The Quantum Approximate Optimization Algorithm Needs to See the Whole Graph: Worst Case Examples

Edward Farhi, David Gamarnik, Sam Gutmann

Outline of argument

1. random d-regular graph, almost all edge-p-neighborhoods are trees if $$(d-1)^{2p} < n^A$$ for $$A < 1$$.
2. Then consider a bipartite graph (completely satisfiable), so the best possible is “cut all edges”
3. so max cut on bipartite random d-regular graph (still mostly trees) matches performance of general random d-regular graphs
4. 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.

Is this true for classical algorithms also?

tags: QAOA  local-algorithms