Chapter 2 Top Terms’ Relationship
2.3 Calculation of paths
2.3.3 Paths ϑ between adjacent pants
In this section we describe the holonomy of the path ϑ = ϑ(P, P0) between two
pairs of pants, sayP andP0, with initial point b=b(P)⊂∆0(P) and ending point b0 =b(P0) ⊂∆0(P0). This path, together with the pathsγ and υ, defines a set of
generators for the fundamental groupoidπ1,2k(Σ, B).
If ∂P is glued to ∂0P0, then there is an obvious path ϑ=ϑ(P, P0;, 0;σ) on Σ
from bto b0 crossing the pants curve σ=σi. As discussed in Section 2.2.2, we will
identify Σ with the quotient Sµ/∼ defined in Section 2.2.1. In particular, P ∪P0
is identified withStS0/∼, where S, S0 ⊂P are the truncated surfaces defined in Section 2.2.1 and where ∼ is the equivalence relation given by the attaching maps
along the annuliA⊂S and A0⊂S0.
We can describeϑby defining, first, a path inH and a path inH0, which project
to a path in StS0/∼, see Section 2.2.1 and 2.2.2.
Recall that
H∞=H∞(µ, ν) ={z∈H|=µ−ν
2 <=z < =
µ+ν
2 } ⊂H,
whereµ=µi ∈His the gluing parameter associated to the pants curve σ=σi and
1 1 0 F(b0) B0 b0 µ B0=F(B0) µ 1 2=F 21 F 11
Figure 2.8: The pathδ1 betweenb0 andB0, the pathδ2 =F(δ20)−1 betweenB0 and µ−B0 and the path F(δ1−1) betweenµ−B0 and F(b0), whereF =J−1Tµ−1, in the
caseH =H0 =H∞.
defined in Section 2.2.1. Recall also thatS⊂Pis the surfacePwith the projection of the horocyclic neighbourhood Ω−1
{z∈H|=z≥ =µ2+ν}of deleted.
Consider in H the straight lines δ1 = δ1,∞ between b0 = 1+i
√
3
2 and B0 =
1+i(=µ−2ν)
2 and δ2 = δ2,∞ between B0 and µ−B0 = µ−
1+i(=µ−2ν) 2
; see Fig- ure 2.8. If we concatenate these two paths, we have a pathδ∞=δ1·δ2 inHwhich
is contained in the complement of the open horoball {z ∈ H|=z > =µ2+ν} around
∞. Similarly, let alsoδ = Ω−1(δ∞) andδi, = Ω−1(δi) with= 0,1,∞andi= 1,2.
Define the pathδ00 and δi,0 0 inH0 in a similar way, where0 = 0,1,∞ andi= 1,2.
Let ζ: H−→ H/Γ and let ζ0: H−→ H/Γ. Then the path δ descends to a path
δ,S =ζ(δ)⊂S ⊂P and the path δ0 descends to a path δ00,S0 =ζ0(δ00) ⊂S0 ⊂P0.
Let alsoδi,,S =ζ(δi,) ⊂S ⊂P and δ0i,0,S0 =ζ0(δ0i,0) ⊂S0 ⊂P0 with , 0 = 0,1,∞
andi= 1,2.
Note that in the stripHtH0/∼, whereH=H and H0 =H0 we have that the
projection of the pathδ2and of the path (δ02)−1(that is,δ20 with reversed orientation)
coincide, see Section 2.2.2 for understanding the identification∼ in detail. In fact, we have that
as you can see from the fact that Ω−1
(B0) = Ω−01(µ−B0) and Ω−1(µ−B0) = Ω−01(B0) in HtH0/∼.
This tells us thatδ,S ∪(δ00,S0)−1 defines a path inStS0/∼.
With these definitions and using the notation of Section 2.2, we can see that
ϑ(P, P0;, 0;σ) is defined as the following union
ϑ(P, P0;, 0;σ) =δ,S∪(δ00,S0)−1⊂StS0/∼.
Unless needed for clarity, we refer to all these paths asϑ(P, P0) or ϑ(P, P0;, 0).
For finding the holonomy and the developing image of ϑ(P, P0), we need to glue
δ withF ◦(δ00)−1, where F = Ω−1J−1Tµ−1Ω0 is the transition function defined in
Section 2.2.2. See Sections 2.1.3.a and 2.1.3.b for the definition of the developing map and of the groupoid holonomy map. Figure 2.8 shows the developing im- age Devµ(ϑ(P, P0;∞,∞;σi)) of ϑ(P, P0;, 0;σi). Now referring to the gluing equa-
tion (2.3) and to the description of the groupoid holonomy map of Section 2.1.3.b, we see that the holonomy ofϑ(P, P0) is given by the following formula:
ρµ ϑ(P, P0;, 0;σ)
= Θ−→0 = Ω−1J−1Tµ−1Ω0, (2.4)
since the path ϑ(P, P0;, 0;σ) can be covered with only two charts and the only
transition map is the map Ω−1
J−1Tµ−1Ω0, whereµ=µi.
Remark 2.3.5. Note that if you look at Figure 2.8 as a picture of two superimposed
copies ofH, sayHandH0, as we did in Figure 2.4, then the point B
0 ∈H, which is
the starting point ofδ2, and the point µ−B0 =F(B0)∈H0, which is the starting
point ofF◦δ20 (whereF = Ω−1J−1Tµ−1Ω0), project to the same point inStS0/∼
under the two different chartsψ1 and ψ2 defined in Section 2.2.2 exactly when the
surfacesS andS0are glued by a Fenchel–Nielsen twistT wσi,<µi−1. This agrees with the description of the marking we made in Section 2.2.3. In fact, in that section we first defined the marking for the surface Σ(µ0), whereµ0 = (µ0
1, . . . , µ0ξ) is defined by
<µ0i = 1, for alli= 1, . . . , ξ, where the seams ofSand the seams ofS0match. Then,
for defining the marking on the general surface Σ(µ), we use a Fenchel–Nielsen twist T wσi,<µi−1 on the annulus aroundσi ∈ PC, for all i= 1, . . . , ξ.
As already noted in Lemma 2.2.3, the gluing parametersµare independent of the
direction of travel (fromP toP0 or vice versa). From (2.4) we have
ρµ ϑ(P0, P;0, )
= Ω−1
so that
ρµ(ϑ(P0, P;0, )−1) = Ω−1TµJΩ0.
Using the identitiesJ−1=−J,T−1
µ =T−µ and TµJ =JT−µ this gives
ρµ ϑ(P0, P;0, )−1
=−ρµ λ(P0, P;0, )
−1
, (2.5)
as one would expect. That fact will be particularly important for our proof in Appendix B.