Pour $\phi \in \mathcal{D}(\mathbb{R})$ tel que $Supp(\phi) \subset [-R,R]$, pourquoi $\int_{|x| \geq \epsilon} \frac{\phi(x)}{x}dx = \int_{R \geq |x| \geq \epsilon} \frac{\phi(x) - \phi(0)}{x}dx$ ?
Asked
Active
Viewed 105 times
0
-
Je pense que l'intégrale de $1/x$ en valeur principale vaut $0$ car l'intégrale est prise symétriquement (par rapport à $0$) dans la définition de $\text{vp}$. – Evariste Jun 07 '21 at 21:23
-
Pourquoi c'est vrai ? En effet, si on teste l'intégrale de 1/x entre -1 et 1, l'intégrale diverge – Romain44 Jun 07 '21 at 21:36
-
Comme l'a dit Cameron, $x \mapsto 1/x$ est impaire, et donc en définissant l'intégrale symétriquement, i.e en prenant $\int_{-1}^1\frac{1}{x}\text{dx} = \lim_{\epsilon \rightarrow 0^+} \left( \int_{-1}^{-\epsilon} \frac{1}{x}\text{dx} + \int_{\epsilon}^1 \frac{1}{x}\text{dx} \right)$ on obtient $0$ car c'est $\lim_{\epsilon \rightarrow 0^+} 0$. Mais là tu ne passes même pas par $0$ dans ton intégrale donc le problème de la divergence ne se pose pas en réalité. – Evariste Jun 07 '21 at 21:49
-
We have the chance to communicate using mathematics. It does not matter if your English is not perfect (mine is far away to be good - I am French). So please, just by courtesy to other users, make a little effort. Thanks & cheers :-) – Claude Leibovici Jun 08 '21 at 01:50
1 Answers
2
(I'm not a French speaker, but I can interpolate when it comes to math, so hopefully I am addressing your question correctly.) There are two parts to your question. First is restricting the set over which you're integrating from $|x|\ge\varepsilon$ to $R\ge|x|\ge\varepsilon$. This is because $\phi$ is compactly supported on $[-R,R]$. Second is the introduction of $\phi(0)$. This is because
$$\int_{R\ge |x|\ge \varepsilon} \frac{\phi(0)}{x} \,dx = \phi(0) \int_{R\ge |x|\ge \varepsilon} \frac{1}{x}\,dx.$$
However $\int_{R\ge|x|\ge\varepsilon}\frac{1}{x}\,dx = 0$ since it is a bounded odd function integrated over a bounded symmetric set, so you have only introduced $0$ to the expression.
Cameron Williams
- 29,432
-
If someone wants to edit this to add a French translation if OP doesn't understand, please feel free to do so. I can slightly read French encased in math, but I couldn't possibly hope to write in it. – Cameron Williams Jun 07 '21 at 21:30
-
Pour la première partie, j'ai compris. En revanche pour la deuxième, je ne vois pas pourquoi c'est vrai car l'intégrale de 1/x entre -1 et 1, entre -2 et 2, entre -3 et 3...... diverge – Romain44 Jun 07 '21 at 21:33
-
You are right about that fact, but keep in mind that you are excluding a neighborhood of the origin. Your integration is over $[-R,-\varepsilon]\cup[\varepsilon,R]$, not $[-R,R]$, so $\frac{1}{x}$ is always well-defined here and has a finite integral (for $\varepsilon > 0$). – Cameron Williams Jun 07 '21 at 21:36
-
-