This thesis investigates asymptotically efficient estimators in the linear Lyapunov causal model. The causal dependence in multidimensional data is controlled by the drift component of an Ornstein-Uhlenbeck process. The covariance matrix of the data is then given by the continuous Lyapunov equation. We consider two explicit examples of directed graphs that determine the sparsity of the drift component and evaluate classical estimators for the drift component, such as the plug-in, maximum likelihood, and Bayesian estimators, for asymptotic efficiency in the models mentioned. For the 2-path, estimators with asymptotic efficiency can be found for all approaches. This is confirmed by simulations. For the 3-path, this is not directly possible due to overdetermination of the covariance-drift component relationship. In this case, too, the theoretical results are supported by simulations.     «

This thesis investigates asymptotically efficient estimators in the linear Lyapunov causal model. The causal dependence in multidimensional data is controlled by the drift component of an Ornstein-Uhlenbeck process. The covariance matrix of the data is then given by the continuous Lyapunov equation. We consider two explicit examples of directed graphs that determine the sparsity of the drift component and evaluate classical estimators for the drift component, such as the plug-in, maximum likeli...     »