We show that the quantum approximate optimization algorithm (QAOA) can construct with polynomially scaling resources the ground state of the fully-connected p-spin Ising ferromagnet, a problem that notoriously poses severe difficulties to a Quantum Annealing (QA) approach, due to the exponentially small gaps encountered at first-order phase transition for ${\rm p} \ge 3$. For a target ground state at arbitrary transverse field, we find that an appropriate QAOA parameter initialization is necessary to achieve a good performance of the algorithm when the number of variational parameters $2{\rm P}$ is much smaller than the system size ${\rm N}$, because of the large number of sub-optimal local minima. Instead, when ${\rm P}$ exceeds a critical value ${\rm P}^*_{\rm N} \propto {\rm N}$, the structure of the parameter space simplifies, as all minima become degenerate. This allows to achieve the ground state with perfect fidelity with a number of parameters scaling extensively with ${\rm N}$, and with resources scaling polynomially with ${\rm N}$.

Main point

Studies r-spin Ising magnet \((\sum_{i=1}^N \sigma_i^z)^r\) with external \(\sigma_x\) magnetic field.

Although there are barriers with quantum annealing (very rough energy landscape), the QAOA_p can do well when p is above a critical value \(\approx N/2\) (because all the minima are degenerate).
When p is below the threshold, you have to choose the parameters well to easily find the ground state.

Results

p=1 QAOA can solve all r Ising magnet with h=0 and N odd (p=2 if N even)

This was known for r=2 Ising magnet (i.e. this is maxcut on a complete graph)

for odd r, \(\beta=\gamma=\pi/4\) is the answer; for even r, \(\gamma\) depends on \(\beta\)

Then QAOA (including the variational parameter search) is simulated to estimate the time needed to converge.