This is an Exercise 3.15, pg 42.

  1. Show that the map between $\Bbb Z/2 \Bbb Z$- graded $C^*$-algebras $$C_0(\Bbb R) \rightarrow M_2(\Bbb C)$$ $$ \varphi: f \mapsto \begin{pmatrix} f(0) & 0 \\ 0 & f(0) \end{pmatrix} $$ is homotopic (through graded $*$-homomoprhisms) to the $0$ homomoprhisms.
  2. Whereas the $*$-homomorphism $$ \psi:f \mapsto \begin{pmatrix} f(0) & 0 \\ 0 & 0 \end{pmatrix} $$ is not null homotopic.

By definition in page 41, the even of $M_2(\Bbb C)$ consists of diagonal elemnts, the odd are those of off diagonals.

My questions would be: 1. How does one construction this homotopy? 2. How does one show the non-homotopy for second part?

  • $\begingroup$ Scaling is not a good idea here, the maps $H(\cdot,t)$ are no homomorphisms for $t\in (0,1)$. $\endgroup$ – MaoWao Mar 4 at 10:15
  • $\begingroup$ Ah, you mean, they don't respect multiplication? That is true. I will edit: I also notice my original grading seems to be wrong. $\endgroup$ – CL. Mar 4 at 10:46

Here is the construction for your first question.

Check that $$f\mapsto\begin{pmatrix}\frac{f(x)+f(-x)}2& \frac{f(x)-f(-x)}2\\ \frac{f(x)-f(-x)}2 &\frac{f(x)+f(-x)}2\end{pmatrix}$$ is a graded $*$ morphism, call it $\phi_x$. Now check that for each $f$ the map $$[0,1]\to M_2(\Bbb C), \qquad t\mapsto \begin{cases}\phi_{1-1/t}(f) & t>0\\ 0 & t=0\end{cases}$$ is continuous (the only difficulty is at $t=0$, but note that $f(1/t)\to 0$ as $t\to0$). This implies that this defines a homotopy on $\mathrm{Hom}_*^{\Bbb Z_2}(C_0(\Bbb R),M_2(\Bbb C))$ where this space has been given the strong operator topology. I assume that this is the topology you want, since it is impossible to have null-homotopic morphisms in the norm topology.

For the second part note that the images of $*$-morphism of $C_0(\Bbb R)$ in $M_2(\Bbb C)$ must be simultaneously diagonalisable. The projection onto the first and second eigenvalue (after choosing such a diagonalisation) will then be characters of $C_0(\Bbb R)$, these are always of the form $f\mapsto f(x)$.

Now if you want your mapping to respect the gradient, there are essentially three different possible scenarios. The first is that both characters are $f\mapsto f(0)$, second is that one is $f\mapsto f(0)$ and the other is the zero character and the third is that one character is $f\mapsto f(x)$ and the other is $f\mapsto f(-x)$. I did not think of a proof for this, but I don't believe it is difficult.

So any homotopy must be of the form $$t\times f\mapsto U(t)\begin{pmatrix} f(x_1(t)) &0 \\ 0 & f(x_2(t))\end{pmatrix} U(t)^*,$$ where $U(t)$ is unitary and $x_1(t), x_2(t)$ obey the above condition and $x_1(0)=0, x_2(0)=\infty$. But continuity will forbid you from changing this arrangement. Thus you cannot deform to zero.

  • $\begingroup$ Hi s.harp, in retrospect I don't really understand your second argument of, I don't really understand what is meant by characters, is there a reference for this? $\endgroup$ – CL. Apr 15 at 11:25
  • $\begingroup$ A character is a (non-zero) $*$-morphism into $\mathbb C$, alternatively a multiplicative probability state. The book by Murphy on C* algebras contains the definition. They are useful since for an abelian $C^*$ algebra $A$ you have $A\cong C_0(\sigma(A))$ where $\sigma(A)$ is the set of characters with the weak* topology. The details of this tell you that for $A=C_0(X)$ the only characters are the point evaluations $\psi_z: C_0(X)\to\Bbb C, f\mapsto f(z)$. $\endgroup$ – s.harp Apr 15 at 11:32
  • $\begingroup$ Ok, thanks, I will read your proof again! I wonder if you may have a look at my most recent question on Cayley Transform. $\endgroup$ – CL. Apr 15 at 11:39

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.