Effects of Imperfections on Quantum Algorithms: A Software Engineering Perspective

Install Option 1: Install dependencies in python virtual environment (Recommended)

Prerequisities (Machine Setup)

• Install of python 3.10.6
• Ubuntu 22.04

Required setup steps

0. Clone this repository
1. Create virtual environment

python3 -m venv .venv
source .venv/bin/activate

2. Install requirementes

pip install -r requirements.txt

Install Option 2: Run Docker Container

Prerequisities

• Installation of Docker engine

Required setup steps

• Build docker image docker build -t qsw-repro .
• Run Container docker run --name qsw-repro -it qsw-repro

Folder structure and files

• ./backend_data/ Contains hardware config dictionaries.
• preprocess_data.ipynb Preparation of (simulated) hardware configurations.
• run_simulation.py Contains methods for running the simulations.
• helper_functions.py Includes a number of helper functions.
• pass_defs.py Contains transpiler passes for custom gates.
• gate_defs.py Contains custom gates definitions.
• exec_commands.sh Examples for the command line interface

How to run simulations

Simulations are started using the run_simulation.py script. You can use the command line iterface (see here or start the run_simulation.py script in an IDE of your choice. In this case you have to set the run_configuration dictionary manually (See the comment in the main function in run_simulation.py)

How to extend the framework :

Add your own backend (e.g. based on real or expected future performance of hardware):

1. In preprocess_data.ipynb, add your own custom configuration (The last notebook cell does exactly that) You may also increase, e.g. the coupling density here.
2. Run the cell and your configuration is prepared and ready to use!

Add your own algorithms (purely quantum):

1. Add algorithms specific arguments in run_simulation.py:

• Augment the parse_args method (See existing code for e.g. grover)
• Add parse_yourAlg method (See existing code for e.g. grover)

2. Prepare your quantum circuit (e.g. create a function your_circuit that takes a run_config as argument and returns your circuit; see e.g. grover_lib.py)

3. In the run method:

• Add case in match run_config["algo"] which sets circuit=your_circuit

4. Optional:

• Add custom metrics of success in the 'format_run' method which can be saved later.

Add your own target function for the variational approximator:

1. Add case "yourFun" to match run_config["fun"] in the run_var method of the run_simulation.py script. (see case "powx2") You can take the existing circuit here or add your own.

2. Add a test case in function test_var

3. Edit helper_functions.py and add yor function to target_fun.

Add your own, totally different variational circuit.

This is a little bit more tricky. The area of variational circuits is broad and enables researchers to design miriads of algorithms. Therefore we advice you to inspect our run_var method in run_simulation.py and then create your own run_yourName method. Note that for our toy example we only measured one qubit for an easy-to-grasp illustration, which is not recommended for more sophisticated designs.

Notes about parameters:

1. In the code the hardware config is saved in the dictionary hw_config and basically contains all hardware specific data as basis gates or coupling maps.
2. The run_config dictionary is used to determine the number of qubits or your algorithm etc. Also your noise method can be set here, if you e.g. want to simulate only some level of pauli_x noise.

How to use the CLI

The command line interface allows you to set your backend (hw_config) and parameters of run_config. In the file exec_commands.sh some examples are shown; We advice you to write your own shell script to avoid manual retyping. Also our interface allows you start several simulations, e.g. for different numbers of qubits in parallel or several repetions of the same algorithm. Note that it is your own responsibility to choose these parameters according to your hardware ressources! Be aware that quantum simulation requires a lot of memory.

Where are my results?

We store every run as a single csv_file in sim_results/alg_name/filename.csv. You can merge the csv files and visualize them using a tool of your choice, e.g. some R library of the tidyverse. Also you may want to inspect the density matrix of your simulations which we store in density_results/alg_name/filename.csv. However you'll have to manually activate this in the save_run method in the run_simulation.py script, because it may occupy too much disk space.

Acknowledgements

Next to Qiskit, we thank the creators of various useful python libraries, most importantly numpy and pandas.

License

Apache License 2.0

In the files, we used the SPDX standard for easy License identification https://spdx.dev/ , where we also explicitely higlight copyright holders and adjustments made, to prior work. We also include the longer version, where we derived code from other sources.