Workshop on Quantum Time Series Analysis

Author: Jacob Cybulski (ironfrown)
Location: Melbourne, Australia
Date: October 2022

Jacob Cybulski is an independent researcher in quantum computing, quantum machine learning, classical machine learning and data visualisation. He also holds the position of Honorary Associate Professor in Quantum Computing, in the School of IT at Deakin University, Melbourne, Australia.

I am a researcher, and so if you wanted to use this work for your own research, publish your derivative work, or use it for commercial purposes, please acknowledge the author's work and cite this material as follows:

• Jacob Cybulski. “Workshop on Quantum Time Series Analysis.” Ironfrown GitHub repository, Melbourne, Australia, 2022.
  https://github.com/ironfrown/qtsa_workshop.

The workshop material was presented in the following events:

• Jacob Cybulski, keynote "Time Series Analysis Using Quantum Machine Learning", Workshop on Quantum Machine Learning, organised in collaboration with QWorld, QPoland, QIndia and Quantum AI Foundation. In association with IEEE Conference Trends in Quantum Computing and Emerging Business Technologies - TQCEBT 2022, 13 October 2022. Access presentation PDF.
• Jacob Cybulski, "Key Concepts in Quantum Time Series Analysis", School of IT, SEBE, Deakin University, Burwood, Australia, 29 Sept 2022. Access presentation PDF.

This repository consists of examples used in the workshop on Quantum Time Series Analysis (QTSA).

• Workshop aims and objectives
• Prerequsite knowledge
• Materials and notebooks
• Workshop tasks
• Installation instructions for IBMQ Labs
• Software compatibility

Aims and objectives

The workshop provides some introductory material and is delivered via online meetings (e.g. Zoom, Teams or Webex). The workshop is practical and sets the following aim:

To provide the participants with knowledge and skills needed to engineer a quantum solution to practical time series analysis problems

It specifically qualifies its objectives as:

We will not seek to explore quantum advantage of QTSA solutions over classical ones but rather aim to gain experience in quantum manipulation and modelling of time series data.

Prerequsite knowledge

The workshop makes some assumptions as to the audience knowledge and skills, i.e.

• Knowledge of Python
• Basic skills in Qiskit
• Fundamentals of Quantum Computing
• Understanding of Variational Quantum Circuits
• Awareness of Quantum Neural Network techniques

Materials and notebooks

The following files are publicly available:

• utils.py - Utility classes and functions
• qtsa_00_utils_v1.0_demo.ipynb - Tests of support functions
• qtsa_01_linreg_v1.6_demo.ipynb - Demo of quantum linear regression
• qtsa_02_serial_model_demo_v8.9.ipynb - Demo of QTSA with serial Fourier transforms
• qtsa_03_parallel_model_problem_v8.9.ipynb - Problem notebook for QTSA with parallel Fourier transforms
• qtsa_04_qnn_v1.2_demo.ipynb - Demo of QTSA with sliding window and a standard QNN
• qtsa_05_sliding_wind_problem_v9.0.ipynb - Problem notebook for QTSA with serial custom QNN

These are sample solutions to the workshop problems (check them, BUT after solving the problems):

• qtsa_03_parallel_model_answer_v8.9.ipynb - Sample solution for QTSA with parallel Fourier transforms
• qtsa_05_sliding_wind_answer_v9.0.ipynb - Sample solution for QTSA with serial custom QNN

Workshop tasks

Typically the following tasks are undertaken by the workshop participants:

• Easy: Study serial quantum Fourier transform TS fit.
  • Test it with various data sets, factors and optimisers.
  • Analyse, compare and find the best combination.
• Medium: Implement a parallel quantum Fourier transform TS fit.
  • Test it with various data sets, factors and optimisers.
  • Analyse, compare and find the best combination.
• Hard: Implement a serial sliding window QNN TS forecaster.
  • Test it with various data sets, factors and optimisers.
  • Analyse, compare and find the best combination.
• Challenge: Modify the SSW QNN for multi-variate TS data.
  • Test it with various data sets, factors and optimisers.
  • Analyse, compare and find the best combination.

Installation instructions for IBMQ Labs

Instructions on using this repository on IBM Quantum (IBMQ):

• Create an IBMQ account and login
  URL: https://quantum-computing.ibm.com/
• Access ironfrown code on GitHub (here):
  URL: https://github.com/ironfrown/qtsa_workshop
• Download workshop files from GitHub:
  Either via git or from GitHub web site (Code > Download ZIP)
  Save it as "qtsa_workshop-main.zip" (includes the enclosing folder)
• On IBMQ select the Lab option (Top left menu > Lab)
• In the IBMQ Labs file system identify the directory for the contents of the archive
• Upload "qtsa_workshop-main.zip" file into this directory (Labs top left icons > Upload files)
• Open a Python Console (Labs top left icons > New file + > Console)
• Type a statement "!unzip qtsa_workshop-main.zip" (Shit-Enter)
• Refresh the contents of the directory (Labs top left icons > Refresh file list)
• The contents of the workshop repository is ready for use within IBMQ Lab

Software compatibility

Workshop examples have been tested on the following system:

Qiskit Software	Version
qiskit-terra	0.21.0
qiskit-aer	0.10.4
qiskit-ibmq-provider	0.19.2
qiskit	0.37.0
qiskit-nature	0.4.2
qiskit-finance	0.3.3
qiskit-optimization	0.4.0
qiskit-machine-learning	0.4.0
System information
Python version	3.8.13
Python compiler	GCC 7.5.0
OS	Linux
Memory (Gb)	62