On the monoid of singular braids (Q1304852)

Let \(\mathbf{SB}_n\) denote the monoid of singular braids, i.e. the monoid generated by \(\sigma_1,\dots,\sigma_{n-1}\), \(\tau_1,\dots,\tau_{n-1}\), with relations \(\sigma_i \tau_i = \tau_i \sigma_i\) (\(1 \leq i \leq n-1\)), \(\sigma_i \sigma_j \sigma_i = \sigma_j \sigma_i \sigma_j\), \(\sigma_i \sigma_j \tau_i = \tau_j \sigma_i \sigma_j\) (\(|i-j|=1\)), \(\sigma_i \sigma_j = \sigma_j \sigma_i\), \(\sigma_i \tau_j = \tau_j \sigma_i\), \(\tau_i \tau_j = \tau_j \tau_i\) (\(|i-j|\geq 2\)), where in all cases \(1 \leq i , j \leq n-1\). This monoid contains the braid group \({\mathbf B}_n\) as its submonoid of units. A conjecture attributed to Birman states that the monoid homomorphism \(\eta\) from \(\mathbf{ SB}_n\) to the integral group ring \({\mathbb Z}{\mathbf B}_n\) of \({\mathbf B}_n\), defined by \(\eta(\sigma_i) = \sigma_i\) and \(\eta(\tau_i)= \sigma_i - \sigma_i^{-1}\), is injective. The main result in this paper is a proof of the validity of Birman's conjecture for the special case where \(n=3\) holds. This is achieved by considering the map \(\psi:\mathbf{SB}_n \to {\mathbb Z}{\mathbf B}_n\) defined by \(\psi(\sigma_i^{\pm}) = \sigma_i^{\pm}\) and \(\psi(\tau_i) = \sigma_i + \sigma_i^{-1}\), which is shown to be injective if and only if \(\eta\) is injective. In particular, a key result (Theorem 3) -- whose proof uses an extension of Artin's combing process -- states that if \(F,G \in \mathbf{SB}_n\) have all their double points on one strand and \(\psi(F) = \psi(G)\) holds, then \(F=G\) holds as well. This is subsequently used to give a combinatorial/geometrical argument involving double points for why the map \(\psi\) is injective in case \(n=3\) holds. Two corollaries of this result are then proven, which indicate where the problems may lie in proving Birman's conjecture in case \(n > 3\) holds.