Let $D(M,N)$ be the set of all possible degrees of maps from $M$ to $N$, $M_1\#M_2$ the connected sum of $M_1$ and $M_2$.

  1. Can $D(M_1\#M_2,N)$ be calculated in terms of $D(M_1,N)$ and $D(M_2,N)$?
  2. Can $D(N,M_1\#M_2)$ be calculated in terms of $D(N,M_1)$ and $D(N,M_2)$?

Here all manifolds are assumed to be of the same dimension $d \ge 3$. I'm especially interested in the case $d=3$.

  • $\begingroup$ Have you given the question much thought? I think (2) can be readily answered by considering a few examples. $\endgroup$ – Ryan Budney Apr 13 '17 at 20:28
  • $\begingroup$ @RyanBudney I immediately see a subset relation for 1, but not the full answer. Question 2 looks rather tricky, unless you are implying the answer is "no". $\endgroup$ – Sebastian Goette Apr 14 '17 at 12:33

I think that looking at surfaces one can see that there is no nice relation between both sides. Let $\Sigma_g = \#^g S^1 \times S^1$ be the oriented closed surface of genus $g$. Then one can show, using simplical volumes aka the Gromov norm (see http://www.map.mpim-bonn.mpg.de/Simplicial_volume), which satisfies $||\Sigma_g|| = 4g-4$, that for $g,h >1$, we have $$D(\Sigma_g,\Sigma_h) = \left\{0,\pm 1, \dots, \pm \left\lfloor \frac{g-1}{h-1}\right\rfloor\right\}.$$ In particular we have $D(\Sigma_g,\Sigma_h) = \{0\}$ for $g < h$. I leave it to you to conclude from this that there is no nice relation (just choose $g$ and $h$ in the right way...).


Here's an elementary observation that gives a relation between the three sets in both cases.

Note that $M_1\# M_2$ admits degree one maps $\alpha : M_1\# M_2 \to M_1$ and $\beta : M_1\#M_2 \to M_2$ (just crush the other summand).

If $f : M_1 \to N$ has degree $d$, then $f\circ\alpha : M_1\# M_2 \to N$ also has degree $d$. Therefore $D(M_1, N) \subseteq D(M_1\# M_2, N)$ and likewise $D(M_2, N) \subseteq D(M_1\# M_2, N)$ so

$$D(M_1, N)\cup D(M_2, N) \subseteq D(M_1\#M_2, N).$$

If $g : N \to M_1\# M_2$ has degree $d$, then $\alpha\circ g : N \to M_1$ also has degree $d$. Therefore $D(N, M_1\# M_2) \subseteq D(N, M_1)$ and likewise $D(N, M_1\# M_2) \subseteq D(N, M_2)$ so

$$D(N, M_1\# M_2) \subseteq D(N, M_1)\cap D(N, M_2).$$

In general, the inclusions are strict as can be seen by using Jens Reinhold's answer: take $M_1 = M_2 = N = \Sigma_2$ for example.

There are also counterexamples in higher dimensions.

Let $d \geq 3$. If $M$ and $N$ are closed, connected, oriented manifolds of dimension $d$, then $\|M\# N\| = \|M\| + \|N\|$; note, this is not the case in dimension two (the torus has Gromov norm zero, but higher genus surfaces do not).

Suppose $M$ has dimension $d$ and $\|M\| \neq 0$. If $f : M \to M\# M$, then

$$\|M\| \geq |\deg f|\|M\# M\| = 2|\deg f|\|M\|$$

so $f$ has degree zero, i.e. $D(M, M\#M) = \{0\}$. Note however that $1 \in D(M\# M, M\#M)$ so

$$D(M, M\#M)\cup D(M, M\# M) \subsetneq D(M\#M, M\#M).$$

Also $1 \in D(M, M)$ so

$$D(M, M\#M) \subsetneq D(M, M)\cap D(M, M).$$

Gromov proved that the Gromov norm of a closed, oriented, hyperbolic $n$-manifold is a non-zero multiple (depending only on $n$) of its volume. In particular, such manifolds have non-zero Gromov norm so they can be used to construct counterexamples in any dimension greater than or equal to $2$.

These counterexamples seem to suggest that the answer to both questions of your questions is no.


One has $D(M_1\#M_2,N) \supseteq D(M_1,N)+D(M_2,N)$ (here, of course, I'm assuming $N,M_1,M_2$ are all connected and oriented as well as having the same dimension). One way of saying it is that there is an oriented $d+1$-cobordism $W$ from $M_1 \sqcup M_2$ to $M_1\#M_2$ by attaching a 1-handle to $(M_1\sqcup M_2)\times I$. Given maps $f_i: M_i\to N$ of degree $d_i$, there is no obstruction to extend $f_1\sqcup f_2$ to $F:W\to N$ since $N$ is connected. Then the degree of $F_{|M_1\#M_2}$ will be $d_1+d_2$.

In case $\pi_{d-1}(N)=0$, one will have $D(M_1\#M_2,N) = D(M_1,N)+D(M_2,N)$. The point is that a map $f:M_1\#M_2\to N$ will extend to a map $F:W\to N$, since one may attach the $d$-handle along the connect sum sphere $S^{d-1}\subset M_1\#M_2$.

For $d=3$, one has $\pi_2(N)\neq 0$ iff $N=S^1\times S^2$ or $N$ is a non-trivial connect sum $N=N_1\#\cdots \# N_k$, where $N_i$ is prime. In the first case, equality will hold again, because the separating connect-sum sphere $S^2\subset M_1\#M_2$ cannot map non-trivially to an essential 2-sphere in $S^2\times S^1$ since it is homologically trivial. So one may extend over $W$ again.


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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