Title: Rounding via Low Dimensional Embeddings

Abstract: A regular graph $G = (V,E)$ is an $(\varepsilon,\gamma)$ small-set expander if for any set of vertices of fractional size at most $\varepsilon$, at least $\gamma$ of the edges that are adjacent to it go outside. In this paper, we give a unified approach to several known complexity-theoretic results on small-set expanders. In particular, we show:
1. Max-Cut: we show that if a regular graph $G = (V,E)$ is an $(\varepsilon,\gamma)$ small-set expander that contains a cut of fractional size at least $1-\delta$, then one can find in $G$ a cut of fractional size at least $1-O\left(\frac{\delta}{\varepsilon\gamma^6}\right)$ in polynomial time.
2. Improved spectral partitioning, Cheeger's inequality and the parallel repetition theorem over small-set expanders. The general form of each one of these results involves square-root loss that comes from certain rounding procedure, and we show how this can be avoided over small set expanders.
Our main idea is to project a high dimensional vector solution into a low-dimensional space while roughly maintaining $\ell_2^2$ distances, and then perform a pre-processing step using low-dimensional geometry and the properties of $\ell_2^2$ distances over it. This pre-processing leverages the small-set expansion property of the graph to transform a vector valued solution to a different vector valued solution with additional structural properties, which give rise to more efficient integral-solution rounding schemes.