Boolean function monotonicity testing requires (almost) $n^{1/2}$ non-adaptive queries

Abstract: We prove a lower bound of $\Omega(n^{1/2 - c})$, for all $c&gt;0$, on the query complexity of (two-sided error) non-adaptive algorithms for testing whether an $n$-variable Boolean function is monotone versus constant-far from monotone. This improves a $\tilde{\Omega}(n<sup>{1/5})$ lower bound for the same problem that was recently given in [CST14] and is very close to $\Omega(n<sup>{1/2})$, which we conjecture is the optimal lower bound for this model.