tensorcircuit

English | 简体中文

TensorCircuit is the next generation of quantum software framework with support for automatic differentiation, just-in-time compiling, hardware acceleration, and vectorized parallelism.

TensorCircuit is built on top of modern machine learning frameworks: Jax, TensorFlow, and PyTorch. It is specifically suitable for highly efficient simulations of quantum-classical hybrid paradigm and variational quantum algorithms in ideal, noisy and approximate cases. It also supports real quantum hardware access and provides CPU/GPU/QPU hybrid deployment solutions since v0.9.

Getting Started

Please begin with Quick Start in the full documentation.

For more information on software usage, sota algorithm implementation and engineer paradigm demonstration, please refer to 70+ example scripts and 30+ tutorial notebooks. API docstrings and test cases in tests are also informative.

The following are some minimal demos.

Circuit manipulation:

import tensorcircuit as tc
c = tc.Circuit(2)
c.H(0)
c.CNOT(0,1)
c.rx(1, theta=0.2)
print(c.wavefunction())
print(c.expectation_ps(z=[0, 1]))
print(c.sample(allow_state=True, batch=1024, format="count_dict_bin"))

Runtime behavior customization:

tc.set_backend("tensorflow")
tc.set_dtype("complex128")
tc.set_contractor("greedy")

Automatic differentiations with jit:

def forward(theta):
    c = tc.Circuit(2)
    c.R(0, theta=theta, alpha=0.5, phi=0.8)
    return tc.backend.real(c.expectation((tc.gates.z(), [0])))

g = tc.backend.grad(forward)
g = tc.backend.jit(g)
theta = tc.array_to_tensor(1.0)
print(g(theta))

More highlight features for TensorCircuit (click for details)

Sparse Hamiltonian generation and expectation evaluation:

n = 6
pauli_structures = []
weights = []
for i in range(n):
    pauli_structures.append(tc.quantum.xyz2ps({"z": [i, (i + 1) % n]}, n=n))
    weights.append(1.0)
for i in range(n):
    pauli_structures.append(tc.quantum.xyz2ps({"x": [i]}, n=n))
    weights.append(-1.0)
h = tc.quantum.PauliStringSum2COO(pauli_structures, weights)
print(h)
# BCOO(complex64[64, 64], nse=448)
c = tc.Circuit(n)
c.h(range(n))
energy = tc.templates.measurements.operator_expectation(c, h)
# -6

Large-scale simulation with tensor network engine

# tc.set_contractor("cotengra-30-10")
n=500
c = tc.Circuit(n)
c.h(0)
c.cx(range(n-1), range(1, n))
c.expectation_ps(z=[0, n-1], reuse=False)

Density matrix simulator and quantum info quantities

c = tc.DMCircuit(2)
c.h(0)
c.cx(0, 1)
c.depolarizing(1, px=0.1, py=0.1, pz=0.1)
dm = c.state()
print(tc.quantum.entropy(dm))
print(tc.quantum.entanglement_entropy(dm, [0]))
print(tc.quantum.entanglement_negativity(dm, [0]))
print(tc.quantum.log_negativity(dm, [0]))

Install

The package is written in pure Python and can be obtained via pip as:

pip install tensorcircuit

We recommend you install this package with tensorflow also installed as:

pip install tensorcircuit[tensorflow]

Other optional dependencies include [torch], [jax], [qiskit] and [cloud].

We also have Docker support.

Advantages

Tensor network simulation engine based
JIT, AD, vectorized parallelism compatible
GPU support, quantum device access support, hybrid deployment support
Efficiency
- Time: 10 to 10^6+ times acceleration compared to TensorFlow Quantum, Pennylane or Qiskit
- Space: 600+ qubits 1D VQE workflow (converged energy inaccuracy: < 1%)
Elegance
- Flexibility: customized contraction, multiple ML backend/interface choices, multiple dtype precisions, multiple QPU providers
- API design: quantum for humans, less code, more power
Batteries included
Tons of amazing features and built in tools for research (click for details)
- Support super large circuit simulation using tensor network engine.
- Support noisy simulation with both Monte Carlo and density matrix (tensor network powered) modes.
- Support approximate simulation with MPS-TEBD modes.
- Support analog/digital hybrid simulation (time dependent Hamiltonian evolution, pulse level simulation) with neural ode modes.
- Support Fermion Gaussian state simulation with expectation, entanglement, measurement, ground state, real and imaginary time evolution.
- Support qudits simulation.
- Support parallel quantum circuit evaluation across multiple GPUs.
- Highly customizable noise model with gate error and scalable readout error.
- Support for non-unitary gate and post-selection simulation.
- Support real quantum devices access from different providers.
- Scalable readout error mitigation native to both bitstring and expectation level with automatic qubit mapping consideration.
- Advanced quantum error mitigation methods and pipelines such as ZNE, DD, RC, etc.
- Support MPS/MPO as representations for input states, quantum gates and observables to be measured.
- Support vectorized parallelism on circuit inputs, circuit parameters, circuit structures, circuit measurements and these vectorization can be nested.
- Gradients can be obtained with both automatic differenation and parameter shift (vmap accelerated) modes.
- Machine learning interface/layer/model abstraction in both TensorFlow and PyTorch for both numerical simulation and real QPU experiments.
- Circuit sampling supports both final state sampling and perfect sampling from tensor networks.
- Light cone reduction support for local expectation calculation.
- Highly customizable tensor network contraction path finder with opteinsum interface.
- Observables are supported in measurement, sparse matrix, dense matrix and MPO format.
- Super fast weighted sum Pauli string Hamiltonian matrix generation.
- Reusable common circuit/measurement/problem templates and patterns.
- Jittable classical shadow infrastructures.
- SOTA quantum algorithm and model implementations.
- Support hybrid workflows and pipelines with CPU/GPU/QPU hardware from local/cloud/hpc resources using tf/torch/jax/cupy/numpy frameworks all at the same time.

Contributing

Status

This project is created and maintained by Shi-Xin Zhang with current core authors Shi-Xin Zhang and Yu-Qin Chen. We also thank contributions from the open source community.