QuPy
QuPy is a quantum circuit simulator for both CPU and GPU.
The features of QuPy are as follows.
- fast: You can experiment with your idea quickly!
- simple: You can experiment with your idea easily!
QuPy supports both Python 3 and Python 2.
QuPy uses CuPy to support GPU.
Install
pip install qupy
or
pip3 install qupy
or
conda install conda-forge::qupy
GPU
If you use a GPU, install cupy and set the following environment variables.
- QUPY_DTYPE
- Data type of the data array.
complex64orcomplex128orcomplex256- If you does not set QUPY_DTYPE, Data type is set
complex128.
- QUPY_GPU
- GPU machine number (integer).
- Set QUPY_GPU if you use GPU.
- If you don't set QUPY_GPU or you set QUPY_GPU=-1, your program will be running on the CPU.
Read the CuPy documentation for using CuPy.
Documents
https://qupy.readthedocs.io/en/latest/qupy.html
Development
Your contribution is welcome!
Example
>>> import numpy as np
>>> from qupy import Qubits
>>> from qupy.operator import H, X, rz, swap
>>> q = Qubits(3)
>>> print(q.get_state())
[1.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j]
>>> q.gate(H, target=0)
>>> q.gate(H, target=1)
>>> print(q.get_state())
[0.5+0.j 0. +0.j 0.5+0.j 0. +0.j 0.5+0.j 0. +0.j 0.5+0.j 0. +0.j]
>>> q.set_state('011')
>>> print(q.get_state())
[0.+0.j 0.+0.j 0.+0.j 1.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j]
>>> q.gate(X, target=2)
>>> print(q.get_state())
[0.+0.j 0.+0.j 1.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j]
>>> q.set_state([0, 1, 0, 0, 0, 0, 0, 0])
>>> print(q.get_state())
[0.+0.j 1.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j]
>>> q.gate(X, target=2)
>>> print(q.get_state())
[1.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j 0.+0.j]
>>> q.gate(H, target=0)
>>> q.gate(H, target=1)
>>> q.gate(X, target=2, control=(0, 1))
>>> q.gate(X, target=0, control=1, control_0=2)
>>> q.gate(swap, target=(0, 2))
>>> q.gate(rz(np.pi / 8), target=2, control_0=1)
>>> print(q.get_state())
[0.49039264-0.09754516j 0.49039264+0.09754516j 0. +0.j
0.5 +0.j 0. +0.j 0. +0.j
0. +0.j 0.5 +0.j ]
>>> iswap = np.array([[1, 0, 0, 0],
... [0, 0, 1j, 0],
... [0, 1j, 0, 0],
... [0, 0, 0, 1]])
>>> q.gate(iswap, target=(2, 1))
>>> print(q.get_state())
[ 0.49039264-0.09754516j 0. +0.j -0.09754516+0.49039264j
0.5 +0.j 0. +0.j 0. +0.j
0. +0.j 0.5 +0.j ]
>>> res = q.project(target=0)
>>> print(res)
0
>>> print(q.get_state())
[ 0.56625665-0.11263545j 0. +0.j -0.11263545+0.56625665j
0.57735027+0.j 0. +0.j 0. +0.j
0. +0.j 0. +0.j ]
>>> q.gate(H, target=1)
>>> q.gate(swap, target=(2, 0), control=1)
>>> print(q.get_state())
[ 0.32075862+0.32075862j 0.40824829+0.j 0.4800492 -0.4800492j
0. +0.j 0. +0.j 0. +0.j
-0.40824829+0.j 0. +0.j ]
>>> res = q.project(target=1)
>>> print(res)
1
>>> print(q.get_state())
[ 0. +0.j 0. +0.j 0.60597922-0.60597922j
0. +0.j 0. +0.j 0. +0.j
-0.51534296+0.j 0. +0.j ]
>>> hamiltonian = {'XXI': 1, 'IIZ': -0.5}
>>> print(q.expect(hamiltonian))
-0.4999999999999999
>>> hamiltonian = np.kron(np.kron(X, X), X)
>>> print(q.expect(hamiltonian))
0.0
Citation
@Misc{QuPy,
author = {Nakanishi, Ken M},
title = {{QuPy}: A quantum circuit simulator for both CPU and GPU},
year = {2018--},
url = "https://github.com/ken-nakanishi/qupy",
note = {[Online; accessed <today>]}
}
License
MIT License (see LICENSE file).