I'm reading Tao's Park City notes on the restriction conjecture http://arxiv.org/abs/math/0311181. He says at some point that the estimate:

there is a constant $C > 0$ such that, for any $R \ge 1$, $\alpha \ge 0$ and test function $f$ supported on $B(x_0, R)$, $||\hat{f}||_{L^q(S^{n-1})} \le C R^{\alpha}||f||_{L^p(B(x_0, R))}$

is equivalent to the estimate

$||\hat{f}||_{L^q(N_{\frac{1}{R}}(S^{n-1})} \le CR^{\alpha - \frac{1}{q}} ||f||_{L^p(B(x_0, R))}$

where $N_{\frac{1}{R}}(S^{n-1})$ is the $1/R$ neighborhood about $S^{n-1}$. The actual proof of this is made as an exercise, and his suggestion for the forward implication here is to translate the estimate by a factor of $O(1/R)$ and then average the estimate over all such translations. However, it's easy to see that if the first estimate holds, then

$||\hat{f}||_{L^q(S^{n-1} + v)} \le CR^{\alpha} ||f||_{L^p(B(x_0, R))}$ because the translation will simply apply a modulation in physical space which won't affect the magnitude of $f$ anyway. I'm not really sure what he means by averaging the translations, but I assume he means something like integrating

$\int_{B(0; \frac{1}{R})} ||\hat{f}||_{L^q(S^{n-1} + v)} dv$ and relating this term to $||\hat{f}||_{L^q(N_{\frac{1}{R}}(S^{n-1}))}$, but I don't really see how to proceed. Any advice would be appreciated.

  • $\begingroup$ A quick comment in case it helps. I haven't looked through the paper yet. The average of a function $f$ over a set $A$ usually means something like $\frac{1}{\mu(A)}\,\int_{\mathrm space}f\,d\mu$. In your case it could mean $\frac{1}{\mu({\mathrm space})}\,\int_{\mathrm space}f\circ \tau_x\,d\mu(x)$. $\tau_x$ being the translation by $x$. $\endgroup$ – InTransit Jul 21 '14 at 8:25

I do not think that it is what Tao means. If you have this inequality for the unit sphere, you also have it for the sphere that is centered at $0$ of radius $1+\varepsilon$. Just use the result for the function obtained from $f$ by the dilation that transforms this sphere into the unit sphere. Next you integrate in $\varepsilon$ in the interval $(-1/R, +1/R)$ to find the required estimate: the integral on $N_{R^{-1}}(S^{n-1})$ is bounded by $R^{-1}$ times the sup of the integrals over these spheres.


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.