xanaduai / sfopenboson

A plugin for Strawberry Fields and OpenFermion, providing methods of simulating bosonic Hamiltonians directly in Strawberry Fields

SFOpenBoson

This Strawberry Fields plugin library allows Strawberry Fields to interface with OpenFermion.

Strawberry Fields is a full-stack Python library for designing, simulating, and optimizing continuous variable (CV) quantum optical circuits.

OpenFermion is an open source package for compiling and analyzing quantum algorithms that simulate fermionic systems.

Features

• Construct bosonic Hamiltonians in OpenFermion, and apply the resulting time propagation using a CV quantum circuit.
• Calculates the time-evolution unitary exactly for Gaussian Hamiltonians – these can then be decomposed into the base CV gate set of Strawberry Fields using the Bloch-Messiah decomposition.
• Particular non-Gaussian gate decompositions, using the Trotter formula, are also supported, including Bose-Hubbard Hamiltonians.
• The Hamiltonians submodule contains important OpenFermion-compatible CV Hamiltonians, including those corresponding to the gate set used in Strawberry Fields.

To get started, please see the online documentation

Installation

Installation of SFOpenBoson, as well as all required Python packages mentioned above, can be done using pip:

$ python -m pip install sfopenboson

Code authors

Josh Izaac.

If you are doing research using Strawberry Fields, please cite our whitepaper:

Nathan Killoran, Josh Izaac, Nicolás Quesada, Ville Bergholm, Matthew Amy, and Christian Weedbrook. Strawberry Fields: A Software Platform for Photonic Quantum Computing. arXiv, 2018. arXiv:1804.03159

Support

• Source Code: https://github.com/XanaduAI/SFOpenBoson
• Issue Tracker: https://github.com/XanaduAI/SFOpenBoson/issues

If you are having issues, please let us know by posting the issue on our Github issue tracker.

We also have a Strawberry Fields Slack channel - come join the discussion and chat with our Strawberry Fields team.

License

SFOpenBoson is free and open source, released under the Apache License, Version 2.0.

sfopenboson's Issues

No module named 'strawberryfields.circuitspecs'

The sfopenboson plugin in Forced quantum harmonic oscillator tutorial here gives the following error

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
<ipython-input-3-c0d26999d831> in <module>()
      2 import strawberryfields as sf
      3 from strawberryfields.ops import *
----> 4 from sfopenboson.ops import GaussianPropagation

/anaconda3/lib/python3.6/site-packages/sfopenboson/ops.py in <module>()
     95 from strawberryfields.backends.shared_ops import sympmat
     96 
---> 97 from strawberryfields.circuitspecs import GaussianSpecs, FockSpecs, TFSpecs
     98 
     99 from .auxillary import trotter_layer, quadratic_coefficients

ModuleNotFoundError: No module named 'strawberryfields.circuitspecs'

when from sfopenboson.ops import GaussianPropagation is imported.

Multiple test functions assume `hbar = 2`

Currently, a large number of test functions assume that hbar=2, and will fail if the default value of HBAR is changed in conftest.py.

These tests should be fixed to make them hbar dependent, and then HBAR set to an unconventional value such as 1.7 to ensure no hbar assumptions are baked in in future.

Bose-Hubbard tests currently fail on Python 3.5 due to quirk in how Python changed dictionary sorting in Python 3.6

Note that this does not affect the library itself, just the way some of the tests perform assertions on Python 3.5.

This can likely be fixed by modifying the tests to compare the keys/values individually, instead of considering the entire object.

VisibleDeprecationWarning from numpy

Note: this is somewhat of a preemptive issue, assuming that the edits from #6 go through

Running:

>>> prog = sf.Program(2)
>>> with prog.context as q:
...     Fock(2) | q[1]
...     BoseHubbardPropagation(H, 1.086, 20) | q

from the Bose Hubbard Demo in the documentation will emit the following warning with numpy v1.21.4

auxillary.py:153: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

I imagine (naively) that the fix involves inserting that dtype=object argument to the two np.array() calls on line 153:

ladders = np.array(list(np.array(group_list, dtype=object)[:, 1]), dtype=object)[:, :, 1]