[I found that my computations below are erroneous, namely $latex \mathrm{Cor} \langle f^{-1}\rangle \mathcal{F} \neq \langle \mathrm{CoCompl} f^{-1}\rangle \mathcal{F}$ in general (the equality holds when $latex \mathcal{F}$ is a set).]
During preparation my preprint of “Funcoids and Reloids” article I a little twisted on the problem of defining compact funcoids which was laying in my drafts.
I propose the following definition of compact funcoids: a funcoid $latex f$ is compact iff $latex \mathrm{im} f = \mathrm{im} \mathrm{Compl} f$ (in prose: the image of completion of the funcoid is the same as the image of the funcoid). I yet don’t know whether this is the best use of the word compact in the theory of funcoids. Maybe (for example) we are to call a funcoid $latex f$ compact if both $latex f$ and $latex f^{-1}$ conform to this formula.
Now I will show how I obtained this formula:
My first attempt to define compactness of funcoids was the following formula which generalizes compactness defined by an addict of fido7.ru.math Victor Petrov (sadly I don’t remember the exact formula given by Victor Petrov and cannot find the original Usenet post):
$latex \forall \mathcal{F} \in \mathfrak{F} : ( \langle f^{-1}\rangle \mathcal{F} \neq \emptyset \Rightarrow \exists \alpha : \{\alpha\} \subseteq \langle f^{-1}\rangle \mathcal{F})$.
Let’s equivalently transform this formula:
$latex \forall \mathcal{F} \in \mathfrak{F} : ( \langle f^{-1}\rangle \mathcal{F} \neq \emptyset \Rightarrow \mathrm{Cor} \langle f^{-1}\rangle \mathcal{F} \neq \emptyset)$;
$latex \forall \mathcal{F} \in \mathfrak{F} : ( \langle f^{-1}\rangle \mathcal{F} \neq \emptyset \Rightarrow \langle \mathrm{CoCompl} f^{-1}\rangle \mathcal{F} \neq \emptyset)$;
$latex \forall \mathcal{F} \in \mathfrak{F} : ( \mathcal{F} \cap^{\mathfrak{F}} \mathrm{dom} f^{-1}\neq \emptyset \Rightarrow \mathcal{F} \cap^{\mathfrak{F}} \mathrm{dom} \mathrm{CoCompl} f^{-1}\neq \emptyset)$;
$latex \mathrm{dom} f^{-1} = \mathrm{dom} \mathrm{CoCompl} f^{-1}$ what is finally equivalent to the formula from the above: $latex \mathrm{im} f = \mathrm{im} \mathrm{Compl} f$.
The most important thing we yet need to know about compact funcoids is the generalization of the theorem saying that to a compact topology corresponds unique uniformity. The generalization may sound like: to a compact funcoid corresponds a unique reloid. I have not yet found how to formulate exactly and prove this generalized theorem. Your comments are welcome.