Abstract
Directed hypergraphs (DHs) capture higher-order, directed interactions. Uniformly sampled random DHs can serve as a null model: Structures that appear more often in your network than in the null model, could be significant. We introduce configuration models for DHs, in which the vertex and arc degrees are fixed, and the connections are randomized. We then sample from those with a Markov chain Monte Carlo method, involving arc swaps. For 8 types of DHs (with/without self-loops, degenerate arcs, double arcs), we (dis)prove uniformity of this sampling method.
Original language | English |
---|---|
Publication status | In preparation - 2023 |
Event | 21st INFORMS Applied Probability Society Conference 2023 - Centre de Congres Prouve, Nancy, France Duration: 28 Jun 2023 → 30 Jun 2023 Conference number: 21 https://informs-aps2023.event.univ-lorraine.fr/ |
Conference
Conference | 21st INFORMS Applied Probability Society Conference 2023 |
---|---|
Country/Territory | France |
City | Nancy |
Period | 28/06/23 → 30/06/23 |
Internet address |
Keywords
- Directed Hypergraphs
- Configuration Model
- Markov chain Monte Carlo
- Uniform sampling