Say I have an embedding: $f: X\to Y$ and a non-vanishing form $\omega$ on $Y$. Are there conditions on $f$ so that the pull-back $f^* \omega$ vanishes everywhere?

Or in a specific case, say $\omega$ is the non-trivial curvature form of some line bundle on $Y$, when does $f^* \omega$ vanish?


I'm not sure if there is a general answer. I'll give two examples from symplectic geometry below. Your specific example on curvature forms seems like it should be a standard result but I don't know it off the top of my head; perhaps someone else can come by and shed some light on it.

Background: A symplectic manifold is a pair $(M, \omega)$ consisting of a smooth manifold $M$ and a closed, nondegenerate $2$-form $\omega \in \Omega^2(M)$; such forms care called symplectic forms. Symplectic manifolds are necessarily even-dimensional (in order for the nondegeneracy condition to possibly be satisfied). If $X$ is a smooth manifold and $$f: X \longrightarrow M$$ is an embedding, then $X$ is called a Lagrangian submanifold of $M$ if $f^\ast \omega = 0$ and $\dim(X) = \frac{1}{2} \dim(M)$. If $(M_1, \omega_1)$ and $(M_2, \omega_2)$ are two symplectic manifolds, a diffeomorphism $$\varphi: M_1 \longrightarrow M_2$$ is called a symplectomorphism if $\varphi^\ast \omega_2 = \omega_1$.

Example 1: If $X$ is any smooth manifold, the cotangent bundle $T^\ast X$ of $X$ has a natural symplectic form $\omega_{\text{can}} = - d\lambda$ where $\lambda$ is the $1$-form on $T^\ast X$ defined pointwise by $$\lambda_{(x,\xi)} = d\pi_{(x,\xi)}^\ast \xi$$ for any $(x, \xi) \in T^\ast X$, where $$\pi: T^\ast X \longrightarrow X$$ is the bundle projection. If $\alpha \in \Omega^1(X)$ is any $1$-form on $X$, then we can define an embedding of $X$ into $T^\ast X$ by $$f_\alpha: X \longrightarrow T^\ast X,$$ $$x \mapsto (x, \alpha_x).$$ Write $X_\alpha$ for the image of $X$ under $f_\alpha$.

Theorem. Let $\alpha \in \omega^1(X)$. Then $X_\alpha$ is a Lagrangian submanifold of $(T^\ast X,\omega_\text{can})$ if and only if $\alpha$ is closed.

See my answer here for the proof and more details. So here we see that $f^\ast_\alpha \omega_{\text{can}} = 0$ if and only if $d\alpha = 0$.

Example 2: Let $(M,\omega)$ be any symplectic manifold. Then $(M \times M,\omega \oplus -\omega)$ is a symplectic manifold, where $$\omega \oplus -\omega = \mathrm{pr}_1^\ast \omega - \mathrm{pr}_2^\ast \omega.$$ Let $$\varphi: M \longrightarrow M$$ be a diffeomorphism. Then we can define an embedding of $M$ into $M \times M$ by $$f_\varphi: M \longrightarrow M \times M,$$ $$x \mapsto (x, \varphi(x)).$$ Write $M_\varphi$ for the image of $M$ under $f_\varphi$.

Theorem. $\varphi$ is a symplectomorphism if and only if $M_\varphi$ is a Lagrangian submanifold of $(M \times M, \omega \oplus -\omega)$.

Proof. We have that \begin{align*} f_\varphi^\ast (\omega \oplus -\omega) & = f_\varphi^\ast \mathrm{pr}_1^\ast \omega - f_\varphi^\ast \mathrm{pr}_2^\ast \omega \\ & = (\mathrm{pr}_1 \circ f_\varphi)^\ast \omega - (\mathrm{pr}_2 \circ f_\varphi)^\ast \omega \\ & = \mathrm{Id}^\ast_M \omega - \varphi^\ast \omega \\ & = \omega - \varphi^\ast \omega. \end{align*} So $M_\varphi$ is Lagrangian if and only if $f_\varphi^\ast (\omega \oplus -\omega) = 0$, which is true if and only if $\omega - \varphi^\ast \omega = 0$, or in other words $$\omega = \varphi^\ast \omega,$$ so that $\varphi$ is a symplectomorphism.

So here we see that the pullback being identically zero depends on $\varphi$ preserving the symplectic form, which seems to be a fundamentally different reason than the previous example. This makes me believe that a general answer might not be possible.


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.