Proof:

Let $ \lambda$ be a singular one cycle say $ \lambda = \sum {n_j}\gamma_j,$ where $ n_j\in \mathbb{Z}$ and $ \gamma_j:[0, 1]\longrightarrow X$. Since $ \partial\lambda = 0,$

$\displaystyle \sum_{j=1}^k n_j(\gamma_j(1) - \gamma_j(0)) = 0. \eqno(32.7)
$

The idea is to complete each of the paths $ \gamma_j$ into a loop at $ x_0$ by means of paths joining $ x_0$ to the ends $ \gamma_j(0)$ and $ \gamma_j(1)$. The only non-trivial part is the book-keeping which has to be done carefully. Let $ S$ denote the set of endpoints

$\displaystyle S = \{\gamma_j(1),\;\gamma_j(0)/\;j = 1, 2 \dots, k\}.
$

For each $ p\in S$, if $ m_p$ denotes the sum of the coefficients of $ p$ in (32.7) then $ m_p$ must be zero. Taking a path $ \beta_p$ in $ X$ joining $ x_0$ and $ p\in S$ we construct for each $ j$ a loop $ \eta_j$ in $ X$ based at $ x_0$ namely,

$\displaystyle \eta_j = \beta_{\gamma_j(0)}*\gamma_j * \beta^{-1}_{\gamma_j(1)}.
$

Finally

$\displaystyle \Pi_X(\eta_1^{n_1}*\eta_2^{n_2}*\cdots *\eta_k^{n_k}) = \sum_{j=1...
...amma_j
-\sum_{j=1}^k n_j(\beta_{\gamma_j(1)} - \beta_{\gamma_j(0)}) = \lambda
$

since

$\displaystyle \sum_{j=1}^k n_j(\beta_{\gamma_j(1)} - \beta_{\gamma_j(0)}) = \sum_{p\in S} m_p\beta_p = 0.
$



nisha 2012-03-20