Assume $f(x)$ is n-bit to n-bit function. Let $F(x)$ be defined as $T$ iterations of $f(x)$, i.e. $F(x) = f^T(x)$.

Quantum algorithm relies on $F(x)$; it calls it $R$ times. What is the best query complexity of the algorithm in terms of calls to $f(x)$:

  1. Can we do better than $R \cdot T$ queries while maintaining negligible quantum memory complexity?

  2. Can we do better than $R \cdot T$ queries with additional quantum memory? If so, then how much? Can we do less than $T$?


That really depends on the function $f$ and the size of $R\cdot T$. Generically, I don't think that you can expect improvements over $R\cdot T$, but improvements are possible in some special cases.

For example, with the function $f$, there's a similar question in classical, and there are instances where speedups are possible, such as modular exponentiation: $f(x)=x\text{ mod }N$. There are better ways of calculating $x^a\text{ mod }N$ than just calculating $F^a(x)$, but it's specific to that function.

If you want really large $R\cdot T$, or non-integer values, there are quantum algorithms based on phase estimation. You want to have a look at this paper.

  • $\begingroup$ thanks. how about 2. ? $\endgroup$ Jun 3 '19 at 3:47
  • $\begingroup$ @la_guesso34 I am not aware of any results in that direction. It is easy to imagine time improvements via parallelism, but probably using the same number of queries. $\endgroup$
    – DaftWullie
    Jun 3 '19 at 5:37
  • $\begingroup$ The logic is as follows: Unitary operator describing $f$ is simply a matrix M. Then $f^T$ is $M^T$. So, to apply $F$ we just need to apply $\tilde{M} = M^T$. $\endgroup$ Jun 4 '19 at 2:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.