Measurement of MaxCut QAOA solution quality with increasing $p$-depth

We solve MaxCut problems using the Quantum Approximate Optimization Algorithm (QAOA) with up to ten trapped $^{171}$Yb$^+$ ions.

Our novel control scheme facilitates native implementation of fully connected, weighted MaxCut graphs.

We present results for $N=3,6,10$ vertex native graphs with QAOA layer number $p\le6$ and report increasing approximation ratio ($\left<C\right>/C_{Max}$) as a function of $p$-depth up to $p=4$ for a $N=10$ vertex graph. Our results employ state preparation and measurement (SPAM) error mitigation to correct for known measurement crosstalk and to increase the observed approximation ratio. We discuss prospects of additional error mitigation at increased $p$-depth.