6
$\begingroup$

Last year I had watched this talk online about problems in $P$ for which an algorithm which runs in some subclass of $P$, say in subquadratic time, would imply $P = NP$, or violate the ETH (exponential time hypothesis). I cannot find this talk and am interested in said problems, for I want to look into quantum algorithms for this group of problems during the semester I'm beginning right now.

$\endgroup$
8
$\begingroup$

Virginia Vassilevska Williams lectured at a bootcamp (link to outline) at the Simons Institute, and presents what may be your memory in the introductory video. The whole workshop is worthwhile; the outline and links to videos of all talks can be found here.

She explains (starting at 42:30 in the video) that if there is a $O\left(n^{2-\varepsilon}\right)$ algorithm for some $\varepsilon>0$ for the Orthogonal Vectors problem, then the Strong Exponential Time Hypothesis (SETH) is false. This would not immediately imply $P=NP$, but it would show an $O\left(2^{n-\epsilon}\right)$ algorithm for every $k$-SAT, which would be better than what we have right now; right now we have algorithms which solve $k$-SAT in $2^{n-O\left(\frac{1}{k}\right)}$, so these algorithms tend to $2^{n}$ for large $k$.

The proof is originally by Ryan Williams, and is quite elegant; I do recommend the lecture.

$\endgroup$

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.