Classical Fourier transformation is a widely used and well established technique. Some examples for this are analyzing solid state material by X-ray crystallography, digital signal processing, or encoding digital signals for internet communication.
Quantum Fourier Analysis is analogous to the classical Fourier transformation, by making use of quantum parallelism. With this we get an exponential speedup. Unfortunately, this directly does not make signal processing faster due to its specific quantum nature.
However, many Quantum Algorithms are relying essentially on Quantum Fourier Transformation (QFT) like e.g., factorization large numbers by Shor’s algorithm [5].
In this Blog we will used QFT to analyze an example piece of music. In doing so,
we want to outline how QFT works in principle. Finally we will link this to arty quantum-images and with this create a film with music and the respective synchronous Quantum analysis.
QFT is an analogue to classical Discrete Fourier transformation. In classical Discrete Fourier Transformation (DFT) you basically take an arbitrary time dependent wave like you see on the left side of the above picture. The amplitude of this wave for a given time interval is defined by a set of numbers x_i. N defines the sampling rate of this wave for a given time interval. In other words, the wave in the time dependent space can be defined by a vector x where x_i in {0, N-1}. If we want to transform x into the frequency space we calculate a new vector y with y_k ; k in {0, N-1} and each element of y is calculated by a sum of discrete sin and cos waves: Quantum Fourier Transformation is almost the same. We take an orthonormal basis |0> … |N-1> and perform a transformation into the Fourier basis.
N = 2 for 1 qubit
N = 4 for 2 qubits
N = 8 for 3 qubits
N= 2^n (n = number of qubits)
This means, that we can in principle perform an exponential number of sampling defined by the qubit count. This seems to be an exponential speedup!
Unfortunately, it is not possible get the QFT coefficients out of the superposition as they are destroyed by a measurement.
The result come from multiple repetition of the measurement and the corresponding statistics as you will see in the below program example.