1
$\begingroup$

Let $A$ be a Banach algebra generated by $a$ which means it is a closure of $span \{1, a, a^2, ... \}$. The problem is to show that its Gelfand spectrum $\Omega(A)$ is homeomorphic to the spectrum of $a$ which is $\sigma_A(a) = \{ \lambda \in \mathbb{C} : a - \lambda \cdot 1 $ is not invertible$ \}$.

So far I can prove that $\sigma_A(a) = \hat a(\Omega (A))$ where $\hat a: \Omega(A) \to \mathbb{C} $ is a Gelfand transform of $a$. It remains to show that $\hat a: \Omega(A) \to \sigma_A(a)$ is actually homeomorphism.

$\endgroup$
2
$\begingroup$

Consider the Gelfand transform of $a$: \begin{align*}\hat{a}:\Omega (A)&\to \sigma_A(a) \\ h &\mapsto \hat{a}(h)=h(a) \end{align*} The map is well defined and surjective, because for any element of a commutative unital Banach algebra we have $$ \lambda \in \sigma_A(a)\iff \exists\;h\in \Omega(A):h(a-\lambda e)=0\iff \exists\;h\in \sigma_A(a):h(a)=\lambda$$ Moreover, since $A$ is generated by $a$, the map is also injective: indeed, since any $h\in \Omega(A)$ is a continuous algebraic homomorphism, the value of $h$ on $a$ uniquely determines $h$ on the closed subalgebra spanned by the element $a$, which by assumption is exactly $A$.

Finally, to show that it is a homeomorphism we only need to prove continuity, since $\Omega(A)$ is compact and $\sigma_A(a)$ is Hausdorff. But this is easy because $\Omega(A)$ has the weak*-topology, so if $\left\langle h_i\right\rangle_{i\in I}$ is a net that converges to $h$ in $\Omega(A)$ then $h_i(a)\to h(a)$ for all $a\in A$. (EDIT: This is also true on any commutative unital Banach algebra. It is the fact that the Gelfand transform of $a$ is also a bijection that allows us to conclude that it is in fact a homeomorphism).

$\endgroup$

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.