\(n\)-dimensional \((S,N)\)-implications (Q2206459)

Our knowledge about any domain contains a lot of expert statements. Experts are often not 100\% confident that their statements are always true. A natural way to gauge how confident we should be about a statement is to ask an expert to estimate the degree of confidence in this statement -- e.g., on a scale from 0 to 1. This is one of the main ideas behind the traditional \textit{fuzzy logic}. To make the result less subjective, a reasonable idea is to ask several experts. In this case, we get several degrees. Often, we have no reason to believe that some experts are more reliable than others. In this case, it does not matter which expert provided which degree: \((0.6,0.7)\) means the same as \((0.7,0.6)\). Thus, each tuple of degrees can be, e.g., sorted in increasing order, as \(0\le \mu_1\le \mu_2\le\ldots\le \mu_n\le 1\). The study of ordered tuples is known as \textit{\(n\)-dimensional fuzzy logic}. Several previous papers analyzed how ``and''-, ``or''-, and ``not''-operations of the traditional \([0,1]\)-based fuzzy logic can be extended to this \(n\)-dimensional case. It turns out that in some cases, the corresponding operations are component-wise (with order reversal in the case of negation). In most other cases, we can compose an \(n\)-dimensional operation based on \(n\) 1-dimensional ones -- provided that these operations satisfy the appropriate inequalities (so that the ordering is preserved). We can also show that for each \(n\)-dimensional operation \(f\), its 1-dimensional projections -- e.g., to tuples \((0,\ldots,0,\mu_i,1,\ldots,1)\) -- preserve the properties of the original \(n\)-dimensional operation \(f\). This paper continues these studies, but its main focus is on extending implication operations. A special emphasis is placed on \((S,N)\)-implication operations of the type \(S(b,N(a))\), where \(N(a)\) is a negation operation and \(S(a,b)\) is an ``or''-operation -- natural fuzzy analogues of the classical definition of implication \(a\to b\) as \(b\vee \neg a\).