E ) ( E ( S ) , E ( X ) ) {\displaystyle {\text{Ext}}_{E_{*}(E)}(E_{*}(\mathbb {S} ),E_{*}(X))}

which contains the E 2 {\displaystyle E_{2}} -page of the Adams–Novikov spectral sequence for X {\displaystyle X} satisfying a list of technical conditions[1]pg 50. To get this page, we must construct the E {\displaystyle E_{*}} -Adams resolution[1]pg 49, which is somewhat analogous to the cohomological resolution above. We say a diagram of the form

X 0 g 0 X 1 g 1 X 2 K 0 K 1 K 2 {\displaystyle {\begin{matrix}X_{0}&\xleftarrow {g_{0}} &X_{1}&\xleftarrow {g_{1}} &X_{2}&\leftarrow \cdots \\\downarrow &&\downarrow &&\downarrow \\K_{0}&&K_{1}&&K_{2}\end{matrix}}}

where the vertical arrows f s : X s K s {\displaystyle f_{s}:X_{s}\to K_{s}} is an E {\displaystyle E_{*}} -Adams resolution if

  1. X s + 1 = Hofiber ( f s ) {\displaystyle X_{s+1}={\text{Hofiber}}(f_{s})} is the homotopy fiber of f s {\displaystyle f_{s}}
  2. E X s {\displaystyle E\wedge X_{s}} is a retract of E K s {\displaystyle E\wedge K_{s}} , hence E ( f s ) {\displaystyle E_{*}(f_{s})} is a monomorphism. By retract, we mean there is a map h s : E K s E X s {\displaystyle h_{s}:E\wedge K_{s}\to E\wedge X_{s}} such that h s ( E f s ) = i d E X s {\displaystyle h_{s}(E\wedge f_{s})=id_{E\wedge X_{s}}}
  3. K s {\displaystyle K_{s}} is a retract of E K s {\displaystyle E\wedge K_{s}}
  4. Ext t , u ( E ( S ) , E ( K s ) ) = π u ( K s ) {\displaystyle {\text{Ext}}^{t,u}(E_{*}(\mathbb {S} ),E_{*}(K_{s}))=\pi _{u}(K_{s})} if t = 0 {\displaystyle t=0} , otherwise it is 0 {\displaystyle 0}

Although this seems like a long laundry list of properties, they are very important in the construction of the spectral sequence. In addition, the retract properties affect the structure of construction of the E {\displaystyle E_{*}} -Adams resolution since we no longer need to take a wedge sum of spectra for every generator.

Construction for ring spectra

The construction of the E {\displaystyle E_{*}} -Adams resolution is rather simple to state in comparison to the previous resolution for any associative, commutative, connective ring spectrum E {\displaystyle E} satisfying some additional hypotheses. These include E ( E ) {\displaystyle E_{*}(E)} being flat over π ( E ) {\displaystyle \pi _{*}(E)} , μ {\displaystyle \mu _{*}} on π 0 {\displaystyle \pi _{0}} being an isomorphism, and H r ( E ; A ) {\displaystyle H_{r}(E;A)} with Z A Q {\displaystyle \mathbb {Z} \subset A\subset \mathbb {Q} } being finitely generated for which the unique ring map

θ : Z π 0 ( E ) {\displaystyle \theta :\mathbb {Z} \to \pi _{0}(E)}

extends maximally. If we set

K s = E F s {\displaystyle K_{s}=E\wedge F_{s}}

and let

f s : X s K s {\displaystyle f_{s}:X_{s}\to K_{s}}

be the canonical map, we can set

X s + 1 = Hofiber ( f s ) {\displaystyle X_{s+1}={\text{Hofiber}}(f_{s})}

Note that E {\displaystyle E} is a retract of E E {\displaystyle E\wedge E} from its ring spectrum structure, hence E X s {\displaystyle E\wedge X_{s}} is a retract of E K s = E E X s {\displaystyle E\wedge K_{s}=E\wedge E\wedge X_{s}} , and similarly, K s {\displaystyle K_{s}} is a retract of E K s {\displaystyle E\wedge K_{s}} . In addition

E ( K s ) = E ( E ) π ( E ) E ( X s ) {\displaystyle E_{*}(K_{s})=E_{*}(E)\otimes _{\pi _{*}(E)}E_{*}(X_{s})}

which gives the desired Ext {\displaystyle {\text{Ext}}} terms from the flatness.

Relation to cobar complex

It turns out the E 1 {\displaystyle E_{1}} -term of the associated Adams–Novikov spectral sequence is then cobar complex C ( E ( X ) ) {\displaystyle C^{*}(E_{*}(X))} .

See also

References

  1. ^ a b c d Ravenel, Douglas C. (1986). Complex cobordism and stable homotopy groups of spheres. Orlando: Academic Press. ISBN 978-0-08-087440-1. OCLC 316566772.
  2. ^ Adams, J. Frank (John Frank) (1974). Stable homotopy and generalised homology. Chicago: University of Chicago Press. ISBN 0-226-00523-2. OCLC 1083550.