Exercise 1.2.9

In the surface M g of genus g , let C be the circle that separates M g into two compact subsurfaces M h and M k obtained from the closed surfaces M h and M k by deleting an open disk from each. Show that M h does not retract onto its boundary circle C , and hence M g does not retract onto C . But show that M g does retract onto the nonseparating circle C in the figure.

Answers

Proof. Suppose C is a retract. Proposition 1.17 implies ι is injective. We claim ι ~ is also. Since ι is injective, ι ~ ( x ) = 0 for x π 1 ( C ) if and only if x is in the derived subgroup of π 1 ( M h ) . But this implies that x = 0 , for any element in the derived subgroup of π 1 ( M h ) is of the form ab a 1 b 1 , and the only element in the image of ι of this form is ι ( 0 ) = 1 , since π 1 ( C ) = . Thus, ι ~ = ι q is injective.

We claim this is a contradiction. We first draw the cell structure of M h :

Note that c is the generator of π 1 ( C ) . We see that the 2-cell is attached using the scheme c [ a 1 , b 1 ] [ a g , b g ] , and so the fundamental group is given by

π 1 ( M h ) = a 1 , , a g , b 1 , , b g , c | c i = 1 g [ a i , b i ]

which has the abelianization

H 1 ( M h ) = a 1 , , a g , b 1 , , b g , c | c = a 1 , , a g , b 1 , , b g ,

since i = 1 g [ a i , b i ] 0 under the abelianization map. This implies that ι ~ ( c ) = 0 , contradicting injectivity. Thus, M g does not retract onto C , for the restriction of this retraction to M h would then be a retract M h C .

We now claim that M g retracts onto the nonseparating circle C . We consider part of our cell structure in the following diagram:

where a i is identified with the circle C . We claim that we can extend a retraction projecting the top copy of a i to the bottom copy in the square on the right side can be extended to a retraction of the whole complex. This retracts the left dashed edge of the square into one point p , and the same applies to b i on the other side of this square. We can map the rest of the polygon outside of the right square onto p ; this is a well-defined map since every edge not equal to a i maps to p and the edge a i stays constant throughout our map. This is moreover a continuous map since an open subset U of a i has preimage of a union of open vertical slices of the right square if it does not contain p , and if it does, has preimage another union of open vertical slices unioned with the rest of the polygon on the left. □

User profile picture
2023-07-24 16:58
Comments