Everywhere equivalent 3-braids (Q2339224)

A diagram \(D\) of a (knot) link \(L\) is called \textit{everywhere equivalent} if all diagrams \(D'\) obtained by changing exactly one crossing in \(D\) have the same (knot) link type. In the particular case, when each diagram obtained by changing one crossing represents the unknot, we have an everywhere trivial diagram. Previously, everywhere equivalent link diagrams have been studied by the author in [Asian J. Math. 17, No. 1, 95--138 (2013; Zbl 1293.57006)]. The purpose of the given paper is to classify all such diagrams which are closures of \(3\)-braids. The main result is Theorem 1.1 which gives a complete description of such diagrams. The proof of Theorem 1.1 is technically quite complicated and uses different combinatorial, knot theoretical, algebraic and computational tools. The proof starts with considering general properties of everywhere equivalent diagrams and can be divided into two parts. In the first part (the case of nonpositive braids), the author analyzes syllable types of \(3\)-braids and obtains restrictions on how syllables of a word are admissible. The next stage is verifying different cases of words of small length which allows to rule out some possible irrelevant syllable types. Finally, the global conditions allow to exclude all the remaining cases of (non)admissible words and complete the proof of the assertion for non positive braids. In the second part (the case of positive braids), the author introduces the notions of adequate and \(\psi\)-everywhere equivalent words (i.e. the words such that all their crossing-switched versions have the same trace of the Burau representation) and proves several auxiliary lemmas. The latter allows to use inductive arguments and reduce any admissible positive \(3\)-braid to one from the class that was managed before or exclude it from consideration. Finally completing the analysis of all possible cases, it is shown that the only cases of everywhere equivalent diagrams of positive braids are those which are given by Theorem 1.1.