VQC-Lite
VQC-Lite is an object-oriented Python library for building variational quantum circuits and related tasks in quantum machine learning (QML).
VQC-Lite may not be the most efficient simulator, but it’s definitely an intuitive, reader-friendly and simple-to-use one. With just a few hundred lines of code and full documentation, one could see how a VQC is built layer by layer, from elementary unitary gates all the way up to a differentiable and optimizable quantum circuit.
VQC-Lite provides two alternative ways to construct VQCs, one involving only standard numpy functions, and one using the framework Pennylane. Both implementations give VQCs of similar functionalities with the same I/O. While the latter is currently more efficiently integrated with Jax for circuit optimization by differentiation, the former provides the additional feature of optimization by sweeping.
Author and Developer: Kevin Shen -> Github -> LinkedIn
Installation
Use the package manager pip to install VQC-Lite.
pip install vqc_lite
Use the requirements.txt sheet to install dependencies
pip install -r requirements.txt
Usage
For full documentation, please visit Documentation. To start with, you may read the two notebooks there.
The notebook An Introduction to VQC gives an introduction to the VQC circuit stacks.
The notebook Experiments with VQC gives an illustrative example application for VQC: uploading a picture to a quantum computer.
Below are some examples of how to use VQC-Lite modules.
import numpy as np
from vqc_lite.circuit_stack_pennylane.circuit_mps import MPS_GU2_P
from vqc_lite.experiments.expressibility import Expressibility_Evaluation
from vqc_lite.experiments.state_preparation import Compression_Adam
# initialize a VQC
circuit = MPS_GU2_P(nl=1, nq=4) # 1 layer of general 2 qubit gates covering 4 qubits
# execute a VQC
params = np.random.uniform(size=45) # 3 gates, each with 15 parameters
params = circuit.params_to_proper_shape(params) # reshape the parameters
psi_out = circuit.run_with_param_input(params) # get output state
# state preparation with VQC
ghz = np.zeros(4 * [2]) # Let's compress a GHZ state as an example
ghz[0, 0, 0, 0] = 1 / np.sqrt(2)
ghz[1, 1, 1, 1] = 1 / np.sqrt(2)
task = Compression_Adam(ghz, circuit, steps=1000) # use Adam optimizer
task.run()
# evaluate expressibility
task = Expressibility_Evaluation(circuit)
task.run()
Contributing
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.