In theory, variable and sequence declarations should be local to their statement. But in practice we often have the following situation (e.g. in MiKoMH/AI/source/game-play/mod/linear-evaluation.en.tex):
i.e. the weights $w_i$ and features $f_i$ are referenced outside the definition, namely in the sparagraph directly afterwards, and the problem could be solved by lifting the declarations ouf of the definition. But that would make the definition non-self-contained (which feels strange).
Here, the reference is relatively local. But we can have long-range referneces like the weights w_i from definition \sref{foo}. or even the m in the Einstain's famous $E=mcˆ2$.
Another situation with long-range reference is continuing (running) examples, as the one in MiKoMH/AI/source/ml/mod/attributes.en.tex, where we pick up the sequence variable \varX later in the section.
We need a good, semantic way of dealing with this.
In theory, variable and sequence declarations should be local to their statement. But in practice we often have the following situation (e.g. in MiKoMH/AI/source/game-play/mod/linear-evaluation.en.tex):
i.e. the weights $w_i$ and features $f_i$ are referenced outside the definition, namely in the sparagraph directly afterwards, and the problem could be solved by lifting the declarations ouf of the definition. But that would make the definition non-self-contained (which feels strange).
Here, the reference is relatively local. But we can have long-range referneces like
the weights w_i from definition \sref{foo}. or eventhe m in the Einstain's famous $E=mcˆ2$.Another situation with long-range reference is continuing (running) examples, as the one in
MiKoMH/AI/source/ml/mod/attributes.en.tex, where we pick up the sequence variable\varXlater in the section.We need a good, semantic way of dealing with this.