i would like to proof that a compact , connected and oriented $n$-manifold $M$ can't retract onto $\partial M$. My tool is the De Rham cohomology.
En effet, if there exists $f:M\to \partial M$ such that $f\circ \iota_{\partial M}=Id_{\partial M}$, then using the functoriality of cohomological map associated, $\iota^{*}_{\partial M}\circ f^{*}=Id_{H^{p}_{dR}(\partial M)}$, thus for each $p\geq 0$, the map $f^{*}:H^{p}_{dR}(\partial M)\to H^{p}_{dR}(M)$ is injective. So, i imagine there is a $p$ where the map injection fails, but with the hypotheses, with $p=n$, $f^{*}\equiv 0$, and i stopped here. Any tips?
First of all if you want to prove your states with the de Rham cohomology you get only that does not exist a smooth retract of $M$ to his boundary because you can use the de Rham cohomology only if your map is smooth.
If $H^{n-1}_{dr}(M)=0$ than the identity map for $p=n-1$ would be zero because $f^*$ is zero but is a contradiction because $\partial M$ is a compact without boundary (n-1)-manifolds and so by Poincarè duality you have that:
$H^{n-1}_{dr}(\partial M)=\mathbb{R}$