$\DeclareMathOperator\MCG{MCG}$Consider the mapping class group $\MCG(\Sigma_2)$ of the closed genus 2 oriented surface $\Sigma_2$. The algebraic-duality theory of $\MCG_2:=\MCG(\Sigma_2)$ is explicitly described by Nathan Broaddus' very useful paper (arXiv link).

Broaddus constructs a beautiful homologically-nontrivial $2$-sphere $B$ in the curve complex of the genus two closed surface with marked curves α1, α2, α3, α4, α5, α6, β1, β2, β3 as labelled in the image below. ((we are indebted to N.Broaddus for his work and graphics.))

Question: I seek three distinct nontrivial elements $φ1,φ2,φ3$ of the mapping class group $\MCG(\Sigma_2)$ with the property that the following chain sum vanishes over $\mathbb{Z}/2$-coefficients:

$$α1 + α2 + α3 + α4 + α5 + α6 + β1 + β2 + β3$$

$$+ α1.φ1 + α2.φ1 + α3.φ1 + α4.φ1 + α5.φ1 + α6.φ1 + β1.φ1 + β2.φ1 + β3.φ1$$

$$+ α1.φ2 + α2.φ2 + α3.φ2 + α4.φ2 + α5.φ2 + α6.φ2 + β1.φ2 + β2.φ2 + β3.φ2$$

$$+α1.φ3 + α2.φ3 + α3.φ3 + α4.φ3 + α5.φ3 + α6.φ3 + β1.φ3 + β2.φ3 + β3.φ3$$

enter image description here

By "vanishing over $\mathbb{Z}/2$-coefficients" we mean something utterly trivial like $\alpha+\beta + \alpha +\beta = 2 \alpha + 2\beta = 0$ (mod 2). Of course if we took $\phi_1 = \phi_2 =\phi_3=id$ or $\phi_1=\phi_2$, $\phi_3=id$ then the chain sum would vanish (mod 2).

Possibly a suitable finite subgroup $G'$ of $\MCG_2$ will have the desired property. For the once-punctured torus $\Sigma_{1,1}$, indeed the order 3 finite subgroup of $PGL(\mathbb{Z}^2)$ "closes" the necessary (Steinberg) symbol.

Believe it or not, but finding such a triple of elements would yield a $\MCG$-equivariant codimension-two spine $Z \hookrightarrow \mathrm{Teich}(\Sigma_2)$ of the Teichmuller space.

The problem is related to stitching footballs from uniform hexagonal panels, or uniform pentagonal panels, or combinations of both. To stitch a football from panels $\{P_i| i\in I\}$ means finding a finite subset $I' \subset I$ for which the singular chain sum $\sum_{i\in I'} P_i$ has singular chain boundary which vanishes mod $2$, so $$\partial(\sum_{i\in I'} P_i)=\sum_{i\in I'} \partial P_i=0$$ over $\mathbb{Z}/2$-coefficients. When $P$ is two-dimensional hexagon or pentagon, the panels have singular boundary $$\partial P= \sum_{e\text{~edge~of~}P} e.$$ We denote the closed convex hull of the football $F:=conv\{P|\text{~panels}\}$. The panels then become closed subsets of the boundary $\partial F$. For instance since the 1960's, the standard football is stitched after Adidas' ``Telstar" design, having twenty white hexagon panels, and twelve black pentagon panels. But in our applications we assume the patches $\{P_i\}_I$ are pairwise isometric to some regular geodesically-flat polygon $P$.

enter image description here

enter image description here


1 Answer 1


I recently returned to this question, and have found a formal solution to Closing the Steinberg symbol in genus two using Mark C. Bell's wonderful curver program.

In low genus it was expected that an interesting finite order subgroup could yield formal solutions. This is the case in $PGL(\mathbb{Z}^2)$. For the mapping class group of genus two, there is an interesting order five subgroup $I=\{Id, \mu, \mu^2, \mu^3, \mu^4\}$ which formally closes the Steinberg symbol $\xi:=[\alpha_1]+[\alpha_2]+[\alpha_3]+[\alpha_4]+[\alpha_5]+[\alpha_6]$. (My notation follows the above image of Broaddus' two sphere). To formally close the specific original Steinberg symbol $\xi'$ given by Broaddus' two-sphere would require finding an appropriate conjugate $x\mu x^{-1}$ for $x\in MCG$. For the applications this is not essential.

We used a jupyter notebook to import Bell's curver program and are strictly studying the multiplicative adjoint action in $MCG$. https://github.com/jhmartel/MCG/blob/master/ClosingSteinbergGenusTwo.ipynb

In python code, the "closing condition $d\xi=0$ mod 2" for the symbol $\xi$ is equivalent to the vanishing of an iterated symmetric difference.

%% Now the next step has always been to use the formal solution to define a convenient chain model $X=\underline{F}=\sum_{\phi \in MCG} \phi.F$, where $F$ is a compact convex "horospherically truncated convex hull" defined over the ``vertices" $I.\xi$. My claim has always been that the singularities constructed from "repulsive" optimal transport programs are excellent candidate equivariant spines of $MCG$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.