1: \begin{abstract}

2: In this work, we compare the performance of the Quantum Approximate Optimization Algorithm (QAOA) with  state-of-the-art classical solvers such as Gurobi and MQLib to solve the combinatorial optimization problem MaxCut on 3-regular graphs. The goal is to identify under which conditions QAOA can achieve “quantum advantage” over classical algorithms,  in terms of both solution quality and time to solution.

3: %The use of approximate solutions exponentially improves the classical convergence speed and puts strict requirements on quantum time to solution. 

4: One might be able to achieve quantum advantage on hundreds of qubits and moderate depth $p$ by sampling the QAOA state at a frequency of order 10 kHz.

5: %To solve this problem, a quantum device needs to execute circuits of $p\leq12$.

6: We observe, however, that classical heuristic solvers are capable of producing high-quality approximate solutions in \textit{linear} time complexity.

7: In order to match this quality for \textit{large} graph sizes $N$, a quantum device must support depth $p>11$.

8: %In this context, we find that given a graph of $\approx 100$ nodes, a quantum computer has to provide a solution in $<1$ msec, and support QAOA depth $p\geq6$ to compete with a simple classical solver. Against the closed-source solver Gurobi, QAOA requires a depth $p\geq11$, with a time to solution of order $1$sec.

9: Otherwise, we demonstrate that 

10: the number of required samples grows exponentially with $N$, hindering the scalability of QAOA with $p\leq11$.

11: These results put challenging bounds on achieving quantum advantage for QAOA MaxCut on 3-regular graphs.

12: Other problems, such as different graphs, weighted MaxCut, maximum independent set, and 3-SAT, may be better suited for achieving quantum advantage on near-term quantum devices.

13: 

14: 	\end{abstract}

15: