For a university project, I am currently implementing the RingCT Ring signatures as used in Monero (before the introduction of Bulletproofs) as laid out in this paper. To close the ring we calculate the last $s$ like this: (where $\alpha$ is the random value initially chosen and $l$ is the order of the elliptic curve group) $$s_j = α − c_j· x_j \bmod l $$
So as I understand the modulo in this case only applies to the $$c_j· x_j$$ otherwise we might get an out of range scalar, but then if this value is bigger then $\alpha$ it could happen that the $s$ is negative? Do I need another modulos here or do I understand something incorrectly?