Alternation, Sparsity and Sensitivity : Bounds and Exponential Gaps (1712.05735v2)

Abstract: $\newcommand{\sp}{\mathsf{sparsity}}\newcommand{\s}{\mathsf{s}}\newcommand{\al}{\mathsf{alt}}$ The well-known Sensitivity Conjecture states that for any Boolean function $f$, block sensitivity of $f$ is at most polynomial in sensitivity of $f$ (denoted by $\s(f)$). The XOR Log-Rank Conjecture states that for any $n$ bit Boolean function, $f$ the communication complexity of a related function $f^{\oplus}$ on $2n$ bits, (defined as $f^{\oplus}(x,y)=f(x\oplus y)$) is at most polynomial in logarithm of the sparsity of $f$ (denoted by $\sp(f)$). A recent result of Lin and Zhang (2017) implies that to confirm the above conjectures it suffices to upper bound alternation of $f$ (denoted $\al(f)$) for all Boolean functions $f$ by polynomial in $\s(f)$ and logarithm of $\sp(f)$, respectively. In this context, we show the following : * There exists a family of Boolean functions for which $\al(f)$ is at least exponential in $\s(f)$ and $\al(f)$ is at least exponential in $\log \sp(f)$. En route to the proof, we also show an exponential gap between $\al(f)$ and the decision tree complexity of $f$, which might be of independent interest. * As our main result, we show that, despite the above gap between $\al(f)$ and $\log \sp(f)$, the XOR Log-Rank Conjecture is true for functions with the alternation upper bounded by $poly(\log n)$. It is easy to observe that the Sensitivity Conjecture is also true for this class of functions. * The starting point for the above result is the observation (derived from Lin and Zhang (2017)) that for any Boolean function $f$ and $m \ge 2$, $deg(f)\le \al(f)deg_2(f)deg_m(f)$ where $deg(f)$, $deg_2(f)$ and $deg_m(f)$ are the degrees of $f$ over $\mathbb{R}$, $\mathbb{F}_2$ and $\mathbb{Z}_m$ respectively. We also show three further applications of this observation.