The Quantum Approximate Optimization Algorithm can be applied to search problems on graphs with a cost function that is a sum of terms corresponding to the edges. When conjugating an edge term, the QAOA unitary at depth p produces an operator that depends only on the subgraph consisting of edges that are at most p away from the edge in question. On random d-regular graphs, with d fixed and with p a small constant time log n, these neighborhoods are almost all trees and so the performance of the QAOA is determined only by how it acts on an edge in the middle of tree. Both bipartite random d-regular graphs and general random d-regular graphs locally are trees so the QAOA's performance is the same on these two ensembles. Using this we can show that the QAOA with $(d-1)^{2p} < n^A$ for any $A<1$, can only achieve an approximation ratio of 1/2 for Max-Cut on bipartite random d-regular graphs for d large. For Maximum Independent Set, in the same setting, the best approximation ratio is a d-dependent constant that goes to 0 as d gets big.
Outline of argument
random d-regular graph, almost all edge-p-neighborhoods are trees if \((d-1)^{2p} < n^A\) for \(A < 1\).
Then consider a bipartite graph (completely satisfiable), so the best possible is “cut all edges”
so max cut on bipartite random d-regular graph (still mostly trees) matches performance of general random d-regular graphs
the performance of bipartite random d-regular graphs is \(1/2 + O(n^{A-1})\), will go to 1/2 for large d.
Implications
constant and shallow depth QAOA on maxcut will only perform at 1/2 plus a d-dependent constant.