2
$\begingroup$

In 11D supergravity we have the Chern-Simons term of the 3-form field $C$ \begin{equation} \int C \land d C \land d C \end{equation} I want to consider this on a spacetime $\mathbb{R}^{1,6}\times S$ where $S$ is a resolution of an $A_n$ singularity. Then $H_2(S)=\mathbb{R}^n$, and I decompose the 3-form as \begin{equation} C = \sum_i^n \omega_i \land A_i \end{equation} where $\omega_i$ form a basis of the harmonic 2-forms on $S$ and $A_i$ are 1-forms on $\mathbb{R}^{1,6}$. Now plugging this into the Chern-Simons term I get \begin{equation} \int_{\mathbb{R}^{1,6}} \int_S \sum_{ijk}^n \omega_i \land A_i \land\omega_j \land dA_j \land\omega_k \land dA_k \overset{?}{=} \sum_{ijk}^n \int_{\mathbb{R}^{1,6}} \int_S \omega_i \land\omega_j\land\omega_k \land A_i \land dA_j \land dA_k \end{equation} at which point I'm not sure how to integrate over $S$, especially given that $\omega_i \land\omega_j\land\omega_k$ is a 6-form in a 4-dimensional space.

Any help would be greatly appreciated.

$\endgroup$
5
  • $\begingroup$ Why do you only use the two-forms of $S$ and one-forms on $\mathbb{R}^{1,6}$? In general $$\Omega^n(X\times Y)=\bigoplus_{p+q=n}\Omega^p(X)\otimes\Omega^q(Y).$$ $\endgroup$ Commented Jul 22, 2022 at 21:54
  • $\begingroup$ @ɪdɪətstrəʊlə you're right. In the expansion I've ignored other terms. The terms I wrote are of interest because of the $\operatorname{U}(1)$ gauge field term arising when reducing to 7D, although I don't know how to do this. I'm not sure how other terms are dealt with, I'm new to this. $\endgroup$
    – Anton B
    Commented Jul 23, 2022 at 21:31
  • $\begingroup$ I've seen in the paper link_Papadopoulos et al_ that we can just write the term $\int_S \omega_i \land \omega_j \land \omega_k = m_{ijk}$, but how can a 6-form be defined in a 4D space? $\endgroup$
    – Anton B
    Commented Jul 23, 2022 at 22:22
  • $\begingroup$ In this paper $\omega$'s are 2-forms in a Calabi-Yau manifold, which is (real) 6-dimensional, so $\omega_i\wedge\omega_j\wedge\omega_k$ is inevitably proportional to the volume form of the CY. $\endgroup$ Commented Jul 23, 2022 at 22:25
  • $\begingroup$ Of course, thanks $\endgroup$
    – Anton B
    Commented Jul 24, 2022 at 8:35

1 Answer 1

4
$\begingroup$

$\newcommand{\d}{\mathrm{d}}\newcommand{\w}{\wedge}\newcommand{\R}{\mathbb{R}}$That integral vanishes because the differential form $\omega_i\w\omega_j\w\omega_k\in\Omega^6(S_4)$ vanishes (you cannot antisymmetrise more that $d$ indices in a $d$-dimensional manifold).

What you could do, instead, is use the fact that $$\Omega^n(X\times Y)=\bigoplus_{p+q=n}\Omega^p(X)\w\Omega^q(Y),$$ to write $$ \Omega^3\!\left(\R^{1,6}\times S_4\right) \ni C= \sum_{\begin{smallmatrix} 0\leq \!\!\!\!&p&\!\leq 3 \\ 0\leq \!\!\!\!&q&\!\!\leq 3\\ p+q&=&3\end{smallmatrix}} A_{[p]} \w \omega_{[q]}, $$ where $A_{[p]}\in\Omega^p\!\left(\R^{1,6}\right)$ and $\omega_{[q]}\in\Omega^q(S_4)$. Suppose that, for some reason, you only consider $\omega_{[q]}\in\mathrm{H}^q(S_4)$ (following e.g. Compactification of D=11 supergravity on spaces of exceptional holonomy that was linked in a comment by the OP). Then the integral $$\int_{\R^{1,6}\times S_4} C\w\d C\w \d C,$$ receives contributions from $$ \omega_{[q]}\w \omega_{[q']}\w \omega_{[q'']} $$ with $q+q'+q''=4$.

Let's expand $\omega_{[q]}$ in an orthonormal basis of the $q$-th cohomology group $\left\{\omega_{I_q}\right\}$, with $I_q\in \left\{1,2,\cdots, b_q=\mathrm{\dim}(\mathrm{H}^q(S_4))\right\}$: $$ \omega_{[q]} = \left<\omega_{[q]},\omega^{I_q}\right>\ \omega_{I_q},$$ so that $$C = \sum_{p+q=3} A^{I_q}\w \omega_{I_q},$$ with implicit summation over $I_q$ and with the coefficients $\left<\omega_{[q]},\omega^{I_q}\right>$ soaked up by $A^{I_q}$.

The desired integral decomposes, then, as \begin{align} \int_{\R^{1,6}\times S_4} C\w\d C\w \d C &= \sum_{\begin{smallmatrix}p+q&=&3 \\p'+q'&=&3 \\ p''+q'' &=&3 \end{smallmatrix}}\int_{\R^{1,6}\times S_4} A^{I_q}\w \omega_{I_q} \w \d A^{J_{q'}}\w \omega_{J_{q'}}\w \d A^{K_{q''}}\w \omega_{K_{q''}} = \\ &= \sum_{\begin{smallmatrix}p+q&=&3 \\p'+q'&=&3 \\ p''+q'' &=&3 \\ q+q'+q''&=&4 \end{smallmatrix}} C_{I_{q},J_{q'},K_{q''}} \int_{\R^{1,6}} A^{I_{q}}\w\d A^{J_{q'}}\w\d A^{K_{q''}}, \end{align} where $$ C_{I_{q},J_{q'},K_{q''}} :=\int_{S_4} \omega_{I_q}\w \omega_{J_{q'}}\w \omega_{K_{q''}}, $$ generalise $C_{IJK}$ (eq. (3)) and $C_{IJk}$ (eq. (9)) of the linked reference. If, furthermore, in your case at hand some of the $C_{I_{q},J_{q'},K_{q''}}$ vanish you can simplify your result further.

$\endgroup$
5
  • $\begingroup$ That's brilliant, thanks very much for the detailed answer! $\endgroup$
    – Anton B
    Commented Jul 24, 2022 at 8:38
  • $\begingroup$ Cheers! you can accept the answer by clicking the check button if you found it satisfactory :) $\endgroup$ Commented Jul 24, 2022 at 10:57
  • $\begingroup$ Just a note to the details of my question: I talked about the resolution of an $A_n$ singularity which I found out, by some result in algebraic geometry (which perhaps someone can fill in here since I don't know exactly myself), that only the zeroth and middle cohomology class of $S$ is non-vanishing (in this case $H^2$), and so in fact in the KK expansion of $C$ there are no other terms and so the Chern-Simons term vanishes. Nevertheless, the answer is extremely useful for a general compactification. Thanks $\endgroup$
    – Anton B
    Commented Jul 24, 2022 at 13:19
  • $\begingroup$ But wait, it doesn't vanish. You can still use the zeroth cohomology. Then you have non-vanishing terms of the form $\omega_{I_2}\wedge\omega_{J_2}\wedge\omega_{K_0}$ (and permutations thereof). $\endgroup$ Commented Jul 24, 2022 at 13:50
  • $\begingroup$ Yes of course. I believe then that this term corresponds to the intersection form on $S$ which, again by the properties of $A_n$ (ADE/ Du Val) singularities, is equivalent to negative the Cartan matrix of the $A_n$ (ADE) group. $\endgroup$
    – Anton B
    Commented Jul 24, 2022 at 19:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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