scientific article; zbMATH DE number 3470314

From MaRDI portal
Publication:4054947





Consider a weakly-initial nondeterministic automaton \( B \) with a designated set of final states \( M \subseteq Z \), where \( Z \) is the set of states of \( B \). Let \( Z_{1} \subseteq Z \) be the set of all possible initial states, and \( \eta \) be a.specified cover of \( Z_{1} \). The external behaviour of \( B \) is given by the set \( h_{M}(z, p) \) consisting of all output words \( q \) produced by \( B \) when the input word \( p \) is applied to \( B \) in the initial state \( z \) and the final state reached by \( B \) is in \( M \). The ability of the experimenter to distinguish output words is described by a relation \( \bar{U} \) of indistinguishability (which is a compatibility relation). This relation holds for the output words \( q, q^{\prime} \) if and only if the experimenter is unable to distinguish between \( q \) and \( q^{\prime} \). The problem of initial state identification is considered. Such a problem is given by the triple \( [B, \bar{U}, \mathfrak N] \), and the information required is the statement that the initial state is a member of a certain set \( N \in \mathfrak N \). It is shown that such diagnosis problems can be equivalently transformed into diagnosis problems where the given automaton is observable and the relation \( \bar{U} \) is the identity (i.e. the analytic ability of the experimenter is no longer bounded). A method for solving such problems is given and the regularity of the experiment sets is proved under certain conditions.












This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4054947)