Skip to content

Latest commit

 

History

History
112 lines (87 loc) · 4.73 KB

README.md

File metadata and controls

112 lines (87 loc) · 4.73 KB
Raw

TenCirChem

TenCirChem

ci codecov pypi doc Binder

English | 简体中文

TenCirChem is an efficient and versatile quantum computation package for molecular properties. TenCirChem is based on TensorCircuit and is optimized for chemistry applications.

TenCirChem-NG is fully compatible with TenCirChem with more new features and bug fixes.

Easy Installation

Getting started with TenCirChem-NG by installing the package via pip:

pip install tencirchem-ng

Some functionality of TenCirChem-NG requires JAX. Please install JAX following installation instructions or simply by pip install tencirchem-ng[jax].

Simple to Use

TenCirChem is written in pure Python, and its use is straightforward. Here's an example of calculating UCCSD:

from tencirchem import UCCSD, M

d = 0.8
# distance unit is angstrom
h4 = M(atom=[["H", 0, 0, d * i] for i in range(4)])

# configuration
uccsd = UCCSD(h4)
# calculate and returns energy
uccsd.kernel()
# analyze result
uccsd.print_summary(include_circuit=True)

Running uccsd.kernel() in the above code determines the optimized circuit ansatz parameters and VQE energy.
TenCirChem also allows the user to supply custom parameters. Here's an example:

import numpy as np

from tencirchem import UCCSD
from tencirchem.molecule import h4

uccsd = UCCSD(h4)
# evaluate various properties based on custom parameters
params = np.zeros(uccsd.n_params)
print(uccsd.statevector(params))
print(uccsd.energy(params))
print(uccsd.energy_and_grad(params))

For more examples and customization, please refer to the documentation

Exciting Features

TenCirChem's features include:

  • Statics module
    • UCC calculation with UCCSD, kUpCCGSD, pUCCD at an extremely fast speed
    • Noisy circuit simulation via TensorCircuit
    • Custom integrals, active space approximation, RDMs, GPU support, etc.
  • Dynamics module
    • Transformation from renormalizer models to qubit representation
    • VQA algorithm based on JAX
    • Built-in models: spin-boson model, pyrazine S1/S2 internal conversion dynamics

Design principle

TenCirChem is designed to be:

  • Fast
    • UCC speed is 10000x faster than other packages
      • Example: H8 with 16 qubits in 2s (CPU). H10 with 20 qubits in 14s (GPU)
      • Achieved by analytical expansion of UCC factors and exploitation of symmetry
  • Easy to hack
    • Avoid defining new classes and wrappers when possible
      • Example: Excitation operators are represented as tuple of int. An operator pool is simply a list of tuple
    • Minimal class inheritance hierarchy: at most two levels
    • Expose internal variables through class attributes

License

TenCirChem is released under Academic Public License. See the LICENSE file for details. In short, you can use TenCirChem-NG freely for non-commercial/academic purpose and commercial use requires a commercial license.

Citing TenCirChem

If this project helps in your research, please cite our software whitepaper:

TenCirChem: An Efficient Quantum Computational Chemistry Package for the NISQ Era

which is also a good introduction to the software.

Research and Applications

Variational basis state encoder

An efficient algorithm to encode phonon states in electron-phonon systems for quantum computation. See examples and the tutorial. Reference paper: https://arxiv.org/pdf/2301.01442.pdf (published in PRR).