Operations fitting triangular-norm-based biresiduation (Q1304216)

In classical (2-valued) logic, the truth value of each logical operation \noindent \(f(a_1,\ldots,a_n)\) does not change if we replace each value \(a_i\) with the equivalent value \(b_i\). In formal terms, in classical logic, we have \(((a_1\Leftrightarrow b_1)\&\ldots\& (a_n\Leftrightarrow b_n))\Rightarrow (f(a_1,\ldots,a_n)\Leftrightarrow f(b_1,\ldots,b_n))\). For fuzzy equivalence \(\Leftrightarrow\), this is not always true, because if \(a_1\) is ``almost'' equivalent to \(b_1\), etc., then these differences accumulate, and the resulting difference between \(f(a_1,\ldots,a_n)\) and \(f(b_1,\ldots,b_n)\) may become too large. In many cases, however, this implication becomes true if in the conjunction \((a_1\Leftrightarrow b_1)\&\ldots\), we repeat each term \(a_i\Leftrightarrow b_i\) several (\(k_i\)) times. An operation \(f\) for which this property is true for some values \(k_i\) is called fitting. In the paper under review, the authors describe which operations are fitting under different fuzzy equivalences. One of the most natural ways to define a fuzzy equivalence \(a\Leftrightarrow b\) is to define it via fuzzy implication, as \(\min(a\Rightarrow b,b\Rightarrow a)\), where, for a given t-norm \(\&\), \(a\Rightarrow b\) can be naturally defined as \(\sup\{z |x\& z\leq y\}\). This natural implication operation is called a residuation, and the resulting equivalent operation is called biresiduation. For several most widely used t-norms, the authors describe all unary and binary operations which are fitting w.r.t. corresponding biresiduations.