Definition 1.1

Let G 𝐺 G italic_G be a nonempty set together with three binary operations: normal-♯ \,\sharp\, , superscript normal-⋅ normal-⇀ \stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}} start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP and superscript normal-⋅ normal-↼ \stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}} start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP . The set G 𝐺 G italic_G is called a trisemigroup if the following two properties hold:

(i)

For each { , , } fragments fragments { , superscript , superscript } \ast\in\{\,\,\sharp\,,\,\stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}},\,% \stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}}\,\} ∗ ∈ { ♯ , start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP , start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP } , ( G , ) 𝐺 (G,\,\ast\,) ( italic_G , ∗ ) is a semigroup.

(ii)

The three binary operations \,\sharp\, , superscript \stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}} start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP and superscript \stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}} start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP satisfy the Hu-Liu triassociative law :

( x y ) z = x ( y z ) fragments fragments ( x superscript y ) z x fragments ( y superscript z ) (x\stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}}y)\,\sharp\,z=x\,\sharp\,(y% \stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}}z) ( italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP italic_y ) ♯ italic_z = italic_x ♯ ( italic_y start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP italic_z ) (1)
( x y ) z = x ( y z ) fragments fragments ( x y ) superscript z x fragments ( y superscript z ) (x\,\sharp\,y)\stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}}z=x\,\sharp\,(y% \stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}}z) ( italic_x ♯ italic_y ) start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP italic_z = italic_x ♯ ( italic_y start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP italic_z ) (2)
x ( y z ) = ( x y ) z fragments x superscript fragments ( y z ) fragments ( x superscript y ) z x\stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}}(y\,\sharp\,z)=(x\stackrel{{% \scriptstyle\leftharpoonup}}{{\cdot}}y)\,\sharp\,z italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP ( italic_y ♯ italic_z ) = ( italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP italic_y ) ♯ italic_z (3)
x ( y z ) = x y z superscript 𝑥 𝑦 𝑧 𝑥 superscript 𝑦 superscript 𝑧 x\stackrel{{\scriptstyle\rightharpoonup}}{{\cdot}}(y\,\sharp\,z)=x\stackrel{{% \scriptstyle\rightharpoonup}}{{\cdot}}y\stackrel{{\scriptstyle\rightharpoonup}% }{{\cdot}}z italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP ( italic_y ♯ italic_z ) = italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP italic_y start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ⇀ end_ARG end_RELOP italic_z (4)
( x y ) z = x y z superscript 𝑥 𝑦 𝑧 𝑥 superscript 𝑦 superscript 𝑧 (x\,\sharp\,y)\stackrel{{\scriptstyle\leftharpoonup}}{{\cdot}}z=x\stackrel{{% \scriptstyle\leftharpoonup}}{{\cdot}}y\stackrel{{\scriptstyle\leftharpoonup}}{% {\cdot}}z ( italic_x ♯ italic_y ) start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP italic_z = italic_x start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP italic_y start_RELOP SUPERSCRIPTOP start_ARG ⋅ end_ARG start_ARG ↼ end_ARG end_RELOP italic_z (5)

for all x , y , z G 𝑥 𝑦 𝑧 𝐺 x,y,z\in G italic_x , italic_y , italic_z ∈ italic_G .


Definition 3.19 .

We say that a monoid S 𝑆 S italic_S is 2-cancellative with respect to a generating set X 𝑋 X italic_X when it has cancellation on monomials of length 2 in the generators in the sense:

x y = x z y = z ; x z = y z x = y , for all x , y , z X . formulae-sequence 𝑥 𝑦 𝑥 𝑧 𝑦 𝑧 𝑥 𝑧 𝑦 𝑧 𝑥 𝑦 for all 𝑥 𝑦 𝑧 𝑋 xy=xz\Longrightarrow y=z;\quad xz=yz\Longrightarrow x=y,\quad\text{for all}% \quad x,y,z\in X. italic_x italic_y = italic_x italic_z ⟹ italic_y = italic_z ; italic_x italic_z = italic_y italic_z ⟹ italic_x = italic_y , for all italic_x , italic_y , italic_z ∈ italic_X .

Definition 9 .

For x x 0 𝑥 subscript 𝑥 0 x\geq x_{0} italic_x ≥ italic_x start_POSTSUBSCRIPT 0 end_POSTSUBSCRIPT (as just described) let σ x subscript 𝜎 𝑥 \sigma_{x} italic_σ start_POSTSUBSCRIPT italic_x end_POSTSUBSCRIPT be the unique solution for σ ( α , β ) 𝜎 𝛼 𝛽 \sigma\in(\alpha,\beta) italic_σ ∈ ( italic_α , italic_β ) to the equation

(11) 𝐚 ( σ ) + log x = 0 . 𝐚 𝜎 𝑥 0 \mathbf{a}(\sigma)+\log x\ =\ 0. bold_a ( italic_σ ) + roman_log italic_x = 0 .

Definition 2.3 .

Define the commutator [ x , y ] 𝑥 𝑦 [x,y] [ italic_x , italic_y ] and the associator ( x , y , z ) 𝑥 𝑦 𝑧 (x,y,z) ( italic_x , italic_y , italic_z ) by:

x y = y x [ x , y ] x y z = ( x y z ) ( x , y , z ) . formulae-sequence 𝑥 𝑦 𝑦 𝑥 𝑥 𝑦 𝑥 𝑦 𝑧 𝑥 𝑦 𝑧 𝑥 𝑦 𝑧 xy=yx\cdot[x,y]\qquad xy\cdot z=(x\cdot yz)\cdot(x,y,z)\ \ . italic_x italic_y = italic_y italic_x ⋅ [ italic_x , italic_y ] italic_x italic_y ⋅ italic_z = ( italic_x ⋅ italic_y italic_z ) ⋅ ( italic_x , italic_y , italic_z ) .

The associator subloop is A ( Q ) := ( x , y , z ) : x , y , z Q fragments A fragments normal-( Q normal-) assign fragments normal-⟨ fragments normal-( x normal-, y normal-, z normal-) normal-: x normal-, y normal-, z Q normal-⟩ A(Q):=\langle(x,y,z):x,y,z\in Q\rangle italic_A ( italic_Q ) := ⟨ ( italic_x , italic_y , italic_z ) : italic_x , italic_y , italic_z ∈ italic_Q ⟩ .

Definition 3.1 .

An element c 𝑐 c italic_c of a loop Q 𝑄 Q italic_Q is a weak inverse property (WIP) element iff for all x Q 𝑥 𝑄 x\in Q italic_x ∈ italic_Q ,

c ( x c ) ρ = x ρ ( c x ) λ c = x λ . formulae-sequence 𝑐 superscript 𝑥 𝑐 𝜌 superscript 𝑥 𝜌 superscript 𝑐 𝑥 𝜆 𝑐 superscript 𝑥 𝜆 c(xc)^{\rho}=x^{\rho}\qquad\qquad(cx)^{\lambda}c=x^{\lambda}\,. italic_c ( italic_x italic_c ) start_POSTSUPERSCRIPT italic_ρ end_POSTSUPERSCRIPT = italic_x start_POSTSUPERSCRIPT italic_ρ end_POSTSUPERSCRIPT ( italic_c italic_x ) start_POSTSUPERSCRIPT italic_λ end_POSTSUPERSCRIPT italic_c = italic_x start_POSTSUPERSCRIPT italic_λ end_POSTSUPERSCRIPT .

Let W ( Q ) 𝑊 𝑄 W(Q) italic_W ( italic_Q ) denote the set of all WIP elements of Q.


Definition A.14 .

Define d 𝑑 d italic_d on I × I 𝐼 𝐼 I\times I italic_I × italic_I by

d ( x , y ) = h ( x ) + h ( y ) - 2 λ ( x , y ) . 𝑑 𝑥 𝑦 𝑥 𝑦 2 𝜆 𝑥 𝑦 d\left(x,y\right)=h\left(x\right)+h\left(y\right)-2\lambda\left(x,y\right). italic_d ( italic_x , italic_y ) = italic_h ( italic_x ) + italic_h ( italic_y ) - 2 italic_λ ( italic_x , italic_y ) .