Addition is exponentially harder than counting for shallow monotone circuits

Let $U{k,N}$ denote the Boolean function which takes as input $k$ strings of $N$ bits each, representing $k$ numbers $a^{{(1)},\dots,a{(k)}$} in ${0,1,\dots,2^{N}-1}$, and outputs 1 if and only if $a^{(1)} + \cdots + a^{(k)} \geq 2^N.$ Let THR${t,n}$ denote a monotone unweighted threshold gate, i.e., the Boolean function which takes as input a single string $x \in {0,1}^n$ and outputs $1$ if and only if $x1 + \cdots + xn \geq t$. We refer to circuits that are composed of THR gates as monotone majority circuits. The main result of this paper is an exponential lower bound on the size of bounded-depth monotone majority circuits that compute $U{k,N}$. More precisely, we show that for any constant $d \geq 2$, any depth-$d$ monotone majority circuit computing $U{d,N}$ must have size $\smash{2^{{\Omega(N{1/d})}}$.} Since $U{k,N}$ can be computed by a single monotone weighted threshold gate (that uses exponentially large weights), our lower bound implies that constant-depth monotone majority circuits require exponential size to simulate monotone weighted threshold gates. This answers a question posed by Goldmann and Karpinski (STOC'93) and recently restated by Hastad (2010, 2014). We also show that our lower bound is essentially best possible, by constructing a depth-$d$, size-$2^{O(N{1/d})}$ monotone majority circuit for $U{d,N}$. As a corollary of our lower bound, we significantly strengthen a classical theorem in circuit complexity due to Ajtai and Gurevich (JACM'87). They exhibited a monotone function that is in AC$^0$ but requires super-polynomial size for any constant-depth monotone circuit composed of unbounded fan-in AND and OR gates. We describe a monotone function that is in depth-$3$ AC$^0$ but requires exponential size monotone circuits of any constant depth, even if the circuits are composed of THR gates.