We extend "stable marriage problem" studied via computer simulation (agent-based) from network perspective by dynamical generalization. We investigate how preferential or random attachment of individuals can affect simulation results such as: proportion of singles in society, number of divorces etc. We took into account attractiveness of individuals and its different types as well as personal taste. Current value of socio-economic pressure p (main model parameter) drives the dynamic of first marriage, remarriage or spontaneously marriage breaks up. Model reflects the behavior of the simplified heterosexual population (frequency of changing partners, the ratio of singles in society). Theoretical agent-based simulation (with populational approaches, e.g. Births and deaths) should be later supplemented by historical values of divorces/marriages in different countries of the world. Stable (constant) society was also implemented to show difference with "living" society. In this model, agents have attribute of attraction. Preferential attachment, known from network theory, was introduced, to mimic selection process. Additionally, totally random attachment (not attraction-depended) was also implemented for contrast.