In the book Equilibrium Statistical Physics: Phases of Matter and Phase Transitions by Marc Baus and Carlos F. Tejero at page 49 (English version) says
3.2 Liouville's Equation
[...]
and hence $$ a(\textbf{r},t)\equiv \int \text{d}q\int\text{d}p a(q,p;\textbf{r},t)\rho(q,p),$$ may be written as $$ a(\textbf{r},t) = \int \text{d}q \int \text{d}p \left[ \text{e}^{-L_N t} a(q,p; \textbf{r},0)\right] \rho (q,p) \overset{?}{=} \int \text{d}q\int\text{d}p a(q,p;\textbf{r},0) \left[ \text{e}^{L_N t} \rho (q,p)\right], $$ where to go from the first to the second expression an integration by parts has been performed (since $L_N$ is a linear differential operator), and it has been admitted that $\rho(q,p)$ and all of its derivatives vanish at the limits of integration.
How can I go from the first to the second expression? I don't understand how the book has made the integration by parts.
*In particular, if $$ \int x y' =xy-\int x' y.$$ Who is who?
Just like this:
from Statistical Mechanics of Nonequilibrium Liquids by Evans and Morriss. (Page 48)