On the quantum hardness of solving isomorphism problems as nonabelian hidden shift problems

We consider an approach to deciding isomorphism of rigid n-vertex graphs (and
related isomorphism problems) by solving a nonabelian hidden shift problem on a
quantum computer using the standard method. Such an approach is arguably more
natural than viewing the problem as a hidden subgroup problem. We prove that
the hidden shift approach to rigid graph isomorphism is hard in two senses.
First, we prove that Omega(n) copies of the hidden shift states are necessary
to solve the problem (whereas O(n log n) copies are sufficient). Second, we
prove that if one is restricted to single-register measurements, an exponential
number of hidden shift states are required.