关于p>1时L^p(T)空间傅里叶系数截断的存在性问题求助

Hey there! This is a classic result tied to Riesz projections and the Hardy space $H^p(T)$, so let's break it down step by step to make it accessible.

Background: What is the function $g$ we're targeting? #

The $g$ you're describing is the analytic projection of $f$ onto the Hardy space $H^p(T)$. Formally, $H^p(T)$ is the subspace of $L^p(T)$ where all negative-index Fourier coefficients are zero. Our goal is to show that for any $f\in L^p(T)$ with $p>1$, this projection exists and stays within $L^p(T)$.

Key Idea: Boundedness of the Riesz Projection #

The core of the proof relies on showing that the Riesz projection operator $P_+: L^p(T)\to H^p(T)$ is bounded when $1<p<\infty$. This operator is defined by:
$$(P_+f)(t) \sim \sum_{n=0}^\infty \hat{f}(n)e^{int}$$
Boundedness means there’s a constant $C_p>0$ (depending only on $p$) such that:
$$|P_+f|_p \leq C_p|f|p$$
If we can confirm this, then since $f\in L^p(T)$, $g=P+f$ will automatically be in $L^p(T)$ (bounded operators map $L^p$ functions to $L^p$ functions), and by definition, it matches $f$’s non-negative Fourier coefficients while setting negative ones to zero.

Proving Boundedness of $P_+$ #

Let’s split this into manageable cases:

Case 1: $p=2$ (Straightforward with Parseval’s Identity)

For $f\in L^2(T)$, Parseval’s identity tells us:
$$|f|2^2 = \sum{n=-\infty}^\infty |\hat{f}(n)|^2$$
For $g=P_+f$, we have:
$$|g|2^2 = \sum{n=0}^\infty |\hat{f}(n)|^2 \leq \sum_{n=-\infty}^\infty |\hat{f}(n)|^2 = |f|2^2$$
So $|P+f|_2 \leq |f|2$, meaning $P+$ is bounded on $L^2(T)$ with constant $C_2=1$.

Case 2: $1<p<\infty$ (Using Duality and Interpolation)

Let $q$ be the conjugate exponent of $p$ (so $\frac{1}{p} + \frac{1}{q} = 1$). We already know $P_+$ is bounded on $L^2(T)$. For $1<p<2$, $q>2$, and we can use duality to extend boundedness:

For any $f\in L^p(T)$ and $h\in L^q(T)$, the inner product satisfies:
$$|\langle P_+f, h\rangle| = |\langle f, P_+h\rangle| \leq |f|p \cdot |P+h|q$$
Since $P+$ is bounded on $L^q(T)$ (we can apply the same logic as the $L^2$ case extended via interpolation), this becomes:
$$|\langle P_+f, h\rangle| \leq |f|_p \cdot C_q|h|q$$
By the duality of $L^p$ and $L^q$, this implies $|P+f|_p \leq C_q|f|p$, so $P+$ is bounded on $L^p(T)$. For $2<p<\infty$, we reverse the roles of $p$ and $q$ to get the same result.

Why $p>1$ is Critical #

The Riesz projection is not bounded on $L^1(T)$. There exist $f\in L^1(T)$ where the series $\sum_{n=0}^\infty \hat{f}(n)e^{int}$ doesn’t converge in $L^1(T)$, meaning no such $g$ exists in $L^1(T)$. That’s why the proposition specifies $p>1$.

Quick Summary of the Proof #

1. Define the Riesz projection $P_+$ to extract the analytic part of $f$ (zero negative Fourier coefficients).
2. Show $P_+$ is bounded on $L^2(T)$ using Parseval’s identity.
3. Extend boundedness to all $1<p<\infty$ via duality and interpolation.
4. Boundedness ensures $g=P_+f\in L^p(T)$, giving the desired function.

If you want to dive deeper into any of these steps—like the interpolation theorem or singular integral representations of $P_+$—just let me know!

内容的提问来源于stack exchange，提问作者Not_a_topologist