Timeline for Temporal generalization of graphs: density vs $n$ and $m$?
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 5, 2021 at 15:03 | history | bounty ended | Matthieu Latapy | ||
| Jan 29, 2021 at 20:49 | comment | added | Tiphaine | Yes, exactly ! :) I was thinking that you could somehow redefine $n$ as something linked to these simultaneous presence times, however, it seems complicated. You would have to count only the times at which $u$ interacts at least with $1, 2,...,k$ other nodes, is that right ? Your "number of nodes" (or rather, the number of possible outcomes in your probability would become something like $\int_t |V_t| \cdot |V_{t-1}|$. I'll see if I can have that expression as $k\cdot (k-1)$ for some $k$, I'll keep you updated. | |
| Jan 29, 2021 at 20:25 | comment | added | Matthieu Latapy | This is interesting! If I understand correctly, you somehow say that vertices would always be present, but at a rate between $0$ and $1$. Then, one may ask that an edge between two vertices at a given time is present with a rate bounded by the ones of these vertices. In a sense, in this way, we always have link streams (vertices are never really absent), which is helping. As you say, this still does not lead to a consistent relation to $n$, it seems, as it does not include the simultaneous presence constraint. Nice idea, though. | |
| Jan 29, 2021 at 20:18 | review | First posts | |||
| Jan 29, 2021 at 20:23 | |||||
| Jan 29, 2021 at 20:13 | history | answered | Tiphaine | CC BY-SA 4.0 |