Consider an algebraic function $\phi$ on $R^{d}$.

By this I mean that there exists a polynomial $P$ with coefficients in $R[x_1,...,x_d]$ (coefficients are polynomials!) such that $P(\phi) = 0$

Let $r$ be a real number >0, I would like to know whether $1/r^{d}$ times $\int_{[-r,r]^d} e^{i \phi}$ tends to zero when $r$ tends to infinity.

Of course it would be enough to show that this integral is bounded by $r^{d-\epsilon}$...

I am not at all familiar with the subject, but from the literature I looked at, I understand that this is the case when $\phi$ itself is a polynomial. I was wondering whether this holds true for any algebraic function.


The answer is no. Consider $\phi(x)=1+1/(x^2 +1)$. Then as $r\rightarrow\infty$ your integral approaches $e^i$.

If you want instead your function $\phi$ to be integral over the polynomial ring, meaning it satisfies a monic polynomial equation, then the answer is yes as long as $\phi$ is not a constant function. To see this, note that you can by partial integration replace the integrand with $e^{i\phi(x)}\frac{\phi''(x)}{\phi'(x)^2}$. Now if $\phi(x)$ grows like $x^a$ then $\frac{\phi''(x)}{\phi'(x)^2}$ grows like $x^{-a}$. The integrality assumption forces $a>0$, which completes the proof.

Note that I assume you want $\phi$ real valued, else all hell breaks loose.


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.