A biochemical network modeled by a multiset grammar may be investigated from a dynamical viewpoint by a linear recurrence system. This interesting connection between computation by a multiset grammar and a (network) recurrent dynamics poses a minimization problem, which turns out to be NP-hard.