I'm trying to understand the proof of the Lefschetz hyperplane theorem in Milnor's book "Morse Theory", page 41 but I can't understand his use of Lefschetz duality.

At this point it has been proven that a non singular affine algebraic variety of complex dimension $k$ has the homotopy type of a CW-complex of dimension less than or equal to $k$. We wish to prove the following:

Let $V$ be an algebraic variety of complex dimension $k$ lying in complex projective space $\mathbb{C} P^n$. Let $P$ be a hyperplane of $\mathbb{C} P^n$ containing all singularities of of $V$. Then the inclusion map $V\cap P \to V$ induces isomorphisms in singular homology in dimensions strictly less than $k-1$. In dimension $k-1$ the induced homomorphism is onto.

The proof goes like this:

Consider the long exact sequence of the pair $(V, V\cap P)$: $$\dots \to H_i (V\cap P) \to H_i (V) \to H_i (V, V \cap P) \to H_{i-1} (V) \to \dots$$ It is sufficient to show that $H_i (V, V \cap P) = 0$ for $i \leq k-1$. By Lefschetz Duality $$H_i(V, V\cap P; \mathbb{Z}) \cong H^{2k-i}(V - (V\cap P);\mathbb{Z})$$ But $V - (V\cap P)$ is a Zariski closed set in the affine space $\mathbb{C} P ^n - P$ and is nonsingular since $P$ contains the singularities of $V$. Hence $V - (V\cap P)$ is a nonsingular affine algebraic variety and so applying our theorem we find $H^{2k-i}(V - (V\cap P);\mathbb{Z})=0$ for $i \leq k-1$.

My confusion is over the application of Lefschetz duality. I was familiar with Poincare duality but not Lefschetz duality so I had to look it up. The statement seems simple enough but I don't understand how what Milnor is saying follows from Lefschetz duality.


First note that complex manifolds are always orientable. Therefore the Poincaré-Alexander-Lefschetz Duality theorem as stated in "Topology and Geometry" by Glen E. Bredon (Chapter VI, Theorem 8.3) says in this case:

If $M$ is a complex manifold of real dimension $2k$ and $L \subset K$ are compact subsets of $M$, then there is an isomorphism $$H^{2k-r}(K, L; \mathbb{Z}) \cong H_r(M \setminus L, M \setminus K; \mathbb{Z}).$$

On page 42 in Milnor's book (in the proof of Theorem 7.4) the author claims that there exists an open neighbourhood $U \subset V$ of $V \cap P$ which has the same homotopy type as $V \cap P$. Since $V \subset \mathbb{C}P^n$ is compact, the set $V \setminus U$ is a compact subset of $V$. We can therefore apply the duality theorem as stated above with $$M := V, \hspace{2mm} K:= V \setminus U, \hspace{2mm} L:= \emptyset$$ to obtain $$H_r(V, V \cap P) \cong H_r(V, U) \cong H^{2k-r}(V \setminus U) \cong H^{2k-r}(V \setminus (V \cap P))$$ as desired.

  • $\begingroup$ Thank you for your response. I gave you +1. However, I am concerned about letting $M=V$ because $V$ may have singularities. $V-P$ is certainly a complex manifold but $V$ may not be. Is there a way to fix this? $\endgroup$
    – Seth
    Aug 5 '14 at 13:38
  • $\begingroup$ Also I am not sure why we introduced $U$ as it seems unnecessary. $\endgroup$
    – Seth
    Aug 5 '14 at 13:40

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.