Of course! For instance, that is exactly what is being done in discrete resonance area of nonlinear wave theory. Nodes represent wavevectors, each storing some energy, edges - the paths, along which energy flows from node to node, and time evolution of energy flow is governed by a system of differential equations, usually one equation per node.
You can consult the book of E.Kartashova for details. A short overview of the present state-of-the-art you find in the latest articles of A.Kartashov and E.Kartashova (or in reverse order).
Of course! For instance, that is exactly what is being done in discrete resonance area of nonlinear wave theory. Nodes represent wavevectors, each storing some energy, edges - the paths, along which energy flows from node to node, and time evolution of energy flow is governed by a system of differential equations, usually one equation per node.
You can consult the book of E.Kartashova for details. A short overview of the present state-of-the-art you find in the latest articles of A.Kartashov and E.Kartashova (or in reverse order).
There's a lot written on graphs and partial differential equations. Let's exploit the Poisson equation d*d phi = rho, or the same in classical language div grad phi = rho as an example. When this is solved in finite dimensional spaces with finite element kind of approaches, one creates a mesh. Let G stand for the incidence data between the nodes and edges. That for every edge e from node i to j, the entry G_ei is -1, G_ej is 1 and otherwise 0. Now, the "discrete counterpart" of the d*d operator (that is, what is written div grad in the classical language) is G^T M G, where G is the mass matrix of so called Whitney 1-forms called also by the name edge elements.
If you like to know more on these matters, check what Alain Bossavit has written about Whitney forms.