Let $f$ be a locally integrable function on an open set $G$ of $\mathbb{R}^{n}$ and $ n\geq2 $. Suppose $\theta$ is in $ C^{\infty}(G) $ such that its laplacian $ \Delta \theta=1 $ everywhere in $\mathbb{R}^{n}$, and let $\phi\in C_{c}^{\infty}(G)$. Clearly $\phi \theta\in C_{c}^{\infty}(G).$ We know that the distributional laplacian of $f$ is defined and is given by $$ \langle \Delta f,(\phi \theta)\rangle=\int_{G}f(x)\Delta[\phi(x) \theta(x)]dx. $$
Now, if I write that this integral is equal to $$ \int_{G}f(x)\phi(x)dx+\int_{G}f(x)[\Delta\phi(x)] \theta(x)dx +\sum\limits^{n}_{j=1}\int_{G}f(x)\frac{\partial \phi(x)}{\partial x_{j}}\frac{\partial\theta(x) }{\partial x_{j}} dx,$$ I obtain absurd results by using some specific functions for $\phi$. There must be somthing wrong with the first integral (by the way I used here the fact that $\Delta\theta\equiv1$). What's wrong with that?