Pos(X) is defined as the set of non negative linear combinations of the elements of X.

L is a linear space, E is a linear subspace.
You can assume L is of finite dimension.

How do they deduce that 0 is in int(conv(A u E)) ? I don't understand their "therefore".