Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am a physicist studying now some supersymmetric sigma models. My question can, however, be reformulated in a purely mathematical language: A twisted de Rham complex involves $d_W = d + dW \wedge $ where $W$ is any even-dimensional form. In all known for me cases the cohomologies of this complex are the same as for the untwisted one. Can one assert that it is always so ? If yes, where is it proven ?

share|cite|improve this question
Are you asking if the de Rham complex $(\Omega_X^\cdot,d)$ is homotopic to the twisted de Rham complex $(\Omega_X^\cdot, d+d_\omega)$, where $d_{\omega} := d+ d\omega \wedge$? I don't see how $d_\omega$ maps $\Omega_X^i$ to $\Omega^{i+1}_X$. Could you elaborate? – Harry Apr 9 '12 at 9:55
I do not know well the mathematical language that you use. Let me explain it in my own terms. For the usual de Rham complex there are some number of even-dimensional forms that are annihilated by both d and d^\dagger. Call this number \beta_{even}. One define analogously \beta_{odd}. The difference \beta_{even} - \beta{odd} is the Euler characteristics of the manifold. For the twisted complex, this difference is not changed. For simple manifolds like S^n I understand how to show that \beta_{even} and \beta_{odd} are not changed SEPARATELY. The question is whether it is true in general ? – Andrei Smilga Apr 9 '12 at 12:08
Ahh I understand the question better now. Unfortunately, I don't know the answer. – Harry Apr 9 '12 at 17:50

I hadn't heard of this definition of twisted deRham cohomology before reading this (I had heard of the one in this question, but they seem to be different) so I don't have full confidence in this. I claim that the answer is "yes" to a question which I hope is equivalent:

Let $X$ be a smooth manifold. The usual definition of $H^i_{DR}(X)$ is "closed $i$-forms modulo exact $i$-forms". In other words, you take those $i$-forms killed by $d$ and quotient by $i$-forms of the form $d \theta$, for $\theta$ an $(i-1)$-form.

Here is the analogous definition in your setting. Let $\Omega^{even}$ and $\Omega^{odd}$ be the spaces of even and odd forms. Let $d_W(\alpha) = d \alpha + (d W) \wedge \alpha$. Define a form to be $d_W$-closed if it is killed by $d_W$, and to be $d_W$-exact if it is of the form $d_W(\theta)$ for some $\theta$. I will show that the space of $d_W$-closed forms, modulo the space of $d_W$-exact forms, has dimension independent of $W$.

Set $$e^W = \sum \frac{W^k}{k!}.$$ If $W$ has no $0$-form component, then this sum only has finitely many terms because high enough powers of $W$ are $0$; if there is a $0$-form in $W$, then the sum above is still convergent.

Observe that $d (e^W) = e^W dW$. This crucially uses that $W$ is even, so $W$ and $dW$ commute. (If $W$ were purely odd, this would also be true, but if $W$ were a mix of even and odd parts, then it needn't be true.)

So $$d \left( e^W \alpha \right) = e^W dW \alpha + e^W d \alpha = e^W d_W \alpha \quad (\ast)$$ using that again that $e^W$ is even.

So $\alpha$ is $d_W$-closed if and only if $e^W \alpha$ is $d$-closed and $\alpha$ is $d_W$-exact if and only if $e^W \alpha$ is exact. So we can identify the space of $d_W$-closed forms modulo the space of $d_W$-exact forms with the space of $d_W$ forms modulo the space of $d_W$ forms by multiplying by $e^W$.

Now, in the ordinary deRham cohomology theory, one shows that the orthogonal complement to the $d$-exact forms is the kernel of $d^{\dagger}$. So, instead of quotienting $\mathrm{Ker}(d)$ by $\mathrm{Image}(d)$, one can instead take the intersection $\mathrm{Ker}(d) \cap \mathrm{Ker}(d^{\dagger})$. Is this also true for $d_W$ and $d_{W}^{\dagger}$ ? In that case, the definition I am working with is the same as yours.

share|cite|improve this answer
Is there any intuition for thinking that studying $e^W$ would be useful for this problem? I mean, I can see how it basically solved almost the whole thing, but I'm wondering what made you think "Looking at $e^W$ might be beneficial!" Is it just that finding forms satisfying $d\omega = 0$ but $\omega \neq d\eta$ for any $\eta$ is essentially solving differential equations, and exponential functions are often helpful in this context? – Jason DeVito Apr 9 '12 at 19:19
Two things. (1) The kind of twisted cohomology I know uses $d+\alpha \wedge$, where $\alpha$ is a $1$-form. If $\alpha = d w$ for a scalar function $w$, then this is ordinary cohomology, by this argument. (2) I started trying to write down a class $\alpha$ in the kernel of $d_W$ for $X=S^n$ and $W$ a $2$-form. We must have $d \alpha_0$, so $\alpha_0$ is constant -- let's make it $1$. So $d \alpha_2 + dW=0$ -- it seems like a good bet that $\alpha_2 = -W$. So $d \alpha_4 - W dW=0$. I triple checked that $d(W \wedge W) =2 W dW$ for $W$ a $2$-form. So it seems likely that $\alpha_4 = W^2/2$. – David Speyer Apr 9 '12 at 19:45
So I had $1-W+W^2/2+\cdots$, and at that point there was an obvious guess. – David Speyer Apr 9 '12 at 19:46
That is the actual history. A clever way to think of it is: We want $d \eta+W \wedge \omega=0$, so we want $\omega = \mathrm{Constant} \cdot e^W$, and when we say "constant", we actually mean "in the kernel of $d$". – David Speyer Apr 9 '12 at 19:48
Actually, I have found the reference [Mathai + Wu, arXiv:0810.4204], where the statement about the equivalence of the twisted de Rham complexes $d + H$ and $d + H +dB$ ($H$ being a closed 3-form) was proven in the same way as David did (with invoking the exponential $e^B$). I would not be surprised though, if somebody noticed this before them... – Andrei Smilga Apr 11 '12 at 12:52

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.