2

The real-valued version of Gelfand duality (ala Johnstone's "Stone Spaces") says every real Banach Algebra is equivalent to the algebra $C_b(Y,\mathbb R)$ of bounded real functions from some compact Hausdorff space $Y$.


Edit: No it doesn't. It only applies to real Banach Algebras with a compatible partial ordering.


It is well known that every complex Banach algebra is a real Banach algebra. Simply keep the same elements but forget about rescaling by complex scalars. Only allow rescaling by real scalars.

Denote by $C_b^{\mathbb R}(X,\mathbb C)$ this 'realisation' of the complex algebra $C_b(X,\mathbb C) $. The above says $C_b^{\mathbb R}(X,\mathbb C) \simeq C_b(Y,\mathbb R)$ for some space $Y$.

The space $Y$ is of course the set of $\mathbb R$-linear multiplictive functionals $C_b(X,\mathbb C) \to \mathbb R$. What do those functionals look like?

Suppose $C_b(X,\mathbb C) \to \mathbb R$ is one such functional. Letting $1: X \to \mathbb C$ be the function equal to $1$ everywhere consider the value of $\phi(i 1)$. We have $$-\phi(1) = \phi(-1) = \phi((i 1)^2)= \phi(i 1)^2 \ge 0$$

and so $$-\phi(1) \ge 0 \implies \phi(1) \le 0.$$ But also $$\phi(1)=\phi(1^2) = \phi(1)^2 \ge 0 $$ which implies $\phi(1) =0$. Thus for each $f$ we have $\phi(f) = \phi(1f) = \phi(1) \phi(f) =0$ and $\phi$ is trivial.

This suggests the space $Y$ is degenerate. But that would imply $C_b^{\mathbb R}(X,\mathbb C) \simeq \mathbb R$ which need not be true.

What is wrong with the above?

Daron
  • 11,639
  • In particular, every complex Banach algebra is a real algebra. how does this follow from the first paragraph? – SBF May 30 '24 at 10:23
  • @SBF It does not follow from the first paragraph. It's just a well known fact. – Daron May 30 '24 at 11:13
  • Are you sure that every real Banach algebra is equivalent to some $C_b(Y,\mathbb{R})$ algebra? For example, consider $\mathbb{C}$ as a real Banach algebra. Then $\mathbb{C}$ is two dimensional, so the only candidate for the space is $C_b({a,b},\mathbb{R})$. But these are not equivalent as Banach agebras. – KeeperOfSecrets May 30 '24 at 14:00
  • @KeeperOfSecrets I certainly am not any more! – Daron May 30 '24 at 14:47
  • 1
    @KeeperOfSecrets Okay I have read Johnstone in more detail. He requires the real algebra to also have a compatible partial order that does not exist for $\mathbb C$ treated as a real vector space. – Daron May 30 '24 at 15:20

0 Answers0