Abstract
A line of work initiated by Terhal and DiVincenzo [Terhal/DiVincenzo, arXiv, 2002] and Bremner, Jozsa, and Shepherd [Bremner/Jozsa/Sheperd, Proc. Royal Soc. A, 2010], shows that restricted classes of quantum computation can efficiently sample from probability distributions that cannot be exactly sampled efficiently on a classical computer, unless the PH collapses. Aaronson and Arkhipov [Aaronson/Arkhipov, J. Theory of Comp., 2013] take this further by considering a distribution that can be sampled efficiently by linear optical quantum computation, that under two feasible conjectures, cannot even be approximately sampled within bounded total variation distance, unless the PH collapses.
In this work we use Quantum Fourier Sampling to construct a class of distributions that can be sampled exactly by a quantum computer. We then argue that these distributions cannot be approximately sampled classically, unless the PH collapses, under variants of the AaronsonArkhipov conjectures.
In particular, we show a general class of quantumly sampleable distributions each of which is based on an "Efficiently Specifiable" polynomial, for which a classical approximate sampler implies an averagecase approximation. This class of polynomials contains the Permanent but also includes, for example, the Hamiltonian Cycle polynomial, as well as many other familiar #Phard polynomials.
Since our distribution likely requires the full power of universal quantum computation, while the AaronsonArkhipov distribution uses only linear optical quantum computation with noninteracting bosons, why is our result interesting? We can think of at least three reasons:
1. Since the conjectures required in [Aaronson/Arkhipov, J. Theory of Comp., 2013] have not yet been proven, it seems worthwhile to weaken them as much as possible. We do this in two ways, by weakening both conjectures to apply to any "Efficiently Specifiable" polynomial, and by weakening the socalled AntiConcentration conjecture so that it need only hold for one distribution in a broad class of distributions.
2. Our construction can be understood without any knowledge of linear optics. While this may be a disadvantage for experimentalists, in our opinion it results in a very clean and simple exposition that may be more immediately accessible to computer scientists.
3. It is extremely common for quantum computations to employ “Quantum Fourier Sampling” in the following way: first apply a classically efficient function to a uniform superposition of inputs, then apply a Quantum Fourier Transform followed by a measurement. Our distributions are obtained in exactly this way, where the classically efficient function is related to a (presumed) hard polynomial. Establishing rigorously a robust sense in which the central primitive of Quantum Fourier Sampling is classically hard seems a worthwhile goal in itself.
BibTeX  Entry
@InProceedings{fefferman_et_al:LIPIcs:2016:6682,
author = {Bill Fefferman and Christopher Umans},
title = {{On the Power of Quantum Fourier Sampling}},
booktitle = {11th Conference on the Theory of Quantum Computation, Communication and Cryptography (TQC 2016)},
pages = {1:11:19},
series = {Leibniz International Proceedings in Informatics (LIPIcs)},
ISBN = {9783959770194},
ISSN = {18688969},
year = {2016},
volume = {61},
editor = {Anne Broadbent},
publisher = {Schloss DagstuhlLeibnizZentrum fuer Informatik},
address = {Dagstuhl, Germany},
URL = {http://drops.dagstuhl.de/opus/volltexte/2016/6682},
URN = {urn:nbn:de:0030drops66829},
doi = {10.4230/LIPIcs.TQC.2016.1},
annote = {Keywords: Quantum Complexity Theory, Sampling Complexity}
}
Keywords: 

Quantum Complexity Theory, Sampling Complexity 
Collection: 

11th Conference on the Theory of Quantum Computation, Communication and Cryptography (TQC 2016) 
Issue Date: 

2016 
Date of publication: 

22.09.2016 