Equivalent statements for proving real parts of functions in disc algebra are uniformly dense in the space of real continuous functions

56 Views Asked by At

I am having trouble understanding why the following two statements are equivalent:

First we denote by $A$ the disc algebra consisting of complex continuous functions in the unit circle such that $\int_{-\pi}^{\pi}f(\theta)e^{in \theta}d\theta =0$ for all $n>0$
(Here $f$ is defined by identifying the continuous function on the circle with a continuous function on [-$\pi,\pi$].)

1) I want to prove that the real parts of functions in $A$ are uniformly dense in the space of real continuous functions on the circle and my book says this equivalent to proving the following:

2) If $\mu$ is a finite real baire measure on the unit circle ( $:=T$) such that $\int_Tfd\mu=0$ for all $f\in A$ then $\mu$ is the zero measure.

I don't understand why proving the second statement implies the first.

Any help will be greatly appreciated.

1

There are 1 best solutions below

5
On BEST ANSWER

You say you want to show (2) implies (1). The most reasonable interpretation is that you want to show (2) implies (1) for any collection $A$ of continuous, complex function on $T$ (if you were only speaking about the $A$ you defined at first, you can just prove (1) and/or (2)). So I will speak with $A$ in that level of generality.

It is obvious then that (2) does not imply (1). Indeed, let $A = \{ip(x) : x \in T\}$ for all polynomials $p \in \mathbb{R}[x]$. Then $\int_T fd\mu = 0$ for all $f \in A$ means $\int_T p(x)d\mu = 0$ for all $f \in A$, which means $\int_T f(x)d\mu = 0$ for any $f \in C(T)$, and so $\mu = 0$. However, $Re[ip(x)] = 0$ for all $ip(x) \in A$ and so the real parts of functions in $A$ are not at all uniformly dense in the set of real continuous functions on $T$.

The implication that you would want to prove is that (1) is implied by the variant of (2):

(2') If $\mu$ is a finite Baire measure on $T$ such that $\int_T Re[f]d\mu = 0$ for each $f \in A$, then $\mu \equiv 0$.

Here is a proof that (2') implies (1): Let $\tilde{A} = \{Re[f] : f \in A\}$. Suppose $\tilde{A}$ is not dense in $C(T)$. Then there is an element of $C(T)^*$ that is $0$ on $\tilde{A}$ but not identically $0$; as $C(T)^* = M(T)$, the finite Borel measures on $T$, we get a measure $\mu \not \equiv 0$ with $\int_T Re[f]d\mu = 0$ for all $f \in A$.