This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

\proceedingstitle

Proceedings of the Nishida Fest (Kinosaki 2003) \conferencestart28 July 2003 \conferenceend8 August 2003 \conferencenameInternational Conference in Homotopy Theory \conferencelocationKinosaki, Japan \editorMatthew Ando \givennameMatthew \surnameAndo \editorNorihiko Minami \givennameNorihiko \surnameMinami \editorJack Morava \givennameJack \surnameMorava \editorW Stephen Wilson \givennameW Stephen \surnameWilson \givennameMitsunori \surnameImaoka \urladdr \volumenumber10 \issuenumber \publicationyear2007 \papernumber10 \startpage187 \endpage193 \MR \Zbl \arxivreference \subjectprimarymsc200055R12 \subjectsecondarymsc200055Q45 \published29 January 2007 \publishedonline29 January 2007 \proposed \seconded \corresponding \version

Hypersurface representation and the image of
the double S3S^{3}–transfer

Mitsunori Imaoka Department of Mathematics Education
Graduate School of Education
Hiroshima University

1–1–1 Kagamiyama
Higashi–Hiroshima 739-8524
Japan
imaoka@hiroshima-u.ac.jp
(24 August 2004; 5 July 2005)
Abstract

We study the image of a transfer homomorphism in the stable homotopy groups of spheres. Actually, we show that an element of order 8 in the 18 dimensional stable stem is in the image of a double transfer homomorphism, which reproves a result due to P J Eccles that the element is represented by a framed hypersurface. Also, we determine the image of the transfer homomorphism in the 16 dimensional stable stem.

keywords:
transfer map
keywords:
stable homotopy
keywords:
hypersurface

1 Introduction and result

Let νπ18s(S0)\nu^{*}\in\pi^{s}_{18}(S^{0}) be an element of order 88 in the stable homotopy groups of spheres. Throughout the paper, πi(X)\pi_{i}(X) (resp. πis(X)\pi^{s}_{i}(X)) denotes the homotopy group (resp. the stable homotopy group) of a space XX, and we use the notations of Toda [10] for elements of πs(S0)\pi^{s}_{*}(S^{0}). Then, using a generators νπ3s(S0)\Z/24\nu\in\pi^{s}_{3}(S^{0})\cong\Z/24 and σπ7s(S0)\Z/240\sigma\in\pi^{s}_{7}(S^{0})\cong\Z/240, ν\nu^{*} is represented by the Toda bracket σ,2σ,ν=σ,ν,σ\langle\sigma,2\sigma,\nu\rangle=-\langle\sigma,\nu,\sigma\rangle with no indeterminacy.

As is known, ν\nu^{*} is not in the image of the homomorphism J0\coπ18(SO)π18s(S0)J_{0}\co\pi_{18}(SO)\to\pi^{s}_{18}(S^{0}) induced by the stable JJ–map J\coSOΩΣS0J\co SO\to\Omega^{\infty}\Sigma^{\infty}S^{0}. We shall show that ν\nu^{*} is in the image of the homomorphism J1\coπ19(ΣSO)π18s(S0)J_{1}\co\pi_{19}(\Sigma SO)\to\pi^{s}_{18}(S^{0}) induced by the adjoint map ΣSOΩΣS1\Sigma SO\to\Omega^{\infty}\Sigma^{\infty}S^{1} to JJ. Actually, we prove that ν\nu^{*} is in a double S3S^{3}–transfer image which is a subgroup of Im(J1)\mbox{Im}(J_{1}).

Eccles [3] has made clear that Im(J1)\mbox{Im}(J_{1}) consists of elements represented by framed hypersurfaces. Such study has also done by Rees [9]. Our result gives the following.

Theorem (Eccles [3, page 168]).

ν\nu^{*} is representable by a framed hypersurface of dimension 1818.

Here, a framed hypersurface of dimension 1818 is a closed manifold of dimension 18 embedded in S19S^{19} and framed in SNS^{N} for sufficiently large NN, and the theorem means that ν\nu^{*} is the framed cobordism class of such a framed hypersurface when we regard π18s(S0)\pi^{s}_{18}(S^{0}) as the framed cobordism group Ω18fr\Omega^{fr}_{18}.

The Bott map B\coΣ3BSpSOB\co\Sigma^{3}BSp\to SO is the adjoint map to the homotopy equivalence BSp×\ZΩ3SOBSp\times\Z\to\Omega^{3}SO, where SO=nSO(n)SO=\bigcup_{n}SO(n) is the rotation group and BSpBSp is the classifying space of the symplectic group Sp=nSp(n)Sp=\bigcup_{n}Sp(n). Let P=nPn\mathbb{H}P^{\infty}=\bigcup_{n}\mathbb{H}P^{n} be the infinite dimensional quaternionic projective space. Then, P=BSp(1)\mathbb{H}P^{\infty}=BSp(1), and there is an inclusion map i\coPBSpi\co\mathbb{H}P^{\infty}\to BSp. We define a map tHt_{H} as the composition

tH\coΣ3PiΣ3BSpBSOJΩΣS0.t_{H}\co\Sigma^{3}\mathbb{H}P^{\infty}\stackrel{{\scriptstyle i}}{{\longrightarrow}}\Sigma^{3}BSp\stackrel{{\scriptstyle B}}{{\longrightarrow}}SO\stackrel{{\scriptstyle J}}{{\longrightarrow}}\Omega^{\infty}\Sigma^{\infty}S^{0}.

We denote by tHs\coPS3t_{H}^{s}\co\mathbb{H}P^{\infty}\to S^{-3} a stable map adjoint to tHt_{H}. It is not certain whether tHst_{H}^{s} is stably homotopic to the stable map constructed by the Boardman’s transfer construction [1] on the principal S3S^{3}–bundle over P\mathbb{H}P^{\infty}. However, since tHt_{H} has the following property in common with the Boardman’s S3S^{3}–transfer map, we call tHt_{H} and tHst_{H}^{s} the S3S^{3}–transfer maps.

Lemma 1.1.

The restriction tH\coS7ΩΣS0t_{H}^{\prime}\co S^{7}\to\Omega^{\infty}\Sigma^{\infty}S^{0} of tHt_{H} to the bottom sphere represents σπ7s(S0)\sigma\in\pi^{s}_{7}(S^{0}) up to sign.

Let μ\coΩΣS0ΩΣS0ΩΣS0\mu\co\Omega^{\infty}\Sigma^{\infty}S^{0}\wedge\Omega^{\infty}\Sigma^{\infty}S^{0}\to\Omega^{\infty}\Sigma^{\infty}S^{0} be the multiplication on the infinite loop space ΩΣS0\Omega^{\infty}\Sigma^{\infty}S^{0} given by composition, and μ\coΣΩΣS0ΩΣS0ΩΣS1\mu^{\prime}\co\Sigma\Omega^{\infty}\Sigma^{\infty}S^{0}\wedge\Omega^{\infty}\Sigma^{\infty}S^{0}\to\Omega^{\infty}\Sigma^{\infty}S^{1} the adjoint map to μ\mu. Then, we set

tH(2)=μΣ(tHtH)\coΣ7PPΩΣS1,t_{H}(2)=\mu^{\prime}\circ\Sigma(t_{H}\wedge t_{H})\co\Sigma^{7}\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty}\to\Omega^{\infty}\Sigma^{\infty}S^{1},

and call tH(2)t_{H}(2) a double S3S^{3}–transfer map. Then, our main result may be stated as follows:

Theorem 2.

There exists an element aπ12(PP)a\in\pi_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty}) satisfying

tH(2)(Σ7a)=νinπ18s(S0).t_{H}(2)_{*}(\Sigma^{7}a)=\nu^{*}\ \ \ in\ \ \pi_{18}^{s}(S^{0}).

Applying a result due to Eccles and Walker [4], we have Im(tH(2))Im(J1)\text{\rm Im}(t_{H}(2)_{*})\subset\text{\rm Im}(J_{1}) (see \fullrefProp2 and Corollary2.2). Hence, it turns out that ν\nu^{*} is represented by a framed hypersurface of dimension 1818 as is stated in the first theorem.

In Carlisle et al [2], effective use of the S1S^{1}–transfer homomorphism τ\coπi1s(\CP)πis(S0)\tau\co\pi^{s}_{i-1}(\C P^{\infty})\to\pi^{s}_{i}(S^{0}) shows that certain elements are represented by framed hypersurfaces. In this respect, ν\nu^{*} is not in the image of the double S1S^{1}–transfer homomorphism τ2\coπ16s(\CP\CP)π18s(S0)\tau_{2}\co\pi^{s}_{16}(\C P^{\infty}\wedge\C P^{\infty})\to\pi^{s}_{18}(S^{0}). In fact, the image of τ2\tau_{2} in π18s(S0)\pi^{s}_{18}(S^{0}) is equal to \Z/4{2ν}\Z/4\{2\nu^{*}\} by Imaoka [5, Theorem 10].

By Toda [10], π16s(S0)=\Z/2{η}\Z/2{ηρ}\pi_{16}^{s}(S^{0})=\Z/2\{\eta^{*}\}\oplus\Z/2\{\eta\circ\rho\}. Here, ηπ1s(S0)\Z/2\eta\in\pi^{s}_{1}(S^{0})\cong\Z/2 is a generator, η\eta^{*} is an element in the Toda bracket σ,2σ,η\langle\sigma,2\sigma,\eta\rangle and ρ\rho is a generator of the image of J0\coπ15(SO)π15s(S0)J_{0}\co\pi_{15}(SO)\to\pi_{15}^{s}(S^{0}). Mahowald [6] has constructed an important family ηiπ2is(S0)\eta_{i}\in\pi_{2^{i}}^{s}(S^{0}) for i3i\geq 3, and ηη4(modηρ)\eta^{*}\equiv\eta_{4}\pmod{\eta\circ\rho}.

By \fullrefLem1, we see the following:

Lemma 1.3.

Im(tH(2))=0{\rm Im}(t_{H}(2)_{*})=0 in π16s(S0)\pi_{16}^{s}(S^{0}).

Mukai [8, Theorem 2] has shown that both ν\nu^{*} and η\eta^{*} are in the image of the S1S^{1}–transfer homomorphism τ\coπi1s(\CP)πis(S0)\tau\co\pi_{i-1}^{s}(\C P^{\infty})\to\pi_{i}^{s}(S^{0}). Morisugi [7, Corollary E] has shown that all Mahowald’s elements are in the image of the S3S^{3}–transfer homomorphism πs(Q)πs(S0)\pi_{*}^{s}(Q_{\infty})\to\pi_{*}^{s}(S^{0}) given for the quaternionic quasi-projective space Q=nQnQ_{\infty}=\bigcup_{n}Q_{n}.

In contrast with these, we have the following:

Theorem 4.

Im((tHs)\coπ13s(P)π16s(S0))=\Z/2{ηρ}{\rm Im}((t_{H}^{s})_{*}\co\pi_{13}^{s}(\mathbb{H}P^{\infty})\to\pi_{16}^{s}(S^{0}))=\Z/2\{\eta\circ\rho\}, and thus ηIm(tHs)\eta^{*}\not\in\text{\rm Im}(t_{H}^{s})_{*}.

We also remark the following noted by Eccles [3], where J2\coπ18(Σ2SO)π16s(S0)J_{2}\co\pi_{18}(\Sigma^{2}SO)\to\pi_{16}^{s}(S^{0}) is the homomorphism induced by the map Σ2SOΩΣS2\Sigma^{2}SO\to\Omega^{\infty}\Sigma^{\infty}S^{2} adjoint to the stable JJ–map.

Proposition 1.5.

ηIm(J2)\eta^{*}\in{\rm Im}(J_{2}).

In \fullrefsec2 we prove \fullrefLem1 and \fullrefThA, and in \fullrefsec3 we prove \fullreftH20, \fullrefPropB and \fullrefeta*2.

2 Proof of \fullrefThA

We first prove \fullrefLem1.

Proof of \fullrefLem1.

Let ξ\xi be the canonical left \mathbb{H}–line bundle over P\mathbb{H}P^{\infty}, and put \tildeξ=ξ1\tilde{\xi}=\xi-1_{\mathbb{H}}. Also, let \tildeξ1\tilde{\xi_{1}}^{*} be the adjoint to the restriction of \tildeξ\tilde{\xi} to P1=S4\mathbb{H}P^{1}=S^{4}. Then, the restriction of the classifying map φ\varphi of \tildeξ1\tildeξ\tilde{\xi}_{1}^{*}\otimes_{\mathbb{H}}\tilde{\xi} to the bottom sphere S8=S4S4S^{8}=S^{4}\wedge S^{4} represents a generator of π8(BSO)\Z\pi_{8}(BSO)\cong\Z. As is known, the adjoint map Σ4PBSO\Sigma^{4}\mathbb{H}P^{\infty}\to BSO to the composition Bi\coΣ3PSOB\circ i\co\Sigma^{3}\mathbb{H}P^{\infty}\to SO is homotopic to φ\varphi. Since σ=J0(ι7)\sigma=J_{0}(\iota_{7}) for a generator ι7π7(SO)\iota_{7}\in\pi_{7}(SO), tHt_{H}^{\prime} represents σ\sigma up to sign. ∎

Let H(m)\coΣSOSOΣSOH(m)\co\Sigma SO\wedge SO\to\Sigma SO be the map defined by the Hopf construction on the multiplication m\coSO×SOSOm\co SO\times SO\to SO, and J\coΣSOΩΣS1J^{*}\co\Sigma SO\to\Omega^{\infty}\Sigma^{\infty}S^{1} the adjoint map to the stable JJ–map J\coSOΩΣS0J\co SO\to\Omega^{\infty}\Sigma^{\infty}S^{0}. Then, Eccles and Walker have shown the following [4, Proposition 2.2], in which μ\coΣ(ΩΣS0ΩΣS0)ΩΣS1\mu^{\prime}\co\Sigma(\Omega^{\infty}\Sigma^{\infty}S^{0}\wedge\Omega^{\infty}\Sigma^{\infty}S^{0})\to\Omega^{\infty}\Sigma^{\infty}S^{1} is the adjoint map to the multiplication μ\coΩΣS0ΩΣS0ΩΣS0\mu\co\Omega^{\infty}\Sigma^{\infty}S^{0}\wedge\Omega^{\infty}\Sigma^{\infty}S^{0}\to\Omega^{\infty}\Sigma^{\infty}S^{0}.

Proposition 2.1.

The composition JH(m)\coΣSOSOΩΣS1J^{*}\circ H(m)\co\Sigma SO\wedge SO\to\Omega^{\infty}\Sigma^{\infty}S^{1} is homotopic to μΣ(JJ)\mu^{\prime}\circ\Sigma(J\wedge J).

Recall that the homomorphism J1\coπi(ΣSO)πi1s(S0)J_{1}\co\pi_{i}(\Sigma SO)\to\pi^{s}_{i-1}(S^{0}) is induced by JJ^{*}. Since (JJ)(BiBi)(J\wedge J)\circ(B\circ i\wedge B\circ i) is equal to tHtHt_{H}\wedge t_{H}, by composing it with the maps of \fullrefProp2, JH(m)Σ(BiBi)\coΣ(Σ3PΣ3P)ΩΣS1J^{*}\circ H(m)\circ\Sigma(B\circ i\wedge B\circ i)\co\Sigma(\Sigma^{3}\mathbb{H}P^{\infty}\wedge\Sigma^{3}\mathbb{H}P^{\infty})\to\Omega^{\infty}\Sigma^{\infty}S^{1} is homotopic to tH(2)=μΣ(tHtH)t_{H}(2)=\mu^{\prime}\circ\Sigma(t_{H}\wedge t_{H}). Hence, we have the following.

Corollary 2.2.

Im(tH(2))Im(J1)\text{\rm Im}(t_{H}(2)_{*})\subset\text{\rm Im}(J_{1}) in πis(S0)\pi_{i}^{s}(S^{0}).

Let βiH4i(P;\Z)\Z\beta_{i}\in H_{4i}(\mathbb{H}P^{\infty};\thinspace\Z)\cong\Z be a generator. Then, we have H12(PP;\Z)=\Z{β1β2}\Z{β2β1}H_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty};\thinspace\Z)=\Z\{\beta_{1}\otimes\beta_{2}\}\oplus\Z\{\beta_{2}\otimes\beta_{1}\}.

Lemma 2.3.

There exists an element aπ12(PP)a\in\pi_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty}) satisfying h(a)=β1β2β2β1h(a)=\beta_{1}\otimes\beta_{2}-\beta_{2}\otimes\beta_{1} for the Hurewicz homomorphism hh.

Proof.

Notice that π12(PP)π12s(PP)\pi_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})\cong\pi^{s}_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty}). Since P2=S4νe8\mathbb{H}P^{2}=S^{4}\cup_{\nu}e^{8}, the 1212–skeleton of PP\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty} has a cell structure S8νν(e12e12)S^{8}\cup_{\nu\vee\nu}(e^{12}\cup e^{12}). Thus, we have the following exact sequence:

π12s(S8)=0π12s(PP)ψπ12s(S12S12)φπ11s(S8).\pi^{s}_{12}(S^{8})=0\to\pi_{12}^{s}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})\stackrel{{\scriptstyle\psi}}{{\longrightarrow}}\pi^{s}_{12}(S^{12}\vee S^{12})\stackrel{{\scriptstyle\varphi}}{{\longrightarrow}}\pi^{s}_{11}(S^{8}).

Here, φ\coπ12s(S12S12)=\Z{ι1}\Z{ι2}π11s(S8)=\Z/24{ν}\varphi\co\pi^{s}_{12}(S^{12}\vee S^{12})=\Z\{\iota_{1}\}\oplus\Z\{\iota_{2}\}\to\pi^{s}_{11}(S^{8})=\Z/24\{\nu\} satisfies φ(ι1)=φ(ι2)=ν\varphi(\iota_{1})=\varphi(\iota_{2})=\nu. Therefore, π12s(PP)=\Z{a}\Z{b}\pi^{s}_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})=\Z\{a\}\oplus\Z\{b\} for some elements aa and bb satisfying ψ(a)=ι1ι2\psi_{*}(a)=\iota_{1}-\iota_{2} and ψ(b)=24ι2\psi_{*}(b)=24\iota_{2}. Then, aa is a required element up to sign, since we have {ψ(β1β2),ψ(β2β1)}={h(ι1),h(ι2)}\{\psi_{*}(\beta_{1}\otimes\beta_{2}),\psi_{*}(\beta_{2}\otimes\beta_{1})\}=\{h(\iota_{1}),h(\iota_{2})\} up to sign. ∎

Proof of \fullrefThA.

Let aπ12(PP)a\in\pi_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty}) be the element in \fullrefLem4. Since π12(PP)π12s(P2P2)\pi_{12}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})\cong\pi_{12}^{s}(\mathbb{H}P^{2}\wedge\mathbb{H}P^{2}) and tH(2)=(tHstHs)Σt_{H}(2)_{*}=(t_{H}^{s}\wedge t_{H}^{s})_{*}\Sigma^{\infty} for the suspension isomorphism Σ\coπ19(Σ7P2P2)π12s(P2P2)\Sigma^{\infty}\co\pi_{19}(\Sigma^{7}\mathbb{H}P^{2}\wedge\mathbb{H}P^{2})\to\pi^{s}_{12}(\mathbb{H}P^{2}\wedge\mathbb{H}P^{2}), we identify aa with Σaπ12s(P2P2)\Sigma^{\infty}a\in\pi^{s}_{12}(\mathbb{H}P^{2}\wedge\mathbb{H}P^{2}). Then, it is sufficient to show (tHstHs)(a)=ν(t_{H}^{s}\wedge t_{H}^{s})_{*}(a)=\nu^{*} up to sign.

The following diagram is stably homotopy commutative up to sign:

S12aP2P2tHs1S3P21tHsS61qqS4S8i1P2S8tHs1S3S8,\begin{CD}S^{12}@>{a}>{}>\mathbb{H}P^{2}\wedge\mathbb{H}P^{2}@>{t_{H}^{s}\wedge 1}>{}>S^{-3}\wedge\mathbb{H}P^{2}@>{1\wedge t_{H}^{s}}>{}>S^{-6}\\ @V{}V{\simeq}V@V{}V{1\wedge q}V@V{}V{q}V\\ S^{4}\wedge S^{8}@>{i\wedge 1}>{}>\mathbb{H}P^{2}\wedge S^{8}@>{t_{H}^{s}\wedge 1}>{}>S^{-3}\wedge S^{8},\end{CD}

where qq is the collapsing map to the top cell. The left hand square is stably homotopy commutative because (1q)(a)(1\wedge q)_{*}(a) is a generator of π4s(P2)\pi^{s}_{4}(\mathbb{H}P^{2}) by \fullrefLem4. Hence, q(tHs1)a(tHsi)1σq\circ(t_{H}^{s}\wedge 1)\circ a\simeq(t_{H}^{s}\circ i)\wedge 1\simeq\sigma by \fullrefLem1, and thus (tHs1)a(t_{H}^{s}\wedge 1)\circ a is a coextension of σ\sigma. On the other hand, since tHs\coP2S3t_{H}^{s}\co\mathbb{H}P^{2}\to S^{-3} is an extension of σ\coS4S3\sigma\co S^{4}\to S^{-3} by \fullrefLem1,

(tHstHs)(a)=(1tHs)(tHs1)(a)σ,ν,σ={ν}(t_{H}^{s}\wedge t_{H}^{s})_{*}(a)=(1\wedge t_{H}^{s})_{*}(t_{H}^{s}\wedge 1)_{*}(a)\in\langle\sigma,\nu,\sigma\rangle=\{-\nu^{*}\}

by definition of the Toda bracket. Thus, we have completed the proof. ∎

3 Transfer image in π16s(S0)\pi^{s}_{16}(S^{0})

We remark that

π17(Σ7PP)π10s(PP)π10s(S8)=\Z/2{η2}.\pi_{17}(\Sigma^{7}\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})\cong\pi^{s}_{10}(\mathbb{H}P^{\infty}\wedge\mathbb{H}P^{\infty})\cong\pi^{s}_{10}(S^{8})=\Z/2\{\eta^{2}\}.

Then, using \fullrefLem1,

tH(2)(η2)=((tHs)(η))2=(ση)2=0,t_{H}(2)_{*}(\eta^{2})=((t_{H}^{s})_{*}(\eta))^{2}=(\sigma\circ\eta)^{2}=0,

and thus we have \fullreftH20.

Lastly, we prove \fullrefPropB and \fullrefeta*2.

Proof of \fullrefPropB.

As mentioned in \fullrefsec1, π16s(S0)=\Z/2{η}\Z/2{ηρ}\pi_{16}^{s}(S^{0})=\Z/2\{\eta^{*}\}\oplus\Z/2\{\eta\circ\rho\} by Toda [10], and we shall show

Im((tHs)\coπ13s(P)π16s(S0))=\Z/2{ηρ}.\text{\rm Im}((t_{H}^{s})_{*}\co\pi_{13}^{s}(\mathbb{H}P^{\infty})\to\pi_{16}^{s}(S^{0}))=\Z/2\{\eta\circ\rho\}.

Notice that π13s(P)π13s(P3)\pi_{13}^{s}(\mathbb{H}P^{\infty})\cong\pi_{13}^{s}(\mathbb{H}P^{3}) and the attaching map φ\coS11P2\varphi\co S^{11}\to\mathbb{H}P^{2} of the top cell in P3\mathbb{H}P^{3} is stably a coextension of 2νπ3s(S0)2\nu\in\pi_{3}^{s}(S^{0}). We consider the following exact sequence:

π14s(S12)φπ13s(P2)iπ13s(P3)qπ13s(S12)φπ12s(P2),\pi_{14}^{s}(S^{12})\stackrel{{\scriptstyle\varphi_{*}}}{{\longrightarrow}}\pi_{13}^{s}(\mathbb{H}P^{2})\stackrel{{\scriptstyle i_{*}}}{{\longrightarrow}}\pi_{13}^{s}(\mathbb{H}P^{3})\stackrel{{\scriptstyle q_{*}}}{{\longrightarrow}}\pi_{13}^{s}(S^{12})\stackrel{{\scriptstyle\varphi_{*}}}{{\longrightarrow}}\pi_{12}^{s}(\mathbb{H}P^{2}),

where ii is the inclusion P2P3\mathbb{H}P^{2}\subset\mathbb{H}P^{3} and qq is the collapsimg map P3S12\mathbb{H}P^{3}\to S^{12} to the top cell.

Since P2S4νe8\mathbb{H}P^{2}\approx S^{4}\cup_{\nu}e^{8}, we have

π12s(P2)\displaystyle\pi_{12}^{s}(\mathbb{H}P^{2}) =\Z/2{(i0)(ν¯)}\Z/2{(i0)(ϵ)}\displaystyle=\Z/2\{(i_{0})_{*}(\overline{\nu})\}\oplus\Z/2\{(i_{0})_{*}(\epsilon)\}
andπ13s(P2)\displaystyle\text{and}\quad\pi_{13}^{s}(\mathbb{H}P^{2}) =\Z/2{(i0)(μ)}\Z/2{(i0)(ηϵ)}\displaystyle=\Z/2\{(i_{0})_{*}(\mu)\}\oplus\Z/2\{(i_{0})_{*}(\eta\circ\epsilon)\}

for the bottom inclusion i0\coS4P2i_{0}\co S^{4}\to\mathbb{H}P^{2}, where

π8s(S0)=\Z/2{ν¯}\Z/2{ϵ}andπ9s(S0)=\Z/2{μ}\Z/2{ηϵ}\Z/2{ν3}.\pi_{8}^{s}(S^{0})=\Z/2\{\overline{\nu}\}\oplus\Z/2\{\epsilon\}\quad\text{and}\quad\pi_{9}^{s}(S^{0})=\Z/2\{\mu\}\oplus\Z/2\{\eta\circ\epsilon\}\oplus\Z/2\{\nu^{3}\}.

Since φ\varphi is a coextension of 2ν2\nu, we have φ(η)=φη(i0)(ν,2ν,η)(i0)(ϵ)(modi(ν¯))\varphi_{*}(\eta)=\varphi\circ\eta\in(i_{0})_{*}(\langle\nu,2\nu,\eta\rangle)\equiv(i_{0})_{*}(\epsilon)\pmod{i_{*}(\overline{\nu})}. Thus, φ\coπ13s(S12)π12s(P2)\varphi_{*}\co\pi_{13}^{s}(S^{12})\to\pi_{12}^{s}(\mathbb{H}P^{2}) is a monomorphism. Similarly, φ(η2)=φη2\varphi_{*}(\eta^{2})=\varphi\circ\eta^{2} is contained in (i0)(ν,2ν,η2)(i0)(ηϵ)(mod(i0)(ν3))(i_{0})_{*}(\langle\nu,2\nu,\eta^{2}\rangle)\equiv(i_{0})_{*}(\eta\circ\epsilon)\pmod{(i_{0})_{*}(\nu^{3})}. Hence, we have π13s(P)π13s(P3)=\Z/2{(i0)(μ)}\pi_{13}^{s}(\mathbb{H}P^{\infty})\cong\pi_{13}^{s}(\mathbb{H}P^{3})=\Z/2\{(i_{0})_{*}(\mu)\}.

Since tHs\coΣ3PS0t_{H}^{s}\co\Sigma^{3}\mathbb{H}P^{\infty}\to S^{0} is an extension of σ\coS7S0\sigma\co S^{7}\to S^{0} by \fullrefLem1, we conclude that Im((tHs)\coπ13s(P)π16s(S0))=\Z/2{σμ}=\Z/2{ηρ}((t_{H}^{s})_{*}\co\pi_{13}^{s}(\mathbb{H}P^{\infty})\to\pi_{16}^{s}(S^{0}))=\Z/2\{\sigma\circ\mu\}=\Z/2\{\eta\circ\rho\}. ∎

Proof of \fullrefeta*2.

As mentioned in \fullrefsec1,

ησ,2σ,ηπ16s(S0),\eta^{*}\in\langle\sigma,2\sigma,\eta\rangle\subset\pi_{16}^{s}(S^{0}),

and the indeterminacy of σ,2σ,η\langle\sigma,2\sigma,\eta\rangle is \Z/2{ηρ}\Z/2\{\eta\circ\rho\}.

Recall that σ=J0(ι7)π7s(S0)\sigma=J_{0}(\iota_{7})\in\pi_{7}^{s}(S^{0}) up to sign for a generator ι7π7(SO)\Z\iota_{7}\in\pi_{7}(SO)\cong\Z. By Toda [10, Proposition 5.15, (5.16), Proposition 5.1], there exist elements σπ14(S7)\sigma^{\prime}\in\pi_{14}(S^{7}) and η16π17(S16)\eta_{16}\in\pi_{17}(S^{16}) which suspend to 2σ2\sigma and η\eta, respectively. Then, ι7σ=0\iota_{7}\circ\sigma^{\prime}=0 since π14(SO)=0\pi_{14}(SO)=0. Also, by [10, Theorem 7.1], (Σ2σ)η16=0(\Sigma^{2}\sigma^{\prime})\circ\eta_{16}=0. Hence, an unstable Toda bracket {Σ2ι7,Σ2σ,η16}\{\Sigma^{2}\iota_{7},\Sigma^{2}\sigma^{\prime},\eta_{16}\} is defined in π18(Σ2SO)\pi_{18}(\Sigma^{2}SO), and it satisfies

J2({Σ2ι7,Σ2σ,η16})σ,2σ,ηϕ.J_{2}(\{\Sigma^{2}\iota_{7},\Sigma^{2}\sigma^{\prime},\eta_{16}\})\cap\langle\sigma,2\sigma,\eta\rangle\neq\phi.

Hence, η\eta^{*} or η+ηρ\eta^{*}+\eta\circ\rho is in Im(J2){\rm Im}(J_{2}). But, since ηρIm(J0)\eta\circ\rho\in{\rm Im}(J_{0}) and Im(J0)Im(J2){\rm Im}(J_{0})\subset{\rm Im}(J_{2}), we have the required result. ∎

References

  • [1] J M Boardman, Stable homotopy theory, mimeographed notes, University of Warwick (1966)
  • [2] D Carlisle, P Eccles, S Hilditch, N Ray, L Schwartz, G Walker, R Wood, Modular representations of GL(n,p)\mathrm{GL}(n,p), splitting Σ(P××P)\Sigma(\mathbb{C}P^{\infty}\times\cdots\times\mathbb{C}P^{\infty}), and the β\beta–family as framed hypersurfaces, Math. Z. 189 (1985) 239–261 \xoxMR779220
  • [3] P J Eccles, Filtering framed bordism by embedding codimension, J. London Math. Soc. (2)(2) 19 (1979) 163–169 \xoxMR527750
  • [4] P J Eccles, G Walker, The elements β1\beta_{1} are representable as framed hypersurfaces, J. London Math. Soc. (2)(2) 22 (1980) 153–160 \xoxMR579819
  • [5] M Imaoka, The double complex transfer at the prime 22, Topology Appl. 72 (1996) 199–207 \xoxMR1406309
  • [6] M Mahowald, A new infinite family in 2πs{~}_{2}\pi_{*}^{s}, Topology 16 (1977) 249–256 \xoxMR0445498
  • [7] K Morisugi, Lifting problem of η\eta and Mahowald’s element ηj\eta_{j}, Publ. Res. Inst. Math. Sci. 25 (1989) 407–414 \xoxMR1018509
  • [8] J Mukai, On stable homotopy of the complex projective space, Japan. J. Math. ((N.S.)) 19 (1993) 191–216 \xoxMR1231514
  • [9] E Rees, Framings on hypersurfaces, J. London Math. Soc. (2)(2) 22 (1980) 161–167 \xoxMR579820
  • [10] H Toda, Composition methods in homotopy groups of spheres, Annals of Mathematics Studies 49, Princeton University Press, Princeton, N.J. (1962) \xoxMR0143217