MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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

Hi, I'm actually physics student and I've not been able to understand how the following integration has been performed:


The fields e and A have support on these discrete structures. The $su(2)$-valued 1-form field $e$ is represented by the assignment of an $e \in su(2)$ to each 1-cell in $\Delta$. The connection field A is represented by the assignment of group elements $g_e \in SU(2)$ to each edge in $\mathcal{J}_\Delta$

The partition function is defined by

${\cal Z}(\Delta)=\int \prod_{f \in {\cal J}_{\Delta}} de_f\prod_{e \in {\cal J}_{\Delta}} dg_e e^{i {\rm Tr}\left[e_f U_f\right]}$

where $de_f$ is the regular Lebesgue measure on $R^3$, $dg_e$ is the Haar measure on $SU(2)$, and $U_f$ denotes the holonomy around faces, i.e., $U_f=g^1_e\dots g^{N}_e$ for $N$ being the number of edges bounding the corresponding face. Since $U_f \in SU(2)$ we can write it as $U_f=u^0_f\ {{1}} + F_f$ where $u^0_f\in C$ and $F_f \in su(2)$. $F_f$ is interpreted as the discrete curvature around the face $f$. Clearly ${\rm Tr}[e_f U_f]={\rm Tr}[e_f F_f]$. An arbitrary orientation is assigned to faces when computing $U_f$. We use the fact that faces in ${\cal J}_{\Delta}$ are in one-to-one correspondence with $1$-cells in $\Delta$ and label $e_f$ with a face subindex.

Integrating over $e_f$, we obtain

${\cal Z}(\Delta)=\int \ \prod_{e \in {\cal J}_{\Delta}} dg_e \prod_{f \in {\cal J}_{\Delta}}{\delta}(g^1_e\dots g^{N}_e),$

where $\delta$ corresponds to the delta distribution defined on ${\cal L}^2(SU(2))$.


The details are not important, what I need to understand is how the integration over the $e_f$'s has been performed


share|cite|improve this question
up vote 7 down vote accepted

Whoever performed that integration is using the following fact from Fourier analysis:

The "delta function supported at the position 0" is the Fourier transform of the constant function 1.

$\delta(x) = \frac{1}{2\pi}\int_{\mathbb{R}} 1 e^{ikx}dk$

You can prove this by approximating the delta-function with a sequence of Gaussian bump functions. Fourier transforming Gaussians inverts the variance, so as the Gaussians approach a "delta function spike", their Fourier transforms approach a constant.

share|cite|improve this answer
I know this fact; that the delta function is the Fourier transform of 1, but what bothers me here is that I have a trace in the exp! – Pedro May 13 '10 at 12:54
Trace defines a non-degenerate inner product on the Lie algebra $\mathfrak{su}(2)$. – userN May 13 '10 at 13:09
So the expression on the exp is like a product of two elements in $\mathfrak{su}(2)$?? – Pedro May 13 '10 at 13:33
a sum of products, yes. – userN May 13 '10 at 14:07
Merci beaucoup :-) – Pedro May 13 '10 at 15:00

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.