Findings
Neighbor Selection: Less connected, but more functional
In this thread of works, we ask:
how the performance of a cooperative multi-agent network can be drammatically enhanced with relatively less communications, which can be very expensive (e.g., underwater communication)?
We have established the following results:
- Global network connectivity via local neighbor selection.
- Convergence rate of reduced network can be drammatically enhanced.
Magic Eigenvector
Relative Tempo Theorem (Linking Local Relative State and Fiedler Vector, Ref.2)
Consider the following ordinary differential equations on a simple graph \(\mathcal{G} = (\mathcal{V},\mathcal{E},A)\),
\[\dot{\boldsymbol{x}}_{i}(t)=-\sum_{j=1}^{|\mathcal{V}|}a_{ij}\left(\boldsymbol{x}_{i}(t)-\boldsymbol{x}_{j}(t)\right), i\in\mathcal{V},\]where \(\mathcal{V}=\{1,2,\cdots,n\}\), \(A=[a_{ij}] \in \mathbb{R}^{n \times n}\) is the adjacency matrix of \(\mathcal{G}\).
Then
\[\lim_{t \rightarrow \infty} \frac{\|\dot{\boldsymbol{x}}_{i}(t)\|}{\|\dot{\boldsymbol{x}}_{j}(t)\|} = \frac{\| [\boldsymbol{v}_{2}(L)]_i\|}{\| [\boldsymbol{v}_{2}(L)]_j\|},\]where \(L\) and \(\boldsymbol{v}_{2}(L)\) are the Laplacian and the Fiedler vector of \(\mathcal{G}\), respectively.
Related Papers
- Haibin Shao, Lulu Pan, Mehran Mesbahi, Yugeng Xi and Dewei Li. Relative tempo of distributed averaging on networks. Automatica, 2019, 105:159-166.
- Haibin Shao, Lulu Pan, Mehran Mesbahi, Yugeng Xi, Dewei Li. Distributed Neighbor Selection in Multi-agent Networks. arXiv preprint arXiv:2107.12022.
- Haibin Shao, Merhan Mesbahi, Dewei Li, and Yugeng Xi. Inferring centrality from network snapshots. Scientific Reports, 2017, 7(1):1-13.