I'm trying to understand this paper, and I see this integral (page 2):
$$ \int_{B\ \cap\ \mathcal{C}} (1 - y)dy, $$ where $\mathcal{C}$ is the curve given by $x = ye^{1-y}$ and $B$ can be any measurable set in $[0, 1]^2$.
This seems like a line integral, but the definition of a line integral that I know requires a differentiable parametrization of the curve. But
- The direction of the curve is not specified, so should the integral be positive or negative? (I know it should be positive, but what's the mathematical reason?)
- What if $B$ is such that $B \cap \mathcal{C}$ is not connected? No parametrization from an interval to $B \cap \mathcal{C}$ could be continuous, let alone differentiable.
I though of simply projecting $B \cap \mathcal{C}$ onto the $y$ axis and taking a one dimensional Lebesgue integral over the projection, but according to this question the projected set may not be measurable.
So what is the definition of an integral like this?