2
$\begingroup$

It is Higon's notes on Index Theory, Ex 3.16, pg 43 where he describes a group structure (where I suppose works in more general context) of the graded $*$-algebra homotopy classes of map $$ [S,K(H)]$$

Here $S=C_0(\Bbb R)$ is the graded $*$-algebra given by even and odd functions as grading.

$K(H)$ is the $C^*$ algebra of operators on a graded Hilbert space $H=H_0 \oplus H_1$ whose homogenous subspaces are both infinite dimensional.


He claims this structure is given by the map

$$K(H) \oplus K(H) \rightarrow K(H \oplus H) \cong K(H)$$

associated to a graded unitary isomorphism $$H \oplus H \cong H \quad (*)$$


My questions:

(0) Previously in the notes it is written that $K(H)$ is the space of compact operators. With the same notation, it is defined as space of graded $*$ operators of $H=H_0 \oplus H_1$. I believe compact condition is still required? Where is it used?

(i) how is this independent of choice of the a graded unitary isomoprhism at $(*)$

(ii) How does such an isomoprhism in $(*)$ exist?

(iii) What's wrong with simply defining $(f+g)_x:= f_x+g_x$ point wise?


@MaoWao, In retrospect I still don't understand how two maps compose, Suppose $$ f \mapsto \begin{pmatrix} f_1 & 0 \\ 0 & f_2 \end{pmatrix}, g \mapsto \begin{pmatrix} g_1 & 0 \\ 0 & g_2 \end{pmatrix} $$ where $f_1, f_2 , g_1, g_2 \in K(H)$ , can we define addition by $$ f_1 +g_2 = \begin{pmatrix} f_1 & \\ & f_2 \\ && g_1 \\ &&& g_2 \end{pmatrix} $$ simply regarding it as a larger matrix, since we regard $K(H) = M_\infty(\Bbb C)$.

$\endgroup$
  • $\begingroup$ Note that $K(H)$ is a vectorspace and contractible. Thus $[X,K(H)]$ is trivial if you mean this to just be continuous maps up to homotopy equivalence. $\endgroup$ – s.harp Mar 4 at 15:40
  • $\begingroup$ Then from $C_0(X)$ to $K(H)$? $\endgroup$ – s.harp Mar 4 at 16:08
  • $\begingroup$ Ok, so i tried to make the question more general, let me edit to the original case. Thanks a lot for pointing out. $\endgroup$ – CL. Mar 4 at 16:17
3
$\begingroup$

(0) You could do the same for all bounded linear operators instead of the compact ones, but compact operators are much more relevant to $K$-theory (or rather $M_\infty(\mathbb{C})=\bigcup_n M_n(\mathbb{C})$, but this is not even a $C^\ast$-algebra, so one considers its completion $K(\ell^2)$ instead).

(i) If $u\colon H\to H$ is a graded unitary isomorphism, then the $\ast$-homomorphism $\alpha $ and $u^\ast \alpha(\cdot)u$ are homotopic through the $\ast$-homorphisms $u_t^\ast \alpha(\cdot)u_t$, where $(u_t)$ is a continuous path of graded unitaries connecting $u$ and $1$. You can use this to see that different graded unitaries in $(\ast)$ give rise to homotopic $\ast$-homomorphisms in the construction of the group structure.

(ii) This is "just" arithmetic with cardinal numbers. Remember that a unitary isomorphism of Hilbert spaces is just a bijection between fixed orthonormal bases. Thus, if $(e_i)_{i\in I}$ is an orthonormal basis of $H_0$, you need to find a bijection $I\to I\sqcup I$, and the same for $H_1$. This is possible because $I$ is assumed to be infinite. In the most important case when $H_0, H_1$ are separable, you can for example split the natural numbers into even and odd ones.

(iii) The sum of $\ast$-homomorphisms is usually not an $\ast$-homomorphism (you break multiplicativity).

$\endgroup$
  • $\begingroup$ Hi MaoWao, in retrospect, I sitll don't understand hwo the group structure, may you spell out an example? Thank you so much. I would be happy to give a bounty. $\endgroup$ – CL. Apr 14 at 20:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.