In the paper Illustrating a problem in the self-financing condition in two 2010-2011 papers on funding, collateral and discounting (2012) (link) from Brigo, Buescu, Pallavicini and Liu, the above authors give a detailed description on how to work with the gain process, and how it differs from the price process.
Namely, given an asset $A$, we identify two processes with it, $P^{A}$ (price process) and $D^{A}$ (dividend process). The Gain process is then defined by $G^{A}=P^{A} + D^{A}$. Further ahead, they apply this definition to particular assets (I modified slightly their notation): $A_{1}$ a repo contract (with risky underlying asset having price $S_{t}$), $A_{2}$ a collateral position (with price being $C$) and $A_{3}$ a funding position, whose price is denoted by $\alpha$.
Now, the authors identify $P^{A_{1}} = 0$ (I guess here the repo rate chosen is such that the price of the contract is zero), $P^{A_{2}}=C$ and $P^{A_{3}}=\alpha$.
Then they claim: and the gain processes $dG^{A_{1}} = dS + (r_{D}-r_{R})Sdt$, $dG^{A_{2}}=r_{C}Cdt$, $dG^{A_{3}}=r_{F}\alpha dt$, where $r_{*}$ is some rate (not relevant for the question).
The question is, how do they get this? I thought, for example for $A_{3}$, that $dG^{A_{3}} = dP^{A_{3}}+dD^{A_{3}} = d\alpha + r_{F}\alpha dt$.
I hope my question is clear.