Mountford, ThomasGauthier, Damien2006-09-202006-09-20200710.5075/epfl-thesis-3674https://infoscience.epfl.ch/handle/20.500.14299/234119urn:nbn:ch:bel-epfl-thesis3674-2We investigate the behavior of probabilities of large deviations above the mean versus large deviations below the mean for random additive functionals in a variety of random media models. Among others, we treat the large deviations of the point-to-hyperplane first-passage percolation on Zd and both the last-passage and the first-passage oriented percolation in Zd-1 × Z+. We also consider the large deviations of the parabolic Anderson model with white noise potential with respect to the Lyapunov exponent of the model.enLarge deviations, first-passage percolationlast-passage percolationoriented percolationparabolic Anderson modelGaussian processesLyapunov exponentsBorell’s inequalityblock argumentGrandes déviationspercolation de premier passagepercolation de dernier passagepercolation orientéemodèle d'Anderson paraboliqueprocessus gaussiensexposants de Lyapunovinégalité de Borellargument de type blockthesis::doctoral thesis