Skip to content

Generate hierarchical quantum circuits for Neural Architecture Search.

License

Notifications You must be signed in to change notification settings

Aaron-Robertson/hierarqcal

 
 

Repository files navigation

HierarQcal

dalle image

HierarQcal is a quantum circuit builder that simplifies circuit design, composition, generation, scaling, and parameter management. It provides an intuitive and dynamic data structure for constructing computation graphs hierarchically. This enables the generation of complex quantum circuit architectures, which is particularly useful for Neural Architecture Search (NAS), where an algorithm can determine the most efficient circuit architecture for a specific task and hardware. HierarQcal also facilitates the creation of hierarchical quantum circuits, such as those resembling tensor tree networks or MERA, with a single line of code. The package is open-source and framework-agnostic, it includes tutorials for Qiskit, PennyLane, and Cirq. Built to address the unique challenges of applying NAS to Quantum Computing, HierarQcal offers a novel approach to explore and optimize quantum circuit architectures.


A robot building itself with artificial intelligence, pencil drawing - generated with Dall E 2


Unitary Fund

Quick example

Building a Quantum Convolutional Neural Network (QCNN) with one line of code:

from hierarqcal import Qinit, Qcycle, Qmask
hierq = Qinit(8) + (Qcycle(mapping=u) + Qmask("right", mapping=v))*3

Modular and hierarchical circuit building:

### Reverse binary tree
from hierarqcal import Qinit, Qcycle, Qmask
# motif: level 1
m1_1 = Qcycle(stride=2)
m1_2 = Qmask(pattern="left")
# motif: level 2
m2_1 = m1_1 + m1_2
# motif: level 3
m3_1 = Qinit(8) + m2_1 * 3

$m^3_1\rightarrow \text{QCNN}:$

# extending follows naturally, repeating the above circuit 5 times is just:
m3_1 * 5

Installation

PyPI version

HierarQcal is hosted on pypi and can be installed via pip:

# Based on the quantum computing framework you use, choose one of:
pip install hierarqcal[cirq]
# or
pip install hierarqcal[qiskit]
# or
pip install hierarqcal[pennylane]

The package is quantum computing framework independent, there are helper functions for Cirq, Qiskit and Pennylane to represent the circuits in their respective frameworks. You can also use the the package independent of any framework, if you want to do this just run:

pip install hierarqcal

Tutorial and Documentation

There is a quickstart tutorial containing code examples for qiskit, cirq and pennylane:

For an overview of the package there is this blogpost which might be worht a read. Altough the syntax has changed since then, the overall functionality is still the same. There is also this paper on the arXiv which describes some of the use cases of the package. For specific details see the documentation.

Contributing

We welcome contributions to the project. Please see the contribution guidelines and code of conduct for more information.

License

BSD 3-Clause "New" or "Revised" License, see LICENSE for more information.

Citation

@article{lourens2023hierarchical,
      title={Hierarchical quantum circuit representations for neural architecture search},
      url = {https://arxiv.org/abs/2210.15073},
      author={Matt Lourens and Ilya Sinayskiy and Daniel K. Park and Carsten Blank and Francesco Petruccione},
      year={2023},
      eprint={2210.15073},
      archivePrefix={arXiv},
      primaryClass={quant-ph}
}

About

Generate hierarchical quantum circuits for Neural Architecture Search.

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%