In a two-dimensional domain Ω ⊂ R2, we consider the wave equation with variable velocity c(x1, x2) degenerating on the boundary Γ = ∂Ω as the square root of the distance to the boundary, and construct an asymptotic solution of the Cauchy problem with localized initial data. This problem is related to the so-called “run-up problem” in tsunami wave theory. One main idea (also used by the authors in earlier papers in the one-dimensional case and the two-dimensional case with c2(x1, x2) = x1) is that the (singular) curve Γ is a caustic of special type. We use this idea to introduce a generalization of the Maslov canonical operator covering the problem with degeneration and obtain efficient formulas for the asymptotic solutions.
1 March 2010 in Russian Journal of Mathematical Physics
In the paper, we consider the Cauchy problem for the inhomogeneous wave equation with variable velocity and with a perturbation in the form of a right-hand side localized in space (near the origin) and in time. In particular, this problem is connected with the question about the creation of tsunami and Rayleigh waves. Using abstract operator theory and in particular Maslov's noncommutative analysis, we show that the solution is separated into two parts: the transient one, which is localized in a neighborhood of the origin and decreases in time and the propagating one, which propagates in space like the wave created by the momentary “equivalent source.” We present several examples covering a wide range of perturbation resulting in rather explicit formulas expressing the solutions it terms of the error function of complex argument.
4 January 2007 in Mathematical and Computer Modelling
We explore the possibility of storage and retrieval of ultrametrically organized patterns in hippocampus, the part of the brain devoted to the memory processes. The ultrametric structure has been chosen for having a good representation of the categories of memory. The storage and retrieval process is the one typical of the hippocampus and it is based on the dynamic of the CA1 neurons under the input from the neurons of the Enthorinal cortex and the Ca3 system. We explore if this real system of neurons exhibits the property of associative memory introduced since a long time in the artificial neural networks. We study how the performance is dependent on the deviation of the system of patterns from ultrametricity. The evolution of the system is simulated by means of a parallel computer and the statistics of storage and retrieval is investigated.