
Proof.
Given a weakly
$\mathcal{J}$
stable,
${M}_{2}$
stable and splitexact functor
$E:LCA\to Ab$
, these properties are inherited by the functor
${E}^{A}:LCA\to Ab$
which assigns
$B\mapsto E\left(A\hat{\otimes}B\right)$
, for any given locally convex algebra
$A$
. This implies that w.l.o.g. it is enough to show that the two evaluation maps from
$E(\mathbb{C}[0,1\left]\right)$
to
$E(\mathbb{C})$
are equal. Using the result of Lemma 3.4.2 and the injectivity of $E\left(j\right):E(\mathbb{C})\to E(\mathcal{K})$ , which followed from weak $\mathcal{J}$ stability and remark 4.1.2 , it suffices to show that we have an equality of $(\mathbb{C}[0,1],\mathcal{K})$ Kasparov modules $$E(\phi ,{u}_{0},P)=E(\phi ,{u}_{1},P).$$ Using the weak $\mathcal{J}$ stability again we conclude that it also suffices to show that the following $(\mathbb{C}[0,1],\mathcal{K}\hat{\otimes}\mathcal{J}))$ Kasparov modules are equal:$$E(\phi \otimes 1,{u}_{0}\oplus 1\oplus ...,P\otimes 1)=E(\phi \otimes 1,{u}_{1}\oplus 1\oplus ...,P\otimes 1).$$ Note that ${u}_{0}{{u}_{1}}^{1}={e}^{ih}$ for some selfadjoint element $h\in {C}^{\infty}\left({S}^{1}\right)$ . Denote by $Z$ the unitary in $\left({C}^{\infty}\right({S}^{1})\hat{\otimes}\mathcal{J}{)}^{+}\subset (D\hat{\otimes}\mathcal{J}{)}^{*}$ given by $$Z={e}^{ih}\oplus 1\oplus 1\oplus ....$$ where $D$ is as in 3.4 .We have to show that $E(\phi \otimes 1,Z,P\otimes 1)=0$ . The element $Z$ is the product of elements $${X}_{1}={e}^{ih}\oplus {e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{8}}\oplus ...$$ and $${X}_{2}=1\oplus {e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{4}}\oplus {e}^{\frac{ih}{8}}\oplus ....$$ Here, the term with $\pm \frac{ih}{{2}^{n}}$ appears precisely ${2}^{n}$ times. The elements ${X}_{i}$ lie in $\left({\mathcal{C}}^{\infty}\right({S}^{1})\hat{\otimes}\mathcal{J}{)}^{+}$ since they are exponentials of elements $ih\otimes {x}_{i}$ with ${x}_{i}\le 2\omega \in \mathcal{J}$ .This is the only place, where we use the assumption that $\mathcal{J}$ is harmonic in an essential way. Note that the elements ${X}_{i}$ commute with ${C}^{\infty}\left({S}^{1}\right)\otimes 1$ and with $P\otimes 1$ modulo $\mathcal{K}\hat{\otimes}\mathcal{J}$ and hence define abstract Kasparov $(\mathbb{C}[0,1],\mathcal{K}\hat{\otimes}\mathcal{J})$ modules relative to $(D\hat{\otimes}\mathcal{J}{)}^{+}$ .Note further that ${X}_{1}={W}_{1}\cdot {W}_{2}$ with $${W}_{1}={e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{2}}\oplus 1\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus ...$$ and $${W}_{2}={e}^{\frac{ih}{2}}\oplus 1\oplus {e}^{\frac{ih}{2}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus {e}^{\frac{ih}{8}}\oplus 1\oplus ....$$ Clearly, ${W}_{2}=(1\otimes Y){W}_{1}^{1}(1\otimes {Y}^{1})$ for some permutation matrix $Y\in \mathcal{\mathcal{B}}\left(H\right)$ .We see that $\left[\phi \right(\mathbb{C}[0,1])\otimes 1,1\otimes Y]=[P\otimes 1,1\otimes Y]=0$ so that abstract Kasparov $(\mathbb{C}[0,1],\mathcal{K}\otimes \mathcal{J})$ modules relative to $D\hat{\otimes}(\mathcal{\mathcal{B}}(H)\mathbb{\u22c9}\mathcal{J})$ are defined, since $\mathcal{K}\hat{\otimes}\mathcal{J}$ is a closed ideal in $D\hat{\otimes}(\mathcal{\mathcal{B}}(H)\mathbb{\u22c9}\mathcal{J})$ . (For a definition of $\mathcal{\mathcal{B}}\left(H\right)\mathbb{\u22c9}\mathcal{J}$ see section 2.2 .) By lemma 3.3.4 (a) wee see that $E(\phi \otimes 1,{X}_{1},P\otimes 1)=E(\phi \otimes 1,{W}_{1}1\otimes Y{W}_{1}^{1}(1\otimes Y{)}^{1},P\otimes 1)=0$ .A similar reasoning applies to ${X}_{2}$ . This finishes the proof. □
5 ${K}_{0}$ of a stable algebra.
5.1 Stabilized functors
In this section we consider splitexact and
${M}_{2}$
stable functors
${E}^{\prime}:LCA\to Ab$
, defined on the category of locally convex algebras.
Definition 5.1.1.
Let
$A$
be a locally convex algebra and let
${E}^{\prime}$
be a
${M}_{2}$
stable and splitexact functor. Let
$\mathcal{J}$
be a Banach ideal. The algebra
$A$
is called weakly
$\mathcal{J}$
stable with respect to the functor
${E}^{\prime}$
if the functor
$B\mapsto {E}^{\prime}\left(B\hat{\otimes}A\right)$
is weakly
$\mathcal{J}$
stable (see definition 4.1.1 ).
Proposition 5.1.2 shows that
$\mathcal{J}$
is weakly
$\mathcal{J}$
stable, so that there is always one obvious weakly
$\mathcal{J}$
stable algebra. Moreover, if
$B$
is weakly
$\mathcal{J}$
stable and
$A$
is any locally convex algebra, the
$A\hat{\otimes}B$
is weakly
$\mathcal{J}$
stable.
Let now
$\mathcal{J}$
be a harmonic Banach ideal and
$A$
be a weakly
$\mathcal{J}$
stable algebra with respect to the functor
${E}^{\prime}$
. In this section we show, as a corollary of theorem 4.2.1 , that the associated
$A$
stabilized functor
$E={E}^{\prime}(?\hat{\otimes}A):LCA\to Ab$
satisfies diffotopy invariance. See the remark after definition 3.1.1 for important examples of splitexact and
${M}_{2}$
stable functors to which such a result could be applied.
Proposition 5.1.2.
The functor
$E:LCA\to Ab$
which assigns
$A\mapsto {E}^{\prime}(A\hat{\otimes}\mathcal{J})$
is weakly
$\mathcal{J}$
stable in the sense of definition 4.1.1 , i.e. the algebra
$\mathcal{J}$
is weakly
$\mathcal{J}$
stable with respect to any
${M}_{2}$
stable and split exact functor.

Proof.
The natural map
$\theta :\mathcal{J}\left(H\right)\hat{\otimes}\mathcal{J}\left(H\right)\u27f6\mathcal{J}\left(H\hat{\otimes}H\right)$
induces a natural map
${\theta}_{A}:E(A\hat{\otimes}\mathcal{J})\to E\left(A\right)$
for every locally convex algebra
$A$
. We want to show that
${\theta}_{A}\circ {j}_{A}=i{d}_{E\left(A\right)}$
for the natural map
${j}_{A}:E\left(A\right)\to E(A\hat{\otimes}\mathcal{J})$
induced by the inclusion
$j:\mathbb{C}\to \mathcal{J}$
, i.e.
${j}_{A}={E}^{\prime}\left(i{d}_{A}\hat{\otimes}j\hat{\otimes}i{d}_{\mathcal{J}}\right)$
. There is an isometry $V$ in $\mathcal{\mathcal{B}}\left(H\hat{\otimes}H\right)$ such that $\theta \circ \left(j\hat{\otimes}i{d}_{\mathcal{J}}\right)\left(x\right)=Ad\left(V\right)=Vx{V}^{*}$ for $x\in \mathcal{J}$ . Choose a second isometry ${V}^{\prime}$ in $\mathcal{\mathcal{B}}\left(H\hat{\otimes}H\right)$ such that $V{V}^{*}+{V}^{\prime}{{V}^{\prime}}^{*}=1$ .Denote by ${O}_{2}$ the algebra generated algebraically by $1\hat{\otimes}V,1\hat{\otimes}{V}^{*}$ and $1\hat{\otimes}{V}^{\prime},1\hat{\otimes}{{V}^{\prime}}^{*}$ in $A\hat{\otimes}\mathcal{\mathcal{B}}\left(H\hat{\otimes}H\right)$ and by $D$ the algebra generated by $A\hat{\otimes}\mathcal{J}$ together with $1\hat{\otimes}V,1\hat{\otimes}{V}^{*}$ and $1\hat{\otimes}{V}^{\prime},1\hat{\otimes}{{V}^{\prime}}^{*}$ inside $A\hat{\otimes}\mathcal{\mathcal{B}}\left(H\hat{\otimes}H\right)$ . We have a split extension $$0\to A\hat{\otimes}\mathcal{J}\to D\to {O}_{2}\to 0$$ It is easy to see that there is a unitary $U$ in ${M}_{2}\left({O}_{2}\right)$ and hence in ${M}_{2}\left(D\right)$ such that $Ux{U}^{*}=\left(1\hat{\otimes}V\right)x\left(1\hat{\otimes}{V}^{*}\right)$ for $x$ in the subalgebra $$\left(\begin{array}{cc}A\hat{\otimes}\mathcal{J}& 0\\ 0& 0\end{array}\right)$$ of ${M}_{2}(A\hat{\otimes}\mathcal{J})$ . Thus, by ${M}_{2}$ stability of $E$ it is clear that $E(Ad(1\hat{\otimes}V\left)\right)=id$ on $E\left(D\right)$ . On the other hand, from the split extension above, we see that $E\left(D\right)=E(A\hat{\otimes}\mathcal{J})\oplus E\left({O}_{2}\right)$ so that the restriction of $Ad\left(1\hat{\otimes}V\right)$ induces the identity on $E(A\hat{\otimes}\mathcal{J})$ also. □
Corollary 5.1.3.
Let
${E}^{\prime}:LCA\to Ab$
be an
${M}_{2}$
stable and split exact functor and let
$A$
be a weakly
$\mathcal{J}$
stable algebra with respect to the functor
${E}^{\prime}$
, for a fixed harmonic Banach ideal
$\mathcal{J}$
. Under these circumstances, the functor
$E={E}^{\prime}(?\hat{\otimes}A)$
is diffotopy invariant.
 Proof. This follows now from 4.2.1 . □
5.2 Algebraic $K$ theory
We continue by applying Corollary 5.1.3 to the algebraic
$K$
theory functor
${K}_{0}$
and identify algebraic
$K$
theory of stable algebras with a suitable group of homotopy classes of maps.
Consider the algebra
$Q\mathbb{C}=\mathbb{C}*\mathbb{C}$
and denote by
$e,\overline{e}$
the two generators
$e={\iota}_{1}\left(1\right),\overline{e}={\iota}_{2}\left(1\right)$
. The argument in [8] , 3.1, shows that
$E(Q\mathbb{C}\hat{\otimes}A)\sim =E\left(A\right)\oplus E\left(A\right)$
and that
$E(q\mathbb{C}\hat{\otimes}A)\sim =E\left(A\right)$
for each
${M}_{2}$
stable and diffotopy invariant functor
$E$
. With the classical description of
${K}_{0}$
in [1] , the generator of
${K}_{0}(q\mathbb{C}\hat{\otimes}\mathcal{J})$
is given by the difference of equivalence classes of the idempotent elements
$p$
and
$\overline{p}$
in
${M}_{2}\left(\right(q\mathbb{C}\hat{\otimes}\mathcal{J}{)}^{+})$
:
$$p=W\left(\begin{array}{cc}{\overline{e}}^{\perp}& 0\\ 0& e\end{array}\right){W}^{1}\text{where}W=\left(\begin{array}{cc}{\overline{e}}^{\perp}& \overline{e}\\ \overline{e}& {\overline{e}}^{\perp}\end{array}\right)$$
and,
$$\overline{p}=\left(\begin{array}{cc}1& 0\\ 0& 0\end{array}\right).$$
Note that
$p\overline{p}\in {M}_{2}(q\mathbb{C}\hat{\otimes}\mathcal{J})$
and that therefore
$\left[p\right]\left[\overline{p}\right]\in {K}_{0}(q\mathbb{C}\hat{\otimes}\mathcal{J})$
.
Proposition 5.2.1.
For each locally convex algebra
$A$
, one has natural isomorphisms
$${K}_{0}(A\hat{\otimes}\mathcal{J})\sim =\langle q\mathbb{C},A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{\infty}\rangle \sim ={lim}_{{\u27f6}_{n}}\langle q\mathbb{C},A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\rangle .$$

Proof.
First of all, note that the righthand side is an abelian semigroup by block sum. The second isomorphism follows from the fact that
$q\mathbb{C}$
is finitely generated and from properties of the fine topology on
${M}_{\infty}$
. We construct a map $\phi :\langle q\mathbb{C},A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{\infty}\rangle \to {K}_{0}(A\hat{\otimes}\mathcal{J}).$ It sends a diffotopy class $\langle h\rangle \in \langle q\mathbb{C},A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{\infty}\rangle $ to the difference of equivalence classes of projections $\left[h\right(p\left)\right]\left[h\right(\overline{p}\left)\right]$ in ${K}_{0}\left(\right(A\hat{\otimes}\mathcal{J}{)}^{+})$ . The difference lies in the direct summand ${K}_{0}(A\hat{\otimes}\mathcal{J})$ .The map is welldefined by the preceding corollary.This map is surjective since all generators of ${K}_{0}^{alg}(A\hat{\otimes}\mathcal{J})$ and their negatives are hit and since the map is clearly a map of abelian semigroups. In order to prove injectivity we use Lemma $7.1$ in [9] which we reprove for sake of completeness.Lemma 5.2.2. Let $\phi :q\mathbb{C}\to {M}_{2}(q\mathbb{C})$ be the restriction of the homomorphism $Q\mathbb{C}\to {M}_{2}\left(\right(Q\mathbb{C}{)}^{+})$ , which sends $e$ to $p$ and $\overline{e}$ to $\overline{p}$ . Then $\phi $ is diffotopic to the inclusion map $\iota :q\mathbb{C}\to {M}_{2}(q\mathbb{C})$ .

Proof.
Let
${\gamma}_{t}:q\mathbb{C}\to {M}_{2}(q\mathbb{C}),t\in [0,\pi /2]$
be the restriction of the homomorphism
${\gamma}_{t}^{\prime}:Q\mathbb{C}\to {M}_{2}\left(\right(Q\mathbb{C}{)}^{+})$
which is defined by
$$\begin{array}{ccc}& {\gamma}_{t}^{\prime}\left(e\right)={W}_{t}\left(\begin{array}{cc}{\overline{e}}^{\perp}& 0\\ 0& e\end{array}\right){W}_{t}& \end{array}$$ $$\begin{array}{ccc}& & \end{array}$$
where $${W}_{t}=\left(\begin{array}{cc}{\overline{e}}^{\perp}& 0\\ 0& {\overline{e}}^{\perp}\end{array}\right)+\left(\begin{array}{cc}\overline{e}cost& \overline{e}sint\\ \overline{e}sint& \overline{e}cost\end{array}\right)$$ For each $t$ the difference ${\gamma}_{t}^{\prime}\left(e\right){\gamma}_{t}^{\prime}\left(\overline{e}\right)$ lies in the ideal ${M}_{2}(q\mathbb{C})$ . Therefore ${\gamma}_{t}$ defines a diffotopy, which connects $\phi $ with $\iota $ . □$$\begin{array}{ccc}& {\gamma}_{t}^{\prime}\left(\overline{e}\right)={W}_{t}\left(\begin{array}{cc}{\overline{e}}^{\perp}& 0\\ 0& \overline{e}\end{array}\right){W}_{t}& \end{array}$$
This means that the there exists a projector $q\in (A\hat{\otimes}\mathcal{J}{)}^{+}\hat{\otimes}{M}_{k}$ and an invertible element $u\in (A\hat{\otimes}\mathcal{J}{)}^{+}\hat{\otimes}{M}_{2n+k}$ such that $u\left({\eta}_{1}^{\prime}\right(p)\oplus q){u}^{1}={\eta}_{2}^{\prime}\left(p\right)\oplus q$ . This element $u$ can even be chosen to be connected to $1$ by a differentiable family ${u}_{t},t\in [1,2]$ , such that $1{u}_{t}\in A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{2n+k}$ for all $t$ .Consider the homomorphisms ${\zeta}_{1},{\zeta}_{2}:q\mathbb{C}\to A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{2n+k}$ defined as restrictions of the maps from $Q\mathbb{C}$ that map $e$ to ${\eta}_{1}^{\prime}\left(p\right)\oplus q$ and ${\eta}_{2}^{\prime}\left(p\right)\oplus q$ , respectively, and $\overline{e}$ to $\overline{p}\oplus q$ . Note that ${\zeta}_{i}={M}_{2}\left({\eta}_{i}\right)\circ \phi $ .According to Lemma 5.2.2 , ${\zeta}_{1}={M}_{2}\left({\eta}_{1}\right)\circ \phi $ is diffotopic to ${\eta}_{1}={M}_{2}\left({\eta}_{1}\right)\circ \iota $ and similarly ${\zeta}_{2}$ is diffotopic to ${\eta}_{2}$ . On the other hand, the family ${\zeta}_{t}$ , $t\in [1,2]$ , of homomorphisms $q\mathbb{C}\to A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{2n+k}$ , obtained as restrictions of the maps from $Q\mathbb{C}$ , which map $e$ to ${u}_{t}\left({\eta}_{1}^{\prime}\right(p)\oplus q){{u}_{t}}^{1}$ and $\overline{e}$ to ${\eta}_{1}^{\prime}\left(\overline{p}\right)\oplus q$ , defines a diffotopy connecting ${\zeta}_{1}$ to ${\zeta}_{2}$ . □ 
Proof.
Let
${\gamma}_{t}:q\mathbb{C}\to {M}_{2}(q\mathbb{C}),t\in [0,\pi /2]$
be the restriction of the homomorphism
${\gamma}_{t}^{\prime}:Q\mathbb{C}\to {M}_{2}\left(\right(Q\mathbb{C}{)}^{+})$
which is defined by
6 Determination of $k{k}^{\mathcal{J}}(\mathbb{C},A)$
6.1 Bivariant $kk$ theory revisited
We now finally have to use the explicit definition of
$k{k}^{alg}$
. To this end we recall some constructions and notation from [10] .
Let
$V$
be a complete locally convex space. Consider the algebraic tensor algebra
$${T}_{alg}V=V\oplus V\otimes V\oplus {V}^{{\otimes}^{3}}\oplus ...$$
with the usual product given by concatenation of tensors. There is a canonical linear map
$\sigma :V\to {T}_{alg}V$
mapping
$V$
into the first direct summand. We equip
${T}_{alg}V$
with the locally convex topology given by the family of all seminorms of the form
$\alpha \circ \phi $
, where
$\phi $
is any homomorphism from
${T}_{alg}V$
into a locally convex algebra
$B$
such that
$\phi \circ \sigma $
is continuous on
$V$
, and
$\alpha $
is a continuous seminorm on
$B$
. We further denote by
$TV$
the completion of
${T}_{alg}V$
with respect to this locally convex structure.
For any locally convex algebra
$A$
we have the natural extension
Here
$\pi $
maps a tensor
${x}_{1}\otimes {x}_{2}\otimes \dots \otimes {x}_{n}$
to
${x}_{1}{x}_{2}\dots {x}_{n}\in A$
and
$JA$
is defined as Ker
$\pi $
. This extension is (uni)versal in the sense that, given any extension
$0\to I\to E\to B\to 0$
of a locally convex algebra
$B$
, admitting a continuous linear splitting, and any continuous homomorphism
$\alpha :A\to B$
, there is a morphism of extensions
The map
$\tau :TA\to E$
is obtained by choosing a continuous linear splitting
$s:B\to E$
in the given extension and mapping
${x}_{1}\otimes {x}_{2}\otimes \dots \otimes {x}_{n}$
to
${s}^{\prime}\left({x}_{1}\right){s}^{\prime}\left({x}_{2}\right)\dots {s}^{\prime}\left({x}_{n}\right)\in E$
, where
${s}^{\prime}:=s\circ \alpha $
. Then
$\gamma $
is the restriction of
$\tau $
.
$$\begin{array}{c}0\to JA\to TA\stackrel{\pi}{\to}A\to 0.\end{array}$$  (3) 
$$\begin{array}{c}\begin{array}{ccccccccc}0& \to & JA& \to & TA& \to & A& \to & 0\\ & & \downarrow \gamma & & \downarrow \tau & & \downarrow \alpha & & \\ 0& \to & I& \to & E& \to & B& \to & 0\end{array}\end{array}$$  (4) 
Choosing
$0\to JB\to TB\to B\to 0$
in place of the second extension
$0\to I\to E\to B\to 0$
, we see that
$A\to JA$
is a functor, i.e. any homomorphism
$\alpha :A\to B$
induces a homomorphism
$J\left(\alpha \right):JA\to JB$
.
Using the universal property of
$TA$
, one can associate a classifying map with any linearly split extension of locally convex algebras of the form
$$0\to I\to {E}_{1}\to {E}_{2}\to \dots \to {E}_{n}\to A\to 0$$
We consider such an extension as a complex, denoting the arrows (boundary maps) by
${\pi}_{i}$
and we say that it is linearly split if there is a continuous linear map
$s$
of degree 1 such that
$s\pi +\pi s=id$
. Every such splitting
$s$
induces a commutative diagram of the form
The leftmost vertical arrow in this diagram is the classifying map for this
$n$
step extension. It depends on
$s$
only up to diffotopy.
$$\begin{array}{ccccccccccccccc}0& \to & I& \to & {E}_{1}& \to & {E}_{2}& \to & ...& \to & {E}_{n}& \to & A& \to & 0\end{array}$$ 
$$\begin{array}{ccccccccccccccc}& & \uparrow & & \uparrow & & \uparrow & & ...& & \uparrow & & \uparrow & & \end{array}$$ 
$$\begin{array}{ccccccccccccccc}0& \to & {J}^{n}A& \to & T\left({J}^{n1}A\right)& \to & T\left({J}^{n2}A\right)& \to & ...& \to & TA& \to & A& \to & 0\end{array}$$ 
Recall then from [10] the following definitions.
Definition 6.1.1.
Let
$A$
and
$B$
be locally convex algebras. For any continuous homomorphism
$\phi :A\to B$
, we denote by
$\langle \phi \rangle $
the equivalence class of
$\phi $
with respect to diffotopy and we set
$$\langle A,B\rangle =\{\langle \phi \rangle \phi \text{is a continuous homomorphism}A\to B\}$$
Given
$n\in \mathbb{Z}$
we set
$$k{k}_{n}^{alg}(A,B)={lim}_{{\u27f6}_{k}}\langle {J}^{kn}A,{\Sigma}^{k}B\hat{\otimes}\mathcal{K}\rangle $$
where the inductive limit is with respect to the natural maps
$$\langle {J}^{k}A,\mathcal{K}\hat{\otimes}B(0,1{)}^{k}\rangle \to \langle {J}^{k+1}A,\mathcal{K}\hat{\otimes}B(0,1{)}^{k+1}\rangle $$
mapping the diffotopy class of
$\alpha $
to the diffotopy class of
${\alpha}^{\prime}$
, where
${\alpha}^{\prime}$
is defined by the commutative diagram
where
${B}^{\prime}=\mathcal{K}\hat{\otimes}B$
.
Thus, by definition of
$k{k}^{\mathcal{J}}$
, we have
$$k{k}_{n}^{alg}(A,B)={lim}_{{\u27f6}_{k}}\langle {J}^{kn}A,{\Sigma}^{k}B\hat{\otimes}\mathcal{J}\hat{\otimes}\mathcal{K}\rangle $$
In the sequel we will also use stabilization by the finite matrix algebras
${M}_{n}$
.
$$\begin{array}{ccccccccc}0& \to & {J}^{k+1}A& \to & T{J}^{k}A& \to & {J}^{k}A& \to & 0\end{array}$$ 
$$\begin{array}{ccccccccc}& & \downarrow {\alpha}^{\prime}& & \downarrow \tau & & \downarrow \alpha & & \end{array}$$ 
$$\begin{array}{ccccccccc}0& \to & {B}^{\prime}(0,1{)}^{k+1}& \to & {B}^{\prime}(0,1{)}^{k}[0,1)& \to & {B}^{\prime}(0,1{)}^{k}& \to & 0\end{array}$$ 
Proposition 6.1.2.
The natural maps, induced by the inclusion
${M}_{n}\to \mathcal{K}$
,
$$\langle {J}^{k}A,{\Sigma}^{k}B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\rangle \to \langle {J}^{k}A,{\Sigma}^{k}B\hat{\otimes}\mathcal{J}\hat{\otimes}\mathcal{K}\rangle $$
define an isomorphism
$\alpha $
of abelian groups:
$${lim}_{{\u27f6}_{k}}{lim}_{{\u27f6}_{n}}\langle {J}^{k}A,{\Sigma}^{k}B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\rangle \sim =k{k}_{0}^{\mathcal{J}}(A,B).$$

Proof.
Let
$V$
be an isometry in
$\mathcal{\mathcal{B}}\left(H\right)$
and
${j}_{n}:\mathcal{J}\to {M}_{n}(\mathcal{J})$
the inclusion map (into the upper left corner). A standard argument shows that
$j\circ AdV:\mathcal{J}\to \mathcal{J}\otimes {M}_{n}$
is diffotopic to
$j$
. Denote by $\theta :\mathcal{J}\hat{\otimes}\mathcal{J}\to \mathcal{J}$ the natural tensor product of operators and by $\phi :\mathcal{K}\to \mathcal{J}$ the natural inclusion. We claim that the inverse $\beta $ to the map $\alpha $ above is induced by $${j}_{n}\circ \theta \circ (i{d}_{\mathcal{J}}\otimes \phi ):\mathcal{J}\hat{\otimes}\mathcal{K}\to \mathcal{J}\otimes {M}_{n}$$ The identity $\beta \circ \alpha =1$ follows from the fact that the composition $$\mathcal{J}\sim =\mathcal{J}\otimes {M}_{n}\to \mathcal{J}\hat{\otimes}\mathcal{K}\stackrel{i{d}_{\mathcal{J}}\hat{\otimes}\phi}{\u27f6}\mathcal{J}\hat{\otimes}\mathcal{J}\stackrel{\theta}{\to}\mathcal{J}$$ is of the form $AdV$ as above.To show that $\alpha \circ \beta =1$ , consider the inclusions ${\iota}_{l},{\iota}_{r}:\mathcal{K}\to \mathcal{\mathscr{H}}\hat{\otimes}\mathcal{K}$ into the left, resp. right, factor using the standard rank 1 projector onto the first basis vector in $H$ . By [10] , 2.2.1, ${\iota}_{l}$ and ${\iota}_{r}$ are diffotopic and in fact both diffotopic to the natural isomorphism $\mathcal{K}\to \mathcal{K}\hat{\otimes}\mathcal{K}$ .We want to show that the composition of the following maps $$\mathcal{J}\hat{\otimes}\mathcal{K}\stackrel{i{d}_{\mathcal{J}}\hat{\otimes}\phi}{\u27f6}\mathcal{J}\hat{\otimes}\mathcal{J}\stackrel{\theta}{\to}\mathcal{J}\to \mathcal{J}\hat{\otimes}\mathcal{K}$$ is diffotopic to the identity. This composition can also be factored as follows $$\mathcal{J}\hat{\otimes}\mathcal{K}\stackrel{i{d}_{\mathcal{J}}\hat{\otimes}{\iota}_{l}}{\u27f6}\mathcal{J}\hat{\otimes}\mathcal{K}\hat{\otimes}\mathcal{K}\stackrel{i{d}_{\mathcal{J}}\hat{\otimes}\phi \hat{\otimes}i{d}_{\mathcal{K}}}{\u27f6}\mathcal{J}\hat{\otimes}\mathcal{J}\hat{\otimes}\mathcal{K}\stackrel{\theta \hat{\otimes}i{d}_{\mathcal{K}}}{\to}\mathcal{J}\hat{\otimes}\mathcal{K}.$$ Replacing in this composition ${\iota}_{l}$ by the diffotopic map ${\iota}_{r}$ we again obtain a map of $\mathcal{J}\hat{\otimes}\mathcal{K}\to \mathcal{J}\hat{\otimes}\mathcal{K}$ of the form $AdV\otimes i{d}_{\mathcal{K}}$ which is diffotopic to id (using the fact that ${j}_{2}:\mathcal{K}\to \mathcal{K}\otimes {M}_{2}$ is diffotopic to the natural isomorphism $\mathcal{K}\sim =\mathcal{K}\otimes {M}_{2}$ ). □
6.2 Main theorem
Our main result in this section is the following computation. Its proof requires some preparation and is given in section 6.3 .
Theorem 6.2.1.
For every locally convex algebra
$A$
and for every harmonic Banach ideal
$\mathcal{J}$
one has
$k{k}_{0}^{\mathcal{J}}(\mathbb{C},A)={K}_{0}(A\hat{\otimes}\mathcal{J})$
.
Remark 6.2.2.
In particular this shows that
${K}_{0}\left(A\hat{\otimes}{\mathcal{\mathcal{L}}}_{p}\right)$
does not depend on
$p$
for
$1<p<\infty $
(see also [14] , 4.1).
Corollary 6.2.3.
Let
$\mathcal{J}$
be a harmonic Banach ideal. The coefficient ring
$k{k}_{*}^{\mathcal{J}}(\mathbb{C},\mathbb{C})$
is isomorphic to
$\mathbb{Z}[u,{u}^{1}]$
.

Proof.
By properties of
$k{k}^{alg}$
we are reduced to a computation of
$k{k}_{0}^{\mathcal{J}}(\mathbb{C},\mathbb{C})$
and
$k{k}_{0}^{\mathcal{J}}(\mathbb{C},\Sigma )$
. By Theorem 6.2.1 these groups are isomorphic to the algebraic
$K$
groups
${K}_{0}(\mathcal{J})$
and
${K}_{0}(\Sigma \mathcal{J})$
. Both algebras appearing are smooth subalgebras of
${C}^{*}$
algebras whose
$K$
theory is wellknown, i.e. $${K}_{0}(\mathcal{J})={K}_{0}\left({\mathcal{\mathcal{L}}}_{\infty}\right)=\mathbb{Z}$$ and $${K}_{0}(\Sigma \mathcal{J})={K}_{0}\left({\mathcal{C}}_{0}\right(\mathbb{R},{\mathcal{\mathcal{L}}}_{\infty}\left)\right)=0.$$ This finishes the proof of the corollary. □
Corollary 6.2.4.
Let
$W$
be the Weyl algebra, i.e the unital algebra with two generators
$x$
and
$y$
satisfying the relation
$xyyx=1$
(with the fine topology). Then for every harmonic Banach ideal
$\mathcal{J}$
we have
$k{k}_{0}^{\mathcal{J}}(\mathbb{C},W)=\mathbb{Z}$
and
$k{k}_{1}^{\mathcal{J}}(\mathbb{C},W)=0$
.
Similarly, the
$\mathcal{J}$
stable
$K$
theory of many other algebras can now be computed as an abelian group rather than just as a module over the coefficient ring.
In order to organize the notation in the following computations, we introduce a category
${H}_{\mathcal{J}}$
. The objects are locally convex algebras and the morphisms between two locally convex algebras are given by
$$[A,B]={lim}_{{\u27f6}_{n}}\langle A,B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\rangle .$$
Given two diffotopy classes of continuous homomorphisms
$\phi :A\to B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}$
and
$\psi :B\to C\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{m}$
, their composition in
${H}_{\mathcal{J}}$
is defined to be the diffotopy class of
$$A\stackrel{\phi}{\to}B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\stackrel{\psi \hat{\otimes}id}{\u27f6}B\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{m}\stackrel{id\hat{\otimes}\theta}{\u27f6}C\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{nm}.$$
It is clear that composition is associative at the level of diffotopy classes. Moreover, note that endofunctors like
$J(?)$
and
$\Sigma \hat{\otimes}?$
descend to endofunctors on
${H}_{\mathcal{J}}$
, since, for example, there is a natural map
$J(A\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n})\to J\left(A\right)\hat{\otimes}\mathcal{J}\hat{\otimes}{M}_{n}$
.
In particular, using the new notation, we have
$$k{k}_{i}^{\mathcal{J}}(A,B)={lim}_{{\u27f6}_{k}}[{J}^{ki}A,{\Sigma}^{k}B]$$
by proposition 6.1.2 and
$${K}_{0}(A\hat{\otimes}\mathcal{J})=[q\mathbb{C},A]$$
by proposition 5.2.1 .
Proposition 6.2.5.
For all locally convex algebras
$A$
and
$B$
one has
$$k{k}_{i}^{\mathcal{J}}(A,B)={lim}_{{\u27f6}_{k}}[{J}^{2ki}A,B]$$
This identity was noted in [10] , remark 8.4. In order to give an explicit proof, we have to introduce some notation. Throughout, we are working in the category
${H}_{\mathcal{J}}$
.
Denote by
${\rho}_{A}:JA\to \Sigma A$
the classifying map of the cone extension of
$A$
. Denote by
${\varepsilon}_{A}^{\prime}:J\Sigma A\to A$
the classifying map of the Toeplitz extension tensored with
$A$
. We define
${\varepsilon}_{A}={\varepsilon}_{A}^{\prime}\circ J\left({\rho}_{A}\right):{J}^{2}A\to A$
. (The map
${\varepsilon}_{A}$
has the important interpretation as the classifying map for the
$2$
step extension given by the Yoneda product of the Toeplitz extension and the cone extension.) Furthermore, note that
$${J}^{2}\left({\varepsilon}_{A}\right)={\varepsilon}_{{J}^{2}\left(A\right)}$$
by Korollar 3.1.1 in [9] , or by Lemma 4.6 in [10] , for all algebras
$A$
. For
$\phi :A\to B$
, we also have the identity
$$\phi \circ {\varepsilon}_{A}={\varepsilon}_{B}\circ {J}^{2}\left(\phi \right).$$
We define inductively
${{\varepsilon}^{\prime}}_{A}^{n}:{J}^{n}{\Sigma}^{n}A\to A$
by setting
${{\varepsilon}^{\prime}}_{A}^{n}={{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}\left({\varepsilon}_{{\Sigma}^{n1}A}^{\prime}\right)$
and
${\rho}_{A}^{n}:{J}^{n}A\to {\Sigma}^{n}A$
by setting
${\rho}_{A}^{n}={\rho}_{{\Sigma}^{n1}A}\circ J\left({\rho}_{A}^{n1}\right)$
. We also define
${\varepsilon}_{A}^{n}:{J}^{2n}A\to A$
by setting
${\varepsilon}_{A}^{n}={\varepsilon}_{A}^{n1}\circ {\varepsilon}_{{J}^{2n2}A}$
. Note that, in the definition of
${\varepsilon}_{A}^{n}$
, all other choices lead to the same definition of
${\varepsilon}_{A}^{n}$
, since
${J}^{2}\left({\varepsilon}_{A}\right)={\varepsilon}_{{J}^{2}A}$
. Note that
${{\varepsilon}^{\prime}}_{A}^{n}\circ {J}^{n}\left({\rho}_{A}^{n}\right)={\varepsilon}_{A}^{n}:{J}^{2n}A\to A$
, as the following induction argument shows.
With this notation, the abelian group
$k{k}^{\mathcal{J}}(A,B)$
is defined as the direct limit of a system of abelian groups
$[{J}^{n}A,{\Sigma}^{n}B]$
via a stabilization map
$$[{J}^{n}A,{\Sigma}^{n}B]\ni \psi \mapsto {\rho}_{{\Sigma}^{n}B}\circ J\left(\psi \right)\in [{J}^{n+1},{\Sigma}^{n+1}B].$$
The
$k$
th stabilization map is given by
$$[{J}^{n}A,{\Sigma}^{n}B]\ni \phi \mapsto {\rho}_{{\Sigma}^{n}B}^{k}\circ {J}^{k}\left(\phi \right)\in [{J}^{n+k}A,{\Sigma}^{n+k}B],$$
which, again, follows from an easy induction argument.
$$\begin{array}{ccc}{{\varepsilon}^{\prime}}_{A}^{n}\circ {J}^{n}\left({\rho}_{A}^{n}\right)& =& {{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}\left({\varepsilon}_{{\Sigma}^{n1}A}^{\prime}\right)\circ {J}^{n}({\rho}_{{\Sigma}^{n1}A}\circ J({\rho}_{A}^{n1}\left)\right)\end{array}$$ 
$$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}({\varepsilon}_{{\Sigma}^{n1}A}^{\prime}\circ J({\rho}_{{\Sigma}^{n1}A})\circ {J}^{2}({\rho}_{A}^{n1}\left)\right)\end{array}$$ 
$$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}({\varepsilon}_{{\Sigma}^{n1}A}\circ {J}^{2}({\rho}_{A}^{n1}\left)\right)\end{array}$$ 
$$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}({\rho}_{A}^{n1}\circ {\varepsilon}_{{J}^{n1}A})\end{array}$$ 
$$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{A}^{n1}\circ {J}^{n1}\left({\rho}_{A}^{n1}\right)\circ {\varepsilon}_{{J}^{2n2}A}\end{array}$$ 
$$\begin{array}{}\end{array}$$ 
The right hand side is defined as the direct limit of abelian semigroups
$[{J}^{2n}A,B]$
via the stabilization map
$$[{J}^{2n}A,B]\in \psi \mapsto {\varepsilon}_{B}\circ {J}^{2}\left(\psi \right)=\psi \circ {\varepsilon}_{{J}^{2n}A}\in [{J}^{2n+2}A,B].$$

Proof.
First of all, we define a map
$[{J}^{n}A,{\Sigma}^{n}B]\to [{J}^{2n}A,B]$
by
$\phi \mapsto {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}\left(\phi \right)\in [{J}^{2n},B].$
This assignment induces a map of directed systems since it is compatible with the stabilization maps. Indeed,
${\rho}_{{\Sigma}^{n}B}\circ J\left(\phi \right)$
is mapped to
${{\varepsilon}^{\prime}}_{B}^{n+1}\circ {J}^{n+1}({\rho}_{{\Sigma}^{n}B}\circ J(\phi \left)\right)$
and the following computation shows that this is the desired result.
$$\begin{array}{ccc}{{\varepsilon}^{\prime}}_{B}^{n+1}\circ {J}^{n+1}({\rho}_{{\Sigma}^{n}B}\circ J(\phi \left)\right)& =& {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}\left({\varepsilon}_{{\Sigma}^{n}B}^{\prime}\right)\circ {J}^{n}\left(J\right({\rho}_{{\Sigma}^{n}B}\circ {J}^{2}\left(\phi \right)\left)\right)\end{array}$$ $$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}({\varepsilon}_{{\Sigma}^{n}B}^{\prime}\circ J({\rho}_{{\Sigma}^{n}B})\circ {J}^{2}(\phi \left)\right)\end{array}$$ $$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}({\varepsilon}_{{\Sigma}^{n}B}\circ {J}^{2}(\phi \left)\right)\end{array}$$ $$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}(\phi \circ {\varepsilon}_{{J}^{n}A})\end{array}$$ $$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}\left(\phi \right)\circ {\varepsilon}_{{J}^{2n}A}\end{array}$$
This implies that there is a welldefined map $${\sigma}_{A,B}:{lim}_{{\u27f6}_{k}}[{J}^{k}A,{\Sigma}^{k}B]\to {lim}_{{\u27f6}_{k}}[{J}^{2k}A,B].$$ Consider a class $\left[\phi \right]\in {lim}_{{\u27f6}_{k}}[{J}^{k}A,{\Sigma}^{k}B]$ which is represented by a homomorphism $\phi :{J}^{n}A\to {\Sigma}^{n}B$ . If the composition ${{\varepsilon}^{\prime}}_{B}^{n}\circ {J}^{n}\left(\phi \right)\circ {\varepsilon}_{{J}^{2n}A}^{m}:{J}^{2m+2n}A\to B$ is homotopic to zero for some $m\in \mathbb{N}$ , then also $\left[{{\varepsilon}^{\prime}}_{B}^{n}\right]\circ \left[{J}^{n}\right(\phi \left)\right]\circ \left[{\varepsilon}_{{J}^{2n}A}^{m}\right]=0$ as element in ${lim}_{{\u27f6}_{k}}[{J}^{k}{J}^{2m+2n}A,{\Sigma}^{k}B].$ Note that, for any $A$ and $B$ , the functor $J$ maps ${lim}_{{\u27f6}_{k}}[{J}^{k}A,{\Sigma}^{k}B]$ isomorphically onto ${lim}_{{\u27f6}_{k}}[{J}^{k}JA,{\Sigma}^{k}JB]$ . Furthermore, $\left[{\varepsilon}_{A}^{\prime}\right]$ and $\left[{\varepsilon}_{A}\right]$ are invertible for all $A$ . This together implies that $\left[\phi \right]=0$ as a class in ${lim}_{{\u27f6}_{k}}[{J}^{k+n}A,{\Sigma}^{k+n}B]$ and hence that ${\sigma}_{A,B}$ is injective.$$\begin{array}{}\end{array}$$ We still have to show that ${\sigma}_{A,B}$ is surjective. Consider a class $\left[\phi \right]\in {lim}_{{\u27f6}_{k}}[{J}^{2k}A,B]$ which is represented by a homomorphism $\phi :{J}^{2n}A\to B$ . The element $\phi $ gives rise to a class in ${lim}_{{\u27f6}_{k}}[{J}^{k}{J}^{2n}A,{\Sigma}^{k}B]$ . The natural map $${lim}_{{\u27f6}_{k}}[{J}^{k}A,{\Sigma}^{k}B]\to {lim}_{{\u27f6}_{k}}[{J}^{k}{J}^{2n}A,{\Sigma}^{k}B]$$ which is induced by the assignment $$[{J}^{k}A,{\Sigma}^{k}B]\ni \psi \mapsto \psi \circ {J}^{k}\left({\varepsilon}_{A}^{n}\right)\in [{J}^{k}{J}^{2n}A,{\Sigma}^{k}B]$$ is an isomorphism. I.e. there is some $k\in \mathbb{N}$ and $\eta :{J}^{k}A\to {\Sigma}^{k}B$ such that $$\eta \circ {J}^{k}\left({\varepsilon}_{A}^{n}\right)={\rho}_{B}^{k}\circ {J}^{k}\left(\phi \right).$$ We claim that $\left[\eta \right]\in {lim}_{{\u27f6}_{k}}[{J}^{k}A,{\Sigma}^{k}B]$ does the job. Indeed, it is mapped to ${{\varepsilon}^{\prime}}_{B}^{k}\circ {J}^{k}\left(\eta \right):{J}^{2k}A\to B.$ Stabilizing yields$$\begin{array}{ccc}{{\varepsilon}^{\prime}}_{B}^{k}\circ {J}^{k}\left(\eta \right)\circ {\varepsilon}_{{J}^{2k}A}^{n}& =& {{\varepsilon}^{\prime}}_{B}^{k}\circ {J}^{k}\left(\eta \right)\circ {J}^{2k}\left({\varepsilon}_{A}^{n}\right)\end{array}$$ $$\begin{array}{ccc}& =& {{\varepsilon}^{\prime}}_{B}^{k}\circ {J}^{k}\left({\rho}_{B}^{k}\right)\circ {J}^{2k}\left(\phi \right)\end{array}$$ $$\begin{array}{ccc}& =& {\varepsilon}_{B}^{k}\circ {J}^{2k}\left(\phi \right)\end{array}$$ $$\begin{array}{ccc}& =& \phi \circ {\varepsilon}_{{J}^{2k}A}^{k}.\end{array}$$
This shows the surjectivity of ${\sigma}_{A,B}$ . It is another easy check to show, that the assignment is compatible with the several composition products. □$$\begin{array}{}\end{array}$$
6.3 Proof of the main theorem
We now proceed by proving Theorem 6.2.1 . By proposition 5.2.1 it suffices to shows that there is a natural isomorphism
$$[q\mathbb{C},A]\sim ={lim}_{{\u27f6}_{k}}[{J}^{2k}\mathbb{C},A].$$
The existence of a natural map will become apparent in the sequel of the proof.

Proof.
There is a natural map
${\alpha}_{1}:{J}^{2}(\mathbb{C})\to q\mathbb{C}$
given as the classifying map of a
$2$
step extension which comes as the Yonedaproduct of the Toeplitz extension and a canonical extension
$$0\to \Sigma q\mathbb{C}\to E\to \mathbb{C}\to 0,$$
where
$E=\{f:[0,1]\to Q\mathbb{C}f\left(0\right)\in \mathbb{C}*0,f\left(1\right)\in 0*\mathbb{C},f\left(0\right)f\left(t\right)\in q\mathbb{C},\forall t\in [0,1]\}.$
Denote by
$\delta :q\mathbb{C}\to \mathbb{C}$
the restriction of
$id*0:\mathbb{C}*\mathbb{C}\to \mathbb{C}$
. It follows from the definition of
${\varepsilon}_{\mathbb{C}}:{J}^{2}(\mathbb{C})\to \mathbb{C}$
that
$\delta \circ {\alpha}_{1}={\varepsilon}_{\mathbb{C}}$
. We define natural maps ${\alpha}_{n}={\alpha}_{1}\circ {\varepsilon}_{{J}^{2}\mathbb{C}}^{n1}\in [{J}^{2n}\mathbb{C},q\mathbb{C}]$ .Since $A\mapsto {K}_{0}(A\otimes \mathcal{J})$ is a halfexact and diffotopy invariant functor by Theorem 4.2.1 , we conclude by Lemma 4.1.5 that there are natural isomorphisms ${K}_{0}({J}^{2n}\mathbb{C}\hat{\otimes}\mathcal{J})\sim ={K}_{0}({\Sigma}^{2n}\mathcal{J})$ . However, the righthand side of the last equation is isomorphic to ${K}_{0}\left({\mathcal{C}}_{0}\right({\mathbb{R}}^{2n}\left)\right)=\mathbb{Z}$ and contains a canonical generator. Denote the canonical generators of ${K}_{0}({J}^{2n}\mathbb{C}\hat{\otimes}\mathcal{J})$ by ${\beta}_{m}\in [q\mathbb{C},{J}^{2n}(\mathbb{C}\left)\right]$ . The following identities are immediate, once we have the alternative description of ${K}_{0}(?\hat{\otimes}\mathcal{J})$ , given in proposition 5.2.1 .
$$\begin{array}{ccc}{\alpha}_{n}\circ {\beta}_{n}& =& i{d}_{q\mathbb{C}}\end{array}$$ $$\begin{array}{ccc}{\varepsilon}_{{J}^{nk}\mathbb{C}}^{k}\circ {\beta}_{n}& =& {\beta}_{nk}\end{array}$$
It is clear from the identities above that the maps ${\alpha}^{*}:li{m}_{n\in \mathbb{N}}[{J}^{2n}\mathbb{C},A]\to [q\mathbb{C},A]$ and ${\beta}^{*}:[q\mathbb{C},A]\to li{m}_{n\in \mathbb{N}}[{J}^{2n}\mathbb{C},A]$ , which are induced from ${\alpha}_{n}$ and ${\beta}_{n}$ by precomposition, are well defined. Furthermore, ${\beta}^{*}\circ {\alpha}^{*}$ is equal to the identity. We now show that ${\alpha}^{*}\circ {\beta}^{*}$ is also equal to the identity.$$\begin{array}{ccc}{\varepsilon}_{\mathbb{C}}^{n}\circ {\beta}_{n}& =& \delta \end{array}$$ $$\begin{array}{ccc}{\beta}_{n}\circ {\alpha}_{n}\circ {\varepsilon}_{{J}^{2n}\mathbb{C}}^{n}& =& {\varepsilon}_{{J}^{2n}\mathbb{C}}^{n}\circ {J}^{2n}({\beta}_{n}\circ {\alpha}_{n})\end{array}$$ $$\begin{array}{ccc}& =& {J}^{2n}\left({\varepsilon}_{\mathbb{C}}^{n}\right)\circ {J}^{2n}({\beta}_{n}\circ {\alpha}_{n})\end{array}$$ $$\begin{array}{ccc}& =& {J}^{2n}({\varepsilon}_{\mathbb{C}}^{n}\circ {\beta}_{n}\circ {\alpha}_{n})\end{array}$$ $$\begin{array}{ccc}& =& {J}^{2n}(\delta \circ {\alpha}_{n})\end{array}$$ $$\begin{array}{ccc}& =& {J}^{2n}(\delta \circ {\alpha}_{1}\circ {\varepsilon}_{{J}^{2}\mathbb{C}}^{n1})\end{array}$$ $$\begin{array}{ccc}& =& {J}^{2n}\left({\varepsilon}_{\mathbb{C}}^{n}\right)\end{array}$$
□$$\begin{array}{ccc}& =& {\varepsilon}_{{J}^{2n}\mathbb{C}}^{n}\end{array}$$
7 Computation of ${K}_{1}(\mathcal{J})$ for harmonic Banach ideals
In this section we want to give a computation of the algebraic
$K$
theory group
${K}_{1}(\mathcal{J})$
for a harmonic Banach ideal
$\mathcal{J}$
(see definition 2.2.2 ). The result contradicts an old result in [14] prop. 4.1. The error in the proof of proposition 4.1 in [14] was brought to our attention by Valqui and Cortin͂as. This concrete computation fits nicely with the far more general structure theorem about higher algebraic
$K$
theory of locally convex algebras stabilized by harmonic Banach ideals, which was obtained by Cortin͂as in [5] using our diffotopy invariance theorem for weakly
$\mathcal{J}$
stable,
${M}_{2}$
stable, splitexact functors.
Let
$A$
be a locally convex algebra. In this section
${A}^{+}$
denotes the unitization by
$\mathbb{Z}$
rather than by
$\mathbb{C}$
. A priori, this difference matters and one has to be careful not to use the complex unitization.
The abelian group
${K}_{1}(\mathcal{J})$
is defined as
$ker\left({K}_{1}\right({\mathcal{J}}^{+})\to {K}_{1}(\mathbb{Z}\left)\right)$
. Denote by
${\mathcal{J}}^{2}$
the (algebraic) square of
$\mathcal{J}$
, i.e. the image of the algebraic tensor product
$\mathcal{J}\otimes \mathcal{J}$
under the multiplication map. We consider the map
${\mathcal{J}}^{+}\to {\mathcal{J}}^{+}/{\mathcal{J}}^{2}$
. Since
${\mathcal{J}}^{+}/{\mathcal{J}}^{2}$
is abelian, we have a naturally defined determinant map
$${\bigcup}_{n\in \mathbb{N}}G{l}_{n}({\mathcal{J}}^{+}/{\mathcal{J}}^{2})\to ({\mathcal{J}}^{+}/{\mathcal{J}}^{2}{)}^{\times}={\mathbb{Z}}^{\times}\mathbb{\u22c9}\mathcal{J}/{\mathcal{J}}^{2}.$$
We get induced maps
$$det:{K}_{1}(\mathcal{J})\to {K}_{1}(\mathcal{J}/{\mathcal{J}}^{2})\to ker({\mathbb{Z}}^{\times}\mathbb{\u22c9}\mathcal{J}/{\mathcal{J}}^{2}\to {\mathbb{Z}}^{\times})=\mathcal{J}/{\mathcal{J}}^{2}$$
(we use here the identification of the additive group
$\mathcal{J}/{\mathcal{J}}^{2}$
with the multiplicative group
$\{1+aa\in \mathcal{J}/{\mathcal{J}}^{2}\}$
).
We want to show that
$det:{K}_{1}(\mathcal{J})\to \mathcal{J}/{\mathcal{J}}^{2}$
is an isomorphism. Note, that it is obviously surjective, since
$det\left({e}^{a}\right)=a+{\mathcal{J}}^{2}$
for any
$a\in \mathcal{J}$
.
Theorem 7.1.
Let
$\mathcal{J}$
be a harmonic Banach ideal. The natural determinant map yields an isomorphism
$${K}_{1}(\mathcal{J})\sim =\mathcal{J}/{\mathcal{J}}^{2}.$$
The preceding result implies that our diffotopy result does not extend to higher algebraic
$K$
theory. The first topological
$K$
theory of the Schatten ideals is wellknown to be zero. The proof of the preceding theorem is given after stating and proving a lemma.
Note that, every class
$\left[z\right]\in {K}_{1}(\mathcal{J})$
is represented by an element
$z\in G{l}_{n}\left({\mathcal{J}}^{+}\right)$
which maps to the identity under the canonical evaluation onto
$G{l}_{n}(\mathbb{Z})$
. Using the next lemma, we are able to show injectivity of the determinant map.
We now proceed with the proof of the theorem.
Lemma 7.2.

(a)
Every invertible element in
${M}_{n}(\mathcal{J}{)}^{+}$
which maps to
$1\in \mathbb{Z}$
is a product of exponentials of elements in
${M}_{n}(\mathcal{J})$
. An invertible element in ${M}_{n}({\mathcal{J}}^{2}{)}^{+}$ which is connected by a norm continuous path of elements, which are invertible in ${M}_{n}({\mathcal{J}}^{2}{)}^{+}$ , to $1\in {M}_{n}({\mathcal{J}}^{2}{)}^{+}$ is a product of exponentials of elements in ${M}_{n}\left({\mathcal{J}}^{2}\right)$ .
 (b) Let $g\in {M}_{n}\left({\mathcal{J}}^{2}\right)$ . The element ${e}^{g}$ is invertible in ${M}_{n}(\mathcal{J}{)}^{+}$ and $\left[{e}^{g}\right]=0$ in ${K}_{1}(\mathcal{J})$ .
 (c) Let $g,h\in {M}_{n}(\mathcal{J})$ . We have that $\left[{e}^{g+h}\right]=\left[{e}^{g}\right]+\left[{e}^{h}\right]$ in ${K}_{1}(\mathcal{J})$ .
 (d) Let $g\in {M}_{n}(\mathcal{J})$ . If $tr\left(g\right)\in {\mathcal{J}}^{2}$ , then $\left[{e}^{g}\right]=0$ in ${K}_{1}(\mathcal{J})$ .

Proof.
Since
${M}_{n}(\mathcal{J})$
is isomorphic to
$\mathcal{J}$
and
$({M}_{n}\mathcal{J}{)}^{2}={M}_{n}{\mathcal{J}}^{2}$
we can restrict our reasoning, for a proof of (a),(b) and (c), to the case
$n=1$
. (a) If $a\in {\mathcal{J}}^{+}$ is close to one, then the logarithmic series converges to an element in $\mathcal{J}$ . An easy calculation shows that the logarithm lies in ${\mathcal{J}}^{2}$ if $a\in ({\mathcal{J}}^{2}{)}^{+}$ (it can be written as a product of $a$ by an element in ${\mathcal{J}}^{+}$ ). The assertion follows by standard arguments using compactness and the fact that the group of invertible elements in ${\mathcal{J}}^{+}$ which map to $1\in \mathbb{Z}$ is connected.(b) The proof of this lemma follows the idea in [14] , 4.1, using a scheme going back to [15] . As in the proof of the homotopy invariance theorem we consider certain invertible elements which we want to represent by commutators. Consider ${M}_{2}(\mathcal{J})$ as acting on $H\oplus H$ and choose an isometry between the second copy of $H$ in this direct sum and ${\oplus}_{n\in \mathbb{N}}H$ . Under this identification, we define $${X}_{1}={e}^{g}\oplus {e}^{g/2}\oplus {e}^{g/2}\oplus {e}^{g/4}\oplus ...$$ and $${X}_{2}=1\oplus {e}^{g/2}\oplus {e}^{g/2}\oplus {e}^{g/4}\oplus ...$$ (again, with ${e}^{\pm g/{2}^{n}}$ term appearing ${2}^{n}$ times) can be considered as elements in ${M}_{2}(\mathcal{J})$ , since $\mathcal{J}$ is harmonic (compare to the proof of Theorem 4.2.1 ). Clearly ${X}_{1}{X}_{2}={e}^{g}\oplus 1$ . In order to show that $\left[{e}^{g}\right]=0$ , it suffices to show $[{e}^{g}\oplus 1]=0$ in ${K}_{1}(\mathcal{J})$ . We want to show that ${X}_{1}$ and ${X}_{2}$ are products of commutators. This implies that ${e}^{g}\oplus 1$ is also a product of commutators and hence finishes the proof.We first concentrate on ${X}_{1}$ . As before we construct matrices $${W}_{1}={e}^{g/2}\oplus {e}^{g/2}\oplus 1\oplus {e}^{g/8}...$$ and $${W}_{2}={e}^{g/2}\oplus 1\oplus {e}^{g/2}\oplus {e}^{g/8}...$$ which satisfy ${W}_{1}{W}_{2}={X}_{1}$ . There is an explicit isomorphism ${M}_{2}(\mathcal{J}{)}^{+}\sim ={M}_{3}(\mathcal{J}{)}^{+}$ (preserving the ideal ${\mathcal{J}}^{2}$ ) such that the matrix ${W}_{1}$ is mapped to a matrix of the form $\gamma \oplus {\gamma}^{1}\oplus 1$ . The element $h=1\gamma $ is in ${\mathcal{J}}^{2}$ (since the exponential series was already convergent in $({\mathcal{J}}^{2}{)}^{+}$ ) and therefore decomposes into $\alpha \beta $ with $\alpha ,\beta \in \mathcal{J}$ . To see this assume that $h={\sum}_{i\le n}{a}_{i}{b}_{i}$ with ${a}_{i}$ and ${b}_{i}$ in $\mathcal{J}$ . Then there are $A$ and $B$ in ${M}_{n}\mathcal{J}$ such that $h\oplus 0=AB$ . Identify then $\mathcal{J}$ with ${M}_{n}(\mathcal{J})$ using $AdV$ for a suitable isometry $V$ .By the proof of Vaserstein's lemma (or by direct computation), we have that $$\gamma \oplus {\gamma}^{1}\oplus 1=\left[\left(\begin{array}{ccc}\gamma & 0& 0\\ 0& 1& 0\\ \gamma \beta & 0& 1\end{array}\right),\left(\begin{array}{ccc}{\gamma}^{1}& 0& {\gamma}^{1}\alpha \\ 0& 1& 0\\ 0& 0& 1\end{array}\right)\right]\left[\left(\begin{array}{ccc}1& 0& 0\\ 0& {\gamma}^{1}& {\gamma}^{1}\alpha \\ 0& 0& 1\end{array}\right),\left(\begin{array}{ccc}1& 0& 0\\ 0& \gamma & 0\\ 0& \gamma \beta & 1\end{array}\right)\right]$$ where $[\cdot ,\cdot ]$ denotes a multiplicative commutator. Thus, ${W}_{1}$ is a product of two multiplicative commutators in ${M}_{2}(\mathcal{J}{)}^{+}$ . A similar reasoning applies to ${W}_{2}$ and matrices occurring in a similar decomposition of ${X}_{2}$ . Thus $\left[{e}^{g}\right]=0$ in ${K}_{1}(\mathcal{J})$ .(c) Since the leading terms in the series expansion of ${e}^{g+h}{e}^{g}{e}^{h}1$ vanish, this element can be written as a sum of four terms where each term is a product of ${g}^{2},gh,hg,{h}^{2}$ respectively, by an element in ${\mathcal{J}}^{+}$ . Therefore ${e}^{g+h}{e}^{g}{e}^{h}$ as well as its inverse lie in $({\mathcal{J}}^{2}{)}^{+}$ . By (a) ${e}^{g+h}{e}^{g}{e}^{h}$ is a product of exponentials of elements in $({\mathcal{J}}^{2}{)}^{+}$ . Using (b), this implies the claim.(d) A matrix with trace in ${\mathcal{J}}^{2}$ is a finite sum of
 $\bullet $ offdiagonal matrices with one entry in $\mathcal{J}$ ,
 $\bullet $ matrices of the form $a\otimes {e}_{ii}a\otimes {e}_{jj}$ with $a\in \mathcal{J}$ and
 $\bullet $ a matrix $a\otimes {e}_{11}$ with $a\in {\mathcal{J}}^{2}$ .
 Proof. Let $A$ be an invertible element in ${M}_{n}\left({\mathcal{J}}^{+}\right)$ which maps to the identity in ${M}_{n}(\mathbb{Z})$ and with determinant zero. By (a) of the preceding lemma it is of the form ${e}^{{h}_{1}}\cdots {e}^{{h}_{k}}$ and $detA={h}_{1}+\cdot \cdot \cdot +{h}_{k}+{\mathcal{J}}^{2}={\mathcal{J}}^{2}$ . The matrix ${e}^{{h}_{1}}\oplus \cdots \oplus {e}^{{h}_{n}}\in {M}_{nk}(\mathcal{J}{)}^{+}$ has the same class in ${K}_{1}(\mathcal{J})$ and the trace of its logarithm is just a lift of its determinant $detA$ to $\mathcal{J}$ and hence in ${\mathcal{J}}^{2}$ . By (d) of the preceding lemma the class in ${K}_{1}(\mathcal{J})$ is zero. This shows injectivity of the determinant map. Surjectivity was obvious, hence the assertion. □
Remark 7.0.1.
It is of course always true that
${K}_{1}(\mathcal{J})$
maps surjectively onto
$\mathcal{J}/[\mathcal{J},\mathcal{J}]$
,
$\mathcal{J}$
being a Banach ideal.
References
Joachim Cuntz, Mathematisches Institut, Einsteinstr.62, 48149 Munster, Germany Email address : cuntz@math.unimuenster.de URL : http://www.math.unimuenster.de/u/cuntz/cuntz Andreas Thom, SFB 478, Hittorfstr.27, 48149 Munster, Germany Email address : thoman@math.unimuenster.de URL : http://www.math.unimuenster.de/u/thoman
 H.Bass, Algebraic $K$ theory, W.A. Benjamin, New York (1968).
 B.Blackadar, $K$ theory for Operator Algebras, SpringerVerlag, Heidelberg / Berlin / New York / Tokyo (1986).
 B.Blackadar and J. Cuntz, Differential Banach algebra norms and smooth subalgebras of ${C}^{*}$ algebras, J. Operator Theory 26 (1991), 255282.
 A.Connes, Noncommutative Geometry, Academic Press (1994).
 G.Cortin͂as, Algebraic $K$ theory of locally convex algebras, in preparation
 J.Cuntz, A class of ${C}^{*}$ algebras generated by isometries, Commun. Math. Phys. 57 (1977), 173185.
 J.Cuntz, Ktheory and ${C}^{*}$ algebras, in ”Algebraic Ktheory, Number theory, Geometry and Analysis”, Springer Lecture Notes in Math. 1046, 5579.
 J.Cuntz, A new look at KKtheory, Ktheory 1 (1987), 3152.
 J.Cuntz, Bivariante $K$ theorie für lokalkonvexe Algebren und der bivariante ChernConnesCharakter, Docum. Math. J. DMV 2 (1997), 139182, http://www.mathematik.unibielefeld.de/documenta
 J.Cuntz, Bivariant $K$ theory and the Weyl algebra, $K$ theory (to appear).
 K.Dykema, T.Figiel, G.Weiss and M.Wodzicki, Commutator structure of operator ideals, Adv. in Mathematics 185 (2004), 179.
 L.Frerick and S.Shkarin, Completeness of certain function spaces, preprint.
 N.Higson, Algebraic $K$ theory of stable ${C}^{*}$ algebras. Adv. in Math. 67 (1988), no. 1, 140 pp.
 M.Karoubi, Homologie de groupes discrets associés à des algèbres d'opérateurs, J. Operator Theory 15 (1986), no. 1, 109–161.
 C.Pearcy and D.Topping, On commutators in ideals of compact operators. Michigan Math. J. 18 (1971) 247–252
 J.Rosenberg, Comparison between algebraic and topological $K$ theory for Banach algebras and ${C}^{*}$ algebras, preprint.
 A.Suslin and M.Wodzicki, Excision in algebraic $K$ theory, Ann. of Math. (2) 136 (1992), no. 1, 51–122.
 J.Tapia, $K$ théorie algébrique négative et $K$ théorie topologique de l'algèbre de Fréchet des opérateurs régularisants, Ann. Sci. Ecole Norm. Sup. (4) 30 (1997), no. 2, 241–277.
 A.Thom, Connective $E$ theory and bivariant homology, 2003, preprint
 F.Treves, Topological vector spaces, distributions and kernels, Academic Press, New York, London, 1967.
 M.Wodzicki, Algebraic $K$ theory and functional analysis, in: Proceedings of the First European Congress of Mathematics, Paris, Vol. II (1992), 485496, Progress in Mathematics vol. 120, Birkhäuser Verlag, Basel (1994).