For Vlasov equation, we have $$\frac {\partial f}{\partial t}=-v\frac{\partial f}{\partial r}-a\frac{\partial f}{\partial v}.$$ This makes sense in the physical sense of the conservation law, which might be expressed as $$\frac{\partial (stuff)}{\partial{t}}=-\nabla \cdot{flux}.$$
However, mathematically speaking, if we forget all about the physical meaning of $f$ and just regard it as some function dependent on $\vec v$ and $\vec r$ which then depend on $t$ (reminiscent of how the Lagrangian relies on generalized coordinates and generalized velocity), we would find that, according to the chain rule for partial derivatives, $$\frac {\partial f}{\partial t}=\frac{\partial r}{\partial t}\cdot\frac{\partial f}{\partial r}+\frac{\partial v}{\partial t}\cdot \frac{\partial f}{\partial v}=v\frac{\partial f}{\partial r}+a\frac{\partial f}{\partial v}.$$ So completely opposite signs on the RHS? What happened?