MHB What is the solution to POTW #281 - Oct 30, 2018?

  • Thread starter Thread starter Euge
  • Start date Start date
Click For Summary
The discussion centers on proving that a complex Borel measure $\mu$ on the torus $\Bbb T^1$ has a Radon-Nikodym derivative with respect to the Lebesgue measure if its Fourier coefficients $\hat{\mu}(n)$ belong to $\ell^1(\Bbb Z)$. Participants are encouraged to explore the implications of the condition on the Fourier coefficients and its relationship to the measure's differentiability. Despite the complexity of the problem, no solutions were provided by the community this week. A solution is available for reference below the main discussion. The thread emphasizes the importance of understanding the connection between Fourier analysis and measure theory.
Euge
Gold Member
MHB
POTW Director
Messages
2,072
Reaction score
245
Here is this week's POTW:

-----
Given a complex Borel measure $\mu$ on the torus $\Bbb T^1$, define the Fourier coefficients of $\mu$ by $\hat{\mu}(n) := \int_{\Bbb T} e^{-2\pi i nx}\, d\mu(x)$, $n\in \Bbb Z$. Show that if the sequence $(\hat{\mu}(n))\in \ell^1(\Bbb Z)$, then $\mu$ has a Radon-Nikyodym derivative with respect to the Lebesgue measure on $\Bbb T$.

-----

Remember to read the https://mathhelpboards.com/showthread.php?772-Problem-of-the-Week-%28POTW%29-Procedure-and-Guidelines to find out how to https://mathhelpboards.com/forms.php?do=form&fid=2!
 
Physics news on Phys.org
No one answered this week’s problem. You can read my solution below.

Define
$$F(x) = \sum_{n \in \Bbb Z} \hat{\mu}(n) e^{2\pi i n x}\quad (x\in \Bbb T)$$
Since $\sum \lvert \hat{\mu}(n)\rvert < \infty$, by the Weierstrass $M$-test it follows that $F$ is a continuous function on $\Bbb T$, and
$$\hat{F}(m) = \sum_{n = -\infty}^\infty \hat{\mu}(n) \int_{\Bbb T} e^{2\pi i (n-m)x}\, dx = \hat{\mu}(m)$$By linearity, $$\int_{\Bbb T} g(x)F(x)\, dx = \int_{\Bbb T} g(x)\, d\mu(x)$$whenver $g$ is a trigonometric polynomial. By density of trig polynomials in $C(T)$, this integral equation holds whenever $g$ is continuous. Finally, since simple functions can be uniformly approximated by continuous functions, the integral equation holds whenever $g$ is a simple function. Therefore, for all measurable subsets $A\subset \Bbb T$, $$\mu(A) = \int_{\Bbb T} 1_A(x)\, d\mu(x) = \int_{\Bbb T} 1_A(x)F(x)\, dx = \int_A F(x)\, dx$$ This shows that $\dfrac{d\mu}{dm} = F$.