I always saw the academic paper writing NMF (non-negative matrix factorization) formula with capital letters, even though, it means about the element. But for the constraint, they wrote in normal letter.
I wonder what is the accurate way. Since I try to write a paper for NMF with constraint, I would like to be sure I should go for all non-capital or all capital (or mix)?
Something like ..
\begin{align*} f^{*}(W,H,\pi,\lambda) &= D^{*}(W,H,\pi) + \Omega^{*}(H,\gamma)\\ &= \sum_{ij} (-v_{ij} \sum_{k} \pi_{ijk} log \frac{w_{ik}h_{kj}}{\pi_{ijk}} + \sum_{k}w_{ik}h_{kj}) + \sum_{kj} \frac{h_{kj}^2}{2\gamma} + \frac{\gamma}{2} \tag{2} \end{align*}
or
\begin{align*} f^{*}(W,H,\pi,\lambda) &= D^{*}(W,H,\pi) + \Omega^{*}(H,\gamma)\\ &= \sum_{ij} (-V_{ij} \sum_{k} \pi_{ijk} log \frac{W_{ik}H_{kj}}{\pi_{ijk}} + \sum_{k}W_{ik}H_{kj}) + \sum_{kj} \frac{H_{kj}^2}{2\gamma} + \frac{\gamma}{2} \tag{2} \end{align*}
I looked at the two papers and it seems that the authors use consistently the following rules: the entries of a matrix $A$ are denoted $A_{i,j}$, but every other real variable is represented by a lower case letter. I would suggest to follow the same rules.