We study the efficiency of quantum algorithms which aim at obtaining phase space distribution functions of quantum systems. Wigner and Husimi functions are considered. Different quantum algorithms are envisioned to build these functions, and compared with the classical computation. Different procedures to extract more efficiently information from the final wave function of these algorithms are studied, including coarse-grained measurements, amplitude amplification and measure of wavelet-transformed wave function. The algorithms are analyzed and numerically tested on a complex quantum system showing different behavior depending on parameters, namely the kicked rotator. The results for the Wigner function show in particular that the use of the quantum wavelet transform gives a polynomial gain over classical computation. For the Husimi distribution, the gain is much larger than for the Wigner function, and is bigger with the help of amplitude amplification and wavelet transforms. We also apply the same set of techniques to the analysis of real images. The results show that the use of the quantum wavelet transform allows to lower dramatically the number of measurements needed, but at the cost of a large loss of information.
I.INTRODUCTIONIn the recent years, the study of quantum information [1] has attracted more and more interest. In this field, quantum mechanics is used to treat and manipulate information. Important applications are quantum cryptography, quantum teleportation and quantum computation. The latter takes advantage of the laws of quantum mechanics to perform computational tasks sometimes much faster than classical devices. A famous example is provided by the problem of factoring large integers, useful for public-key cryptography, which can be solved with exponential efficiency by Shor's algorithm [2]. Another example is the search of an unstructured list, which was shown by Grover [3] to be quadratically faster on quantum devices. In parallel, investigations of the simulation of quantum systems on quantum computers showed that the evolution of a complex wave function can be simulated efficiently for an exponentially large Hilbert space with polynomial resources [4,5,6,7,8,9]. Still, there are many open questions which remain unanswered. In particular, it is not always clear how to perform an efficient extraction of information from such a complex quantum mechanical wave function once it has been evolved on a quantum computer. More generally, the same problem appears for quantum algorithms manipulating large amount of classical data.In the present paper, we study different algorithmic processes which perform this task. We focus on the phase space distribution (Wigner and Husimi functions) [10,11] These functions provide a two-dimensional picture of a one-dimensional wave function, and can be compared directly with classical phase space distributions. They have also been shown in [12,13] to be stable with respect to various quantum computer error models. Different phase space representation which can be...