Menu:

Exploring soft graphs

por el

Dr. Jean-Guy Caputo

INSA Rouen Normandie


Resumen:
We are interested in graphs whose graph laplacian have an eigenvector
with a null component. These null components play an important role when
the graph Laplacian is used to describe miscible flows on a network.
On these null components, any action, control
or observation of the system is impossible; because they are so
important it is useful to detect them in general networks.

The graph laplacian is the matrix of node degrees minus the adjacency matrix : we refer to [3] for definitions and results on graph spectra. In a previous work [1] we called {\it soft node} a vertex corresponding to such a component. In the case of a multiple eigenvalue, any component of an eigenvector may be zero and we call {\it absolute soft node} a vertex with value zero for all eigenvector in the subspace.

Here we call $\lambda$-{\it soft graphs} graphs with a soft node for
an eigenvalue $\lambda$. We present a classification [2] of $\lambda$-soft
graphs containing up to 6 nodes, as well as some particular classes of graphs,
sorted by value of $\lambda$.
This shows a structure within each $\lambda$-soft set with
transformations connecting its members. This
suggests that we can build soft graphs and find eigenvectors with soft
nodes combinatorially.

Viernes 25 de agosto, 2017
5:00 p.m.
Salón 203, Edificio Anexo, IIMAS, UNAM

Informes: coloquiomym@gmail.com, o al 5622-3564.