pyqrack-cuda 1.75.0

pyqrack - Pure Python vm6502q/qrack Wrapper

pyqrack

Pure Python bindings for the pure C++11/OpenCL Qrack quantum computer simulator library

(PyQrack is just pure Qrack.)

For PyQrack-CUDA, the package is distributed as source-only on PyPi, so that the Qrack build step can do its best to try to automatically detect your available CUDA architectures.

If you're looking for Mac ARM support, use the package pyqrack, not pyqrack-cpu. Mac officially "deprecated" OpenCL years ago. Hence, accelerator support is not included in ARM-based Mac wheels, and OpenCL installation is not required on these systems, but, if you have a CUDA accelerator on ARM-based Mac, you could try the package pyqrack-cuda instead.

Performance can benefit greatly from following the Qrack repository "Quick Start" and "Power user considerations."

If you use only an integrated graphics accelerator, like the Intel HD, you probably can't use this CUDA version of PyQrack. Use the OpenCL version of PyQrack instead (by uninstalling all other versions of PyQrack, then installing from main branch or with pip install pyqrack).

Import and instantiate QrackSimulator instances. This simulator can perform arbitrary single qubit and controlled-single-qubit gates, as well as other specific gates like SWAP.

Any 2x2 bit operator matrix is represented by a list of 4 complex floating point numbers, in row-major order.

Single and array "b" parameters represent Pauli operator bases. They are specified according to the enumeration of the Pauli class.

MC[x] and MAC[x] methods are controlled single bit gates, with as many control qubits as you specify via Python list c argument. MCX is multiply-controlled Pauli X, and MACX is "anti-"controlled Pauli X, i.e. "anti-control" activates the gate if all control bits are specifically off, as opposed to on.

To load the required unitaryfund/qrack libraries from a different location, set the PYQRACK_SHARED_LIB_PATH environment variable.

PyQrack has experimental support for PyZX Circuit definitions as an intermediate representation for QrackSimulator. To try this, load a Circuit in PyZX, (use that module to optimize your circuit, as you like,) and create a QrackSimulator() instance using the pyzxCircuit named argument of the constructor, like so:

sim = QrackSimulator(pyzxCircuit=c)

where c is a PyZX circuit object. The circuit will automatically be simulated in the constructed QrackSimulator instance. This also allows loading from QASM and other intermediate representations supported by PyZX.

See https://pyqrack.readthedocs.io/en/latest/ for an API reference.

For custom Qrack build floating-point precision, where options are half, float, double, and quad, set an environment variable via export QRACK_FPPOW=[n] (or as appropriate to your shell) where [n] is the logarithm base 2 of the number of bits in the systemic floating point type (4, 5, 6, or 7, with 5 or float as default, i.e. 2**5=32 for 32-bit float). Your Qrack installation floating-point build option must match this specific value, which might require a custom Qrack build.

Please feel welcome to open an issue, if you'd like help. 😃

Special thanks go to Zeeshan Ahmed, for bug fixes and design suggestions, Ashish Panigrahi, for documentation and design suggestions, WingCode, for documentation, Or Golan, for CI build pipeline tooling, and to the broader community of Qrack contributors, for years of happy Qracking! You rock!