quocs-lib 0.0.54

QuOCS (Quantum Optimal Control Suite) library

The optimization library

QuOCS (Quantum Optimal Control Suite) is a python software package for model- and experiment-based optimizations of quantum processes. It uses the excellent Numpy and Scipy packages as numerical backends. QuOCS aims to provide a user-friendly interface to solve optimization problems. A variety of popular optimal control algorithms are available:

• GRAPE (GRadient Ascent Pulse Engineering) Algorithm
• dCRAB (dressed Chopped RAndom Basis) Algorithm
• AD-GRAPE (Automatic Differentiation) Algorithm - for Mac and Linux only
• Direct Search Algorithm, i.e. Nelder Mead, CMA-ES...

QuOCS is open source and its interface structure allows for user-friendly customizability. It can be used on all Unix-based platforms and on Windows.

Installation

QuOCS is available on pip. You can install QuOCS by doing

pip install quocs-lib

The requirements are:

• setuptools >= 44.0.0
• numpy >= 1.19.1
• scipy >= 1.5.1
• If you want to use the AD Algorithm, the installation of JAX (Autograd and XLA) is required. Please note that this package isn't available for Windows.

Editable mode

If you want to customize the algortihm and basis inside QuOCS, the package has to be installed in the editable mode. You can easily do that with the following commands:

git clone https://github.com/Quantum-OCS/QuOCS.git
cd QuOCS
pip install -e .

Documentation

The possible settings for the JSON file can be found here.

You can find the latest development documentation here.

A selection of demonstration notebooks is available, which demonstrate some of the many features of QuOCS. These are stored in the QuOCS/QuOCS-jupyternotebooks repository here on GitHub.

Example of usage

Using QuOCS is intuitive and simple. The main steps are:

1. Create and load the optimization dictionary. This json file contains all the optimization settings (as an example see this file).

   from quocslib.utils.inputoutput import readjson
   optimization_dictionary = readjson("opt_dictionary.json"))
   

2. Create Figure of Merit object. This is an instance of a class that contains the physical problem to be optimized. In the following, you can see an example of how to define this class. The input and output of get_FoM should not be changed.

   from quocslib.utils.AbstractFoM import AbstractFoM
   # Define problem class
   class OneQubit(AbstractFoM):
   
       def __init__(self, args_dict:dict = None):
           """ Initialize the dynamics variables"""
           if args_dict is None:
               args_dict = {}
           ...
   
       def get_FoM(self, pulses: list = [],
                   parameters: list = [],
                   timegrids: list = []
           ) -> dict:
           # Compute the dynamics and FoM
           ...
   
           return {"FoM": fidelity}
   
   # Create Figure of Merit object
   FoM_object = OneQubit()
   

3. Define the optimizer by initializing it with the uploaded optimization dictionary and FoM object. After that the execution can be run.

   from quocslib.Optimizer import Optimizer
   # Define Optimizer
   optimization_obj = Optimizer(optimization_dictionary,
                                FoM_object)
   # Execute the optimization
   optimization_obj.execute()
   

Complete examples are provided in QuOCS/QuOCS-jupyternotebooks repository or in the tests folders.

Usage with Qudi

If you want to use QuOCS in combination with Qudi, please have a look at this repository with additional files, information and a tutorial.

Contribute

Would you like to implement a new algorithm or do you have in mind some new feature it would be cool to have in QuOCS? You are most welcome to contribute to QuOCS development! You can do it by forking this repository and sending pull requests, or filing bug reports at the issues page. All code contributions are acknowledged in the contributors section in the documentation. Thank you for your cooperation!

Citing QuOCS

If you use QuOCS in your research, please cite the original QuOCS papers that are available here.

Authors and contributors

• Marco Rossignolo
• Thomas Reisser
• Alastair Marshall
• Phila Rembold
• Alice Pagano