gspsa-gradients 1.0.0

Implementation of Guided-SPSA

Guide-SPSA Gradients

This repository implements the Guided-SPSA gradients introduced in "Guided-SPSA: Simultaneous Perturbation Stochastic Approximation assisted by the Parameter Shift Rule by M. Periyasamy et. al."

Features

• Integration: Easily integrate with Qiskit and Tensorflow-Quantum APIs. The code inherits the respective differentiators provided by the APIs. Hence, one can use this implementation in any standard machine learning work that deals with gradient estimation via "expectation values" in Qiskit or TensorFlow-Quanutm.

Setup and Installation

The library requires an installation of python > 3.9 , and following libraries:

• Qiskit version
  • qiskit
  • qiskit-algorithms
• Tensorflow-Quantum version
  • tensorflow
  • tesnroflow-quantum

The package can be installed via pip:

• package without other dependencies

  pip install gspsa-gradients
  

• To install the with qiskit dependency:

  pip install gspsa-gradients[qiskit]
  

• To install the with tfq dependency:

  pip install gspsa-gradients[tfq]
  

• The package gspsa-gradients can be installed locally via:

  git clone https://github.com/maniraman-periyasamy/gspsa-gradients.git
  cd gspsa-gradients
  pip install -e .
  

Usage and Examples

Usage

• Qiskit version

  from gspsa_gradients.qiskit_gradient import GSPSAEstimatorGradient
  from qiskit.primitives import Estimator
  gradient = GSPSAEstimatorGradient(total_steps = num_epochs, estimator=Estimator(), num_observables = len(observables), 
      tau=0.5, spsa_epsilon=0.01, damping_coeff=0.5)
  qnn1 = EstimatorQNN(
      circuit=qc, input_params=inputParams, weight_params=circuitParams, observables=observables, gradient=gradient, input_gradients=False)
  
  ...
  

• TFQ verison

  from gspsa_gradients.tfq_gradient import GSPSAGradient
  diff=GSPSAGradient(total_steps=num_epochs, input_dims=num_inputs, tau=0.5, spsa_epsilon=0.01, damping_coeff=0.5)
  quantum_layer = tfq.layers.ControlledPQC(circuit, observables, differentiator=diff, repetitions=1024) # make repitions to 0 for exact expectation value estimation 
  
  ...
  

Example

Qml examples are provided in the examples folder

Acknowledgements

We use qiskit software framework: https://github.com/Qiskit and tensorflow-quantumsoftware framework: https://github.com/tensorflow/quantum

Citation

If you use the gspsa-gradients or results from the paper, please cite our work as

@misc{periyasamy2024guidedspsa,
      title={Guided-SPSA: Simultaneous Perturbation Stochastic Approximation assisted by the Parameter Shift Rule}, 
      author={Maniraman Periyasamy and Axel Plinge and Christopher Mutschler and Daniel D. Scherer and Wolfgang Mauerer},
      year={2024},
      eprint={2404.15751},
      archivePrefix={arXiv},
      primaryClass={quant-ph}
}

Version History

Initial release (v0.1.0): May 2024

License

Apache 2.0 License