2
$\begingroup$

enter image description here

I am kind of confused with the conjugate linear 1-form, because it seems that to any conjugate linear 1-form $x$ and any $v\in E$, $x(v)=0$, in this sense the conjugation doesn't make sense.

I am considering an example: Let $X$ be a complex manifold of dim $n$ and $E$ be the tangent space of $X$ at $a$: $T_a^{1,0}X=<...,\frac{\partial}{\partial x_j}-i\frac{\partial}{\partial y_j},...>=<...,\frac{\partial}{\partial z_j},...>$, where $\frac{\partial}{\partial z_j}=1/2(\frac{\partial}{\partial x_j}-i\frac{\partial}{\partial y_j})$,

Now with this notation, we have $F$ above should be $(T_a^*X_0)_{\mathbb{C}}=(T_a^*X)^{1,0}\oplus (T_a^*X)^{0,1}=<...,dz_j,...>\oplus <...,d\bar{z_j},...>$, s.t. $dz_i(\frac{\partial}{\partial z_j})=\delta_{ij}$ and $d\bar{z_i}(\frac{\partial}{\partial {z_j}})=0$. And $\wedge^{1,0}F=(T_a^*X)^{1,0}$, $\wedge^{0,1}F=(T_a^*X)^{0,1}$.

But by $d\bar{z_i}(\frac{\partial}{\partial {z_j}})=0$, to any $v\in E=T_a^{1,0}X=<...,\frac{\partial}{\partial z_j},...>$, we have $d\bar{z_i}(v)=0$. But $d\bar{z_i}$ is the conjugate of $d{z_i}$, and $d{z_i}(v)$ might not have to be $0$, thus there is a contradiction. Where did I go wrong?

More generally, let $E=<d_1,..d_n>=<x_1+iy_1,...,x_n+iy_n>$, then $F=<x_1^*,y_1^*,...,x_n^*,y_n^*>\oplus i<x_1^*,y_1^*,...,x_n^*,y_n^*>$, then the $\wedge^{1,0}F$ as the set of complex linear 1-forms should be $<...,1/2(x_j^*-iy_j^*),...>$, then what's the conjugate? It seems that it has a similar problem.

$\endgroup$
1
$\begingroup$

The confusion is at the line $E = T^{1,0}X$. It is true that there is an (complex isomorphism between $(E, J)$ and $(T^{1,0}X, \sqrt{-1})$ given by

$$ v\in E\mapsto v - \sqrt{-1} Jv \in T^{1,0}X,$$

but if you think of $E \cong E\otimes \{1\}$ in $E\otimes \mathbb C$, then $E\cap T^{1,0}X = \{0\}$.

$\endgroup$
  • $\begingroup$ I feel a little confused, here $E$ is a complex space, but it seems you see $E$ as a real space. $\endgroup$ – Danny Jul 31 '18 at 12:59
  • $\begingroup$ Also, for my last part, the general case, the basis of the conjugate should be $\{ ...,1/2(x_j^*+iy_j^*,...\}$, then still we have $1/2(x_j^*+iy_j^*)(x_k+iy_k)=0$, thus the conjugate element acts trivially on $E$. $\endgroup$ – Danny Jul 31 '18 at 13:04
  • $\begingroup$ Yes I am thinking of $E$ as $2n$-dimensional real space and this is confusing. $E$ is complex so it has a complex structure $J$. We do not call it $i$, since that will cause confusion when you consider $E\otimes \mathbb C$. So your $x_j + i y_j$ is actually not in $E$ but only in $E\otimes \mathbb C$. $\endgroup$ – user99914 Jul 31 '18 at 13:13
  • $\begingroup$ Now if $C^n=E=<d_1,..d_n>\cong <x_1, y_1,...,x_n,y_n>=R^{2n}$ and $F=<x_1^*,y_1^*,...,x_n^*,y_n^*>\oplus i<x_1^*,y_1^*,...,x_n^*,y_n^*>$, then how do elements of $F$ act on $E$? Can you give an example? $\endgroup$ – Danny Jul 31 '18 at 13:22
  • $\begingroup$ For example $(x_1^* + i y_1^*) (x_1) = x_1^* (x_1) + i y_1^*(x_1) = 1 + i\cdot 0 = 1$. $\endgroup$ – user99914 Jul 31 '18 at 13:35

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.