Bott and Tu compact cohomology of the circle "differential forms in Algebraic Topology"

308 Views Asked by At

On page 27 of that book, it is claimed that the inclusion map $\delta$ which maps a form from the non-empty intersection of two open covers of the circle to the disjoint union of those covers has a one dimensional kernel:

$$ \delta : H_c^1 (U\cap V) \rightarrow H_c^1 (U \sqcup V) $$ $$ w= (w_1 , w_2) \rightarrow (-(j_U)_* w , (j_V)_* w)$$

where inclusion simply extends the form by zero outside the intersection of U and V. Clearly, $(-(j_U)_* w , (j_V)_* w)$ is zero only if w is zero. So Kernel of $\delta$ is not one dimensional. Mistake? note the $U\cap V$ is two disjoint "arcs".

1

There are 1 best solutions below

5
On BEST ANSWER

It is not true that $(-(j_U)_* \omega , (j_V)_*\omega)=0$ iff $\omega=0$. Note that $(j_U)_*\omega$ and $(j_V)_*\omega$ here are cohomology classes in $H^1_c(U)$ and $H^1_c(V)$, not just $1$-forms. So we have to consider the possibility that they might be coboundaries. A $1$-form on $U$ is a coboundary iff its integral is $0$, and similarly for $V$. So $\ker(\delta)$ consists of exactly those $\omega$ such that $\int_{U\cap V}\omega=0$. Since there is an isomorphism $H^1_c(U\cap V)\to \mathbb{R}\oplus \mathbb{R}$ given by integration over each of the two components of $U\cap V$, $\ker(\delta)$ corresponds to the one-dimensional subspace $\{(x,-x):x\in\mathbb{R})\}\subset\mathbb{R}\oplus\mathbb{R}$.