1: \begin{abstract}

2: It has been known for some time that graph isomorphism reduces to

3: the hidden subgroup problem (HSP).  What is more, most exponential

4: speedups in quantum computation are obtained by solving instances of

5: the HSP.  A common feature of the resulting algorithms is the use of

6: quantum coset states, which encode the hidden subgroup.  An open

7: question has been how hard it is to use these states to solve graph

8: isomorphism.  It was recently shown by Moore, Russell, and

9: Schulman~\cite{MooreRS05} that only an exponentially small amount of

10: information is available from one, or a pair of coset states.  A

11: potential source of power to exploit are entangled quantum

12: measurements that act jointly on many states at once.  We show that

13: entangled quantum measurements on at least $\Omega(n \log n)$ coset

14: states are necessary to get useful information for the case of graph

15: isomorphism, matching an information theoretic upper bound. This may

16: be viewed as a negative result because highly entangled measurements

17: seem hard to implement in general.  Our main theorem is very general

18: and also rules out using joint measurements on few coset states for

19: some other groups, such as $\GL(n,\FF_{p^m})$ and $G^n$ where $G$ is

20: finite and satisfies a suitable property.

21: \end{abstract}

22: