Facultad

2014-03-17

Michael Feischl. Universidad Técnica de Viena, Austria
Adaptiv algorithms for PDEs
Sala 2 de la Facultad de Matemáticas - 17:00 hrs.
Abstract:
Resumen:
Given a partial di_erential equation and a numerical algorithm to solve it, the overall goal is to compute a satisfying approximation of the unknown exact solution while investing a minimal amount of computational power and time. Usually, singularities in the solution and in the data spoil the performance of numerical algorithms, since they require a _ne resolution to get a decent approximation. In this context, adaptive algorithms are designed with the goal to distribute the available resources such that di_cult parts of the problem, i.e. singularities, are resolved in great detail, whereas easy parts of the problem, i.e., smooth solution and data, are treated only super_cially. We present a general approach to adaptivity and apply this approach to the concrete example of non-symmetr

2014-01-16

Manuel Cabezas. Impa, Rio de Janeiro
IMPA, Rio de Janeiro
en la sala 3 de la Facultad de Matematicas - 15:30 Hrs.
Abstract:
Resumen: Activated random walks (ARW) is an interacting particle system in which initially
there are particles distributed with density mu on the lattice Zd. These particles perform random walks and they
fall asleep at a rate lambda>0. The interaction ocurrs when an active particle visits a site occupied
by sleeping particle, instant in which the active particle wakes up the sleeping one.
It is conjectured that there is a critical value of the density mu=mu_c, such that for mu out in the sense that all particles fall asleep forever, while for large mu>mu_c there is an ongoing activity which never dies out.
In this talk we will be dicuss recent results about the behavior of the model at the critical value mu_c of the density.
Joint