qudit 0.4.1

qudit

High performance simulations for qudit systems. To make qudit machine learning, qudit error correction, and qudit circuit simulation easier. Qudit is made fully around pytorch to make it easy to mix and match tools without worrying about type errors.

pip install qudit

Quickstart

In most cases it should not matter if you mix and match pytorch with qudit since most abstractions are built on top of pytorch tensors. The following is two examples to do the same thing, one using the Circuit class and the other manually using the matrices.

Using the Circuit class:

from qudit import Circuit
import torch as pt

C = Circuit(2, dim=2)  # 2 qBits with d=2
G = C.gates[2]

C.gate(G.H, dits=[0])
C.gate(G.CX, dits=[0, 1])

ket0 = pt.zeros(2**2)
ket0[0] = 1.0  # |00>

print(C(ket0))  # [1. 0. 0. 1.]/rt2

Contributing

Qudit has a small shell scripts used to generate documentation and run tests. If you want to contribute, please fork the repo and make a pull request with your changes. The running script is do.

Commands

• ./do help
  • Helper for common workflows (build, tests, docs, etc.)
• ./do build
  • Builds sdist + wheel and runs twine check dist/*.
• ./do deploy
  • Uploads dist/* to PyPI via twine upload using an API token.
  • Token is read from FILE (defaults to .vscode/token.env).
• ./do test
  • Runs the Python test scripts in ./tests.
• ./do prof
  • Profiles ./tests/bench_fast.py with cProfile and opens snakeviz.
• ./do head
  • Runs python ./view/headers.py (used by the docs pipeline).
• ./do docs dev
  • Runs the docs dev server (npm run dev).
• ./do docs build
  • Builds the docs (npm run build).

Common workflows

# run test scripts
./do test

# build packages locally
./do build

# build docs
./do docs build

# docs dev server
./do docs dev

PyPI token setup (for deploy)

Create a file containing only your PyPI API token (no export, no quotes):

• default location: .vscode/token.env
• or specify a different file via FILE=path/to/token.env

Example:

FILE=.vscode/token.env ./do deploy

Acknowledgements

Many many thanks to Sai Sakunthala and R-Phoenix for testing and fixing bugs!

This library is built on top of pytorch, and takes ideas from Qiskit, Cirq, and QuDiet.

Caveats

qudit does not, and will not support openQASM in the near future since hardware is not a goal. There are plans for some amount of interoperability with other qudit simulators, however that will be a long term goal and not a near term one. We will first make qudit stable.