According to [Maturana and Varela 1980], autopoiesis is at the basis of autonomy and cognition. Consequently, some artificial life works aimed at simulating artificial autopoiesis [Beer 2004, McMullin 2004]. Generally, the purpose is to maintain a topological organization of automata. At the opposite, we consider that simulation can override topological aspects because topology has no significance in the computational domain. More important is to preserve self-regeneration and resistance to disturbances of an Abstract Autopoietic Machine (AAM). The long term goal of our work is to address the elucidation of the Rosen conjecture - i.e. that a model of life defined as closure under efficient cause cannot be instantiated by a computer simulation - which is called into question by [Stewart and Miosso 2007] - which say that closure under efficient causation does not represent a theoretical problem for calculability - [Letelier et al. 2003] showed that autopoiesis is less general but very close to (M,R) System. Nevertheless, we can't claim that we simulate (M,R) systems because our model lack on a full loop allowing the use of produced data as an executive code.