The defence will be in French.
Abstract: Frustrated magnetic systems have been studied for many years, in particular using the Monte-Carlo method. In this thesis, we propose to solve frustrated classical models using tensor network methods. Firstly, by extending the transfer matrix method for Ising models with long-range interactions in the case of quasi-one-dimensional and hyperbolic systems. These results will be compared with experimental realizations of artificial spin systems to understand whether the latter are well described by a thermodynamical approach, or whether the model needs to be refined by including longer-range interactions, for example. We will then introduce a new method based on tensor networks to study classical Heisenberg models models in these same contexts of quasi-one-dimensional and hyperbolic systems.