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

Witten, in this paper, equation (3.42), says the integral over the complex plane of the imaginary part of a polynomial g(z)

$$\int dz d\bar z \exp( g(z)-\bar g (\bar z))$$

is convergent. How can I show it? Is there a good reference?

share|cite|improve this question

What Witten write is:

Introduce a complex variable $z$ and a complex-valued polynomial $$g(z) = \sum_{j=0}^n a_n z^n.\qquad (2.7)$$ Now consider the integral $$Z_g=\int|d^2z|\exp\bigl(g(z)-\overline{g(z)}\bigr).\qquad (2.8)$$ This again is a convergent oscillatory integral and a closer analog of complex Chern-Simons theory, with $-ig(z)$ and $ig(z)$ corresponding to the terms $tW$ and $\widetilde{t}\overline{W}$ in the action (2.3).

We stress that no contour integral is intended in (2.8) --- a contour integral could scarcely be intended here as the integrand is not holomorphic! Rather, if $z=u+iv$, with real $u$, $v$, we integrate separately over $u$ and $v$, the integration measure being $|d^2z|=2\,du\,dv$.

(end of quote)

But this is plainly false: Take $g(z)=z$, then $$g(z)-\overline{g(z)}=u+iv-(u-iv)=2iv$$ and the integral $$\int e^{2iv} du dv$$ is not convergent.

Maybe he has forgotten to say that the integral is extended to a measurable set $M\subset \textbf{C}$ of finite Lebesgue measure.

I think a more complex polynomial will not solve the problem.

Perhaps he intended to integrate only with respect to $v$. But then the measure will not be as he said explicitly $du dv$ but only $dv$. The integral will not be absolutely convergent. But this may be salvaged if he consider a polynomial of degree $\ge2$ and take principal value of the integral

Then we may translate in the following way. If $g$ is a complex plynomial of degree $\ge2$ then $$\int_{-\infty}^{+\infty} e^{i\Im g(iv)} dv=\lim_{T\to+\infty}\int_{-T}^T e^{i\Im g(iv)} dv$$ exists. For example taking $g(z)=-\lambda(z^3+z)$, with $\lambda$ real,
we get $g(iv)=i\lambda( v^3/3 -v)$ so that we obtain as particular case Airy's integral $$\int_{-\infty}^{+\infty} e^{i\lambda(v^3/3-v)} dv\qquad (2.6)$$ As the proper Witten write in equation (2.6) of the paper.

share|cite|improve this answer
Thank you very much for your input! Could you suggest a reference where I can look for something related? I'm also OK if it converges as a distribution ... – Yuji Tachikawa Jan 16 '13 at 5:23

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.