1
$\begingroup$

The Hodge star operator $\ast$ acts on the differential forms of a differential manifold sending $\Omega^{k}$ to $\Omega^{N-k}$. If the manifold is complex, then for $p+q=k$, does $\ast$ map $\Omega^{p,q}$ into some $\Omega^{a,b}$, where $a+b=N-k$.

$\endgroup$
8
  • $\begingroup$ The last question is ambiguous, but I think the 2nd sentence is also a question, so you should put a question mark after it. You can find the answer in many places, e.g. Griffiths-Harris page 66. $\endgroup$ Nov 21 '10 at 19:14
  • $\begingroup$ So yes in the sense you mean. $\endgroup$ Nov 21 '10 at 19:18
  • $\begingroup$ The last question is just a rewording of the previous line. Since it is non-essential and apparently confusing I'll delete it. $\endgroup$ Nov 21 '10 at 19:24
  • 1
    $\begingroup$ Since $∗$ is a real operator, to be more precise, you should say that after complexifying the space of forms, and extending $∗$ to be complex linear, then indeed $∗$ maps $\Omega^{p,q}$ to $\Omega^{N-q,N-p}$. Frequently, it is preferable to use $\bar *$, which is the composition of $∗$ with complex conjugation, to map $\Omega{p,q}$ to $\Omega{N−p,N−q}$. This way, we have $\alpha \wedge \bar * \beta = g(\alpha, \beta) vol$, where $g$ is the Hermitian metric. $\endgroup$ Nov 21 '10 at 19:42
  • 1
    $\begingroup$ Our comments seemed to have crossed. Yes, p 82 not 66, and yes. (Be warned that some authors use a different convention, where $N-p,N-q$ get switched. It's a question of whether $*$ is linear or antilinear.) $\endgroup$ Nov 21 '10 at 19:42
4
$\begingroup$

As Abtan requested, I'm converting my comments to an answer:

Suppose that $X$ is an $N$ (complex) dimensional complex manifold endowed with a Hermitean metric, or equivalently a Riemannian metric g satisfying $g(JX,JY)=g(X,Y)$, where $J$ is the complex structure. Let $*$ denote the $\mathbb{C}$-antilinear extension of the Hodge star operator to complex valued forms (some people -- including me -- prefer to write this as $\overline{*}$ as Spiro points out in the comments). Then as one finds on page 82 of Griffiths and Harris, $$*\Omega^{pq}\subset \Omega^{N-q,N-p}$$ where I'm following the notation in the question and writing $\Omega^{pq}$ for the space of $C^\infty$ forms of type $(p,q)$.

$\endgroup$
1
  • $\begingroup$ Isn't the point that the antilinear extension sends p,q to N-p,N-q in contrast to what is written in your answer? $\endgroup$
    – Tomo
    Sep 20 '16 at 16:46

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.