Convolution of two distributions

65 Views Asked by At

Consider the convolution product:

$$H(x)\ast\operatorname{Pf}\dfrac{H(x)}{x},$$

where $\operatorname{Pf}$ denotes pseudo function. This means, that $\operatorname{Pf}\dfrac{H(x)}{x}$ is, as defined in the notes I'm reading, the following distribution:

$$\left(\operatorname{Pf}\frac{H(x)}{x},\phi\right)=\int_0^1 x^{-1}(\phi(x)-\phi(0))dx+\int_1^\infty x^{-1}\phi(x)dx.$$

In that case, I've tried to compute the convolution product directly by the definition:

$$\left(H(x)\ast \operatorname{Pf}\dfrac{H(x)}{x},\phi\right)=\left(H(x),\left(\operatorname{Pf}\dfrac{H(y)}{y},\phi(x+y)\right)\right).$$

For $x$ fixed, we have

$$\left(\operatorname{Pf}\dfrac{H(y)}{y},\phi(x+y)\right)=\int_0^1y^{-1}(\phi(x+y)-\phi(x))dy+\int_1^\infty y^{-1}\phi(x+y)dy.$$

Thus when we apply $H(x)$ we get

$$\left(H(x)\ast \operatorname{Pf}\dfrac{H(x)}{x},\phi\right)=\int_0^\infty \int_0^1y^{-1}(\phi(x+y)-\phi(x))dy dx+\int_0^\infty \int_1^\infty y^{-1}\phi(x+y)dy dx.$$

Now I don't know how to solve this. I've tried substitution it didn't get very far.

I know that the answer is $\ln x H(x)$, but I don't know how to get this.

How can I proceed my computation and show that this product is just $\ln x H(x)$?