garymooney commented 2 years ago

Abstract

Quantum computing is notorious for being unintuitive and difficult to imagine. This project aims to create some kind of connection between a human observer and the complex workings of a quantum algorithm.

To do this, we will express the process of quantum computation as art, in the form of music, video... or a music video!

I mean, this does kind of remind you of a music score right?

Description

The idea is to come up with a way to convert properties of quantum algorithms at each time step of the computation to a (hopefully) tasteful form of visual and audio output.

By using the qiskit simulators, we could query the quantum state at each time step of the quantum algorithm and extract properties. These properties could be used to determine the pixels and MIDI sounds (a form of coding music).

An example of how to do this could be the following (we can chat about other ideas that could work too):

Each state from the simulator can be a note of some musical scale and a superposition of states get played as a chord.
We could extend this by adding noise to the simultations. Using a noisy density matrix simulator, each pure state in the probabilistic mixture of pure states can correspond to their own musical scale (or same scale with offset pitch), with note volumes being proportional to the corresponding state probability in the density matrix. This way, noise in the quantum system will be expressed as notes from different scales, creating a feeling of decoherence in the music.
For the video, qubits in the IBM Q architectures can be mapped to pixel locations, then the RGB values could be determined at each time step by measuring the qubits in the X, Y and Z bases.
As we go, we can iteratively adjust how the data maps to video and audio to improve the output quality for general quantum circuits.

Tools

Sound: The Mido python library (https://mido.readthedocs.io/en/latest/) can be used to create MIDI files (a form of coding music). MIDI files can be rendered to other formats using the midi2audio lib https://github.com/bzamecnik/midi2audio

Image: The png python library (https://pypng.readthedocs.io/en/latest/png.html) can be used to create images using pixel RGB values.

Video: The MoviePy python library (https://pypi.org/project/opencv-python/) can be used to read the images from a folder and turn them into a video, or images can be specified as numpy arrays.

Members

Qiskit Coach: @garymooney - email: mooneyg@unimelb.edu.au @yy00770 @Gyp1127

The project would benefit from a variety of backgrounds. Technical artists or anyone with experience in sound/music or generating MIDI files would be especially fitting.