1
$\begingroup$

The deterministic query complexity $D(f)$ of a symmetric function $f$ is $\Omega(n)$ (except for f = 0 or f = 1). I am wondering if the same result holds for the (bounded-error) randomized query complexity? Do we have $R_{1/4}(f) = \Omega(n)$?

I know it's true for some well-known functions (like the OR function), and that it doesn't hold in the quantum setting (for instance, $Q(OR) = \Theta(\sqrt{n})$ by Grover's algorithm). In fact, $D(f) = O(Q(f)^2)$ so we have at least $R_{1/4}(f) = \Omega(\sqrt{n})$.

$\endgroup$

1 Answer 1

1
$\begingroup$

Ok, I found the answer in this survey: http://homepages.cwi.nl/~rdewolf/publ/qc/dectree.pdf

The sensitivity $s(f)$ of a (nonconstant) symmetric function $f$ is $s(f) \geq \lceil\frac{n+1}{2}\rceil$. However, $D(f) \geq s(f)$ and $R_{1/4}(f) \geq \frac{s(f)}{3}$...

There are other interesting results concerning symmetric functions in this survey (see Section 6.1).

$\endgroup$

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.