spylind 0.22

Simple definition and fast solution of quantum systems described by a Lindblad master equation.

Spylind is a Python library for the symbolic definition and high-performance solution of large ensembles of quantum systems described by the Lindblad master equation.

It is particularly optimized for ensembles with many degrees of freedom (e.g., inhomogeneous broadening), utilizing JAX and TensorFlow for JIT compilation and GPU acceleration.

Core Components

• spylind: Generates symbolic equations of motion (EOM) for density matrix elements from a Hamiltonian and collapse operators.

• spyIVP: A high-performance bridge between symbolic expressions and numerical ODE solvers.

Key Features

• Symbolic-to-Numerical: Automates the transition from SymPy expressions to optimized solver code.

• Multiple Backends:

  • Diffrax (JAX): High-performance, JIT-compiled, and GPU-capable.

  • NumPy (SciPy): Robust, industry-standard CPU integration.

  • TensorFlow: scalable integration for large-scale problems.

• Multi-dimensional Ensembles: Native support for parameter sweeps and spatial dimensions.

• QuTiP Integration: Supports QuTiP Qobj for operator definitions.

Installation

pip install spylind[jax]  # Recommended for performance
pip install spylind[tf]   # For TensorFlow support
pip install spylind[full] # All backends

Quick Example

from spylind import spylind as spl
import qutip as q
import numpy as np
import sympy as sm

# Define a 2-level system symbolically
H = [0.1 * np.pi * q.sigmaz(), [sm.symbols('Omega')/2, q.sigmax()]]
tlist = np.linspace(0, 1.0, 101)

# Solve using the Diffrax backend
res = spl.mesolve(H, q.basis(2,0), tlist,
                  t_dep_fL={'Omega': lambda t: 2*np.pi},
                  e_ops=[q.sigmaz()],
                  backend='diffrax')

• License: 3-clause BSD