I was trying to understand this interesting question by example.

Let me follow their previous discussion and ask: Let a generic nontrivial 2-cocycle $\omega_2^G(g_1,g_2) \in H^2(G,\mathbb{R}/\mathbb{Z})$ in the cohomology group of $G$ with $U(1)=\mathbb{R}/\mathbb{Z}$ coefficient. In otherwords, here the 2-cocycle $\omega_2^G$ is a complex $U(1)=\mathbb{R}/\mathbb{Z}$ function with the norm $|\omega_2^G|=1$ but with a $U(1)$ complex phase satisfying the cocycle condition.

I like to ask how can we trivialize the 2-cocycle $\omega_2(g_1,g_2)$ of $G$ into 2-coboundary if we lift $G$ into a larger group $J$, and given that we know the group homomorphism $r$: $$J \overset{r}{\rightarrow} G.$$ In particular I like to focus on:

$$SU(2) \overset{r}{\rightarrow} SO(3).$$

So that
$$\omega_2^J(j_1,j_2)=\omega_2^G(r(j_1),r(j_2))=\omega_2^G(g_1,g_2) \text{ is trivial in } H^2(J,U(1)).$$ Namely $\omega_2^G(r(j_1),r(j_2))$ becomes a 2-coboundary in $H^2(J,\mathbb{R}/\mathbb{Z})$ for the cohomology group of $J$, but $\omega_2^G(g_1,g_2)$ originally was not a 2-coboundary but was a 2-cocycle for the cohomology group of $G$. We can explicitly write $$ \omega_2^G(g_1,g_2)=\omega_2^G(r(j_1),r(j_2))= \frac{\beta_1^J(j_2)\beta_1^J(j_1)}{\beta_1^J(j_1 j_2)}. $$ Here $\beta_1^J(j_1)$ is a 1-cochain for $j_1, j_2 \in J$, and that $g_1=r(j_1)$, $g_2=r(j_2)\in G$.

Is that true that the 2-cocycle in $$H^2[SO(3),\mathbb{R}/\mathbb{Z}]=\mathbb{Z}/2\mathbb{Z}$$ for $G=SO(3)$ can be trivialize in $J=SU(2)$? How can it be shown? Here we consider the cohomology group of the Lie group $SO(3)$ and $SU(2)$.

  • 1
    $\begingroup$ What sort of cohomology are you taking? Do you mean continuous (or smooth) cohomology, or the group cohomology of the underlying group? $\endgroup$ Nov 4, 2016 at 12:52
  • $\begingroup$ The question you link to seems to consider arbitrary discrete groups, whereas you have Lie groups. This is kinda different! $\endgroup$ Nov 4, 2016 at 12:53
  • $\begingroup$ thanks, I mean the "group cohomology of the underlying group." $\endgroup$
    – miss-tery
    Nov 4, 2016 at 14:02
  • $\begingroup$ Isn't it the case that the extension $$1\to \mathbb{Z}/2\to SU(2)\to SO(3)\to 1$$ corresponds to the nontrivial element in $H^2(SO(3),\mathbb{R}/\mathbb{Z})$? If so, it is more or less tautological that the cocycle trivializes, and it is also quite clear how to write the cochain $\beta$. $\endgroup$
    – Ehud Meir
    Nov 12, 2016 at 12:34
  • $\begingroup$ Does your example only work in $H^2$ or does it work for $H^n$ for other $n$? If it is clear, in either cases, you can write it as an answer even if it is trivial to you, it is non trivial to me still! Thanks. $\endgroup$
    – miss-tery
    Nov 12, 2016 at 20:53

1 Answer 1


Let $G$ be a group, and let $$1\to A\to J\to G\to 1$$ be an extension of groups with an abelian kernel. Choose a set-theoretical lifting $s:G\to J$ of the quotient map $p:J\to G$. Now define a function $\beta:G^2\to A$ by the formula $$\alpha(g,h) = s(g)s(h)s(gh)^{-1}.$$ This formula defines a two cocycle, and this is the two cocycle which corresponds to the above extension. Let us now show that the inflation of $\alpha$ to $J$ is trivial: define $\beta(j) = j\cdot((sp)(j))^{-1}\in A$. Then a direct calculation shows that $\delta\beta=inf(\alpha)$.

  • $\begingroup$ how would you express the 2-cocycle for $H^2[SO(3),R/Z]$? $\endgroup$
    – miss-tery
    Nov 14, 2016 at 0:54
  • $\begingroup$ What is $sp$ in your expression? $\endgroup$
    – miss-tery
    Dec 2, 2016 at 3:00
  • $\begingroup$ I am not sure whether you can explain this $$\beta(j) = j\cdot((sp)(j))^{-1}\in A$$, your notation confuses me -- but I will accept it as an answer -- I hope you can fill in sometime. $\endgroup$
    – miss-tery
    Dec 2, 2016 at 3:01
  • 1
    $\begingroup$ what I mean was the following: $sp$ was just the concatenation of the functions s and p. The inverse is the inverse in the group $J$. $\endgroup$
    – Ehud Meir
    Dec 2, 2016 at 10:48
  • 2
    $\begingroup$ A related but harder challenging question: mathoverflow.net/questions/265953 $\endgroup$
    – wonderich
    Mar 30, 2017 at 21:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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