RU and \((U,N)\)-implications satisfying modus ponens (Q274463)

One of the most important classes of fuzzy implication functions are R-implications, generated from t-norms, and \((S,N)\)-implications, obtained from fuzzy negations and t-conorms. These families can be generalized for uninorms to RU-implications and \((U,N)\)-implications, respectively. In this paper, the authors investigate under which conditions implications from the both classes based on uninorms satisfy the modus ponens property, also called \(T\)-conditionality, i.e., under which conditions they satisfy the inequality \(T(x,I(x,y))\leq y\) for all \(x,y\in[0,1]\), where \(T\) is a continuous t-norm. For RU-implications, it is shown that \(T\)-conditionality depends only on the underlying t-norm \(T_U\) of the uninorm used to obtain the RU-implication. The authors analyze situations when this uninorm \(T_U\) is continuous and thus, taking into account well-known characterization of continuous t-norms, considering four possible cases: minimum t-norm, strict t-norms, nilpotent t-norms and ordinal sum t-norms based on continuous Archimedean t-norms. For \((U,N\))-implications the situation is different, the underlying t-norm \(T_U\) is never relevant and only the region out of this t-norm is important. In this case, the authors present a full characterization only if \(T\) is the minimum t-norm and some partial results for any other t-norm.