rovibrational-excitation 0.1.2

Rovibrational wave-packet simulation toolkit

rovibrational-excitation

High-performance Python package for time-dependent quantum dynamics of linear molecules (rotation × vibration) driven by femtosecond–picosecond laser pulses.

CPU / GPU (CuPy)	Numba-JIT RK4 propagator	Lazy, cached dipole matrices

---

Key features

• Runge–Kutta 4 (RK-4) propagators for the Schrödinger and Liouville–von Neumann equations (complex128, cache-friendly).
• Lazy, high-speed construction of transition-dipole matrices
  (rovibrational_excitation.dipole.*)
  • rigid-rotor + harmonic / Morse vibration
  • Numba (CPU) or CuPy (GPU) backend
• Vector electric-field objects with Gaussian envelopes, chirp, optional sinusoidal modulation.
• Batch runner for pump–probe / parameter sweeps with automatic directory creation, progress-bar and compressed output (.npz).
• 100 % pure-Python, no compiled extension to ship (Numba compiles at runtime).

---

Installation

# From PyPI  (stable)
pip install rovibrational-excitation          # installs sub-packages as well

# Or from GitHub (main branch, bleeding-edge)
pip install git+https://github.com/yourname/rovibrational-excitation.git

CuPy (optional) – for GPU acceleration

pip install cupy-cuda12x     # pick the wheel that matches your CUDA

---

Quick start : library API

import numpy as np
import rovibrational_excitation as rve

# --- 1. Basis & dipole matrices ----------------------------------
c_vacuum = 299792458 * 1e2 / 1e15  # cm/fs
debye_unit = 3.33564e-30                       # 1 D → C·m
Omega01_rad_phz = 2349*2*np.pi*c_vacuum
Delta_omega_rad_phz = 25*2*np.pi*c_vacuum
B_rad_phz = 0.39e-3*2*np.pi*c_vacuum
Mu0_Cm = 0.3 * debye_unit                      # 0.3 Debye 相当
Potential_type = "harmonic"  # or "morse"
V_max = 2
J_max = 4

basis = rve.LinMolBasis(
            V_max=V_max,
            J_max=J_max,
            use_M = True,
            omega_rad_phz = Omega01_rad_phz,
            delta_omega_rad_phz = Delta_omega_rad_phz
            )           # |v J M⟩ direct-product

dip   = rve.LinMolDipoleMatrix(
            basis, mu0=Mu0_Cm, potential_type=Potential_type,
            backend="numpy", dense=True)            # CSR on GPU

mu_x  = dip.mu_x            # lazy-built, cached thereafter
mu_y  = dip.mu_y
mu_z  = dip.mu_z

# --- 2. Hamiltonian ----------------------------------------------
H0 = rve.generate_H0_LinMol(
        basis,
        omega_rad_phz       = Omega01_rad_phz,
        delta_omega_rad_phz = Delta_omega_rad_phz,
        B_rad_phz           = B_rad_phz,
)

# --- 3. Electric field -------------------------------------------
t  = np.linspace(-200, 200, 4001)                   # fs
E  = rve.ElectricField(tlist=t)
E.add_Efield_disp(
        envelope_func=rve.core.electric_field.gaussian_fwhm,
        duration=50.0,             # FWHM (fs)
        t_center=0.0,
        carrier_freq=2349*2*np.pi*c_vacuum,   # rad/fs
        amplitude=1.0,
        polarization=[1.0, 0.0],   # x-pol.
)

# --- 4. Initial state |v=0,J=0,M=0⟩ ------------------------------
from rovibrational_excitation.core.states import StateVector
psi0 = StateVector(basis)
psi0.set_state((0,0,0), 1.0)
psi0.normalize()

# --- 5. Time propagation (Schrödinger) ---------------------------
psi_t = rve.schrodinger_propagation(
            H0, E, dip,
            psi0.data,
            axes="xy",              # Ex→μx, Ey→μy
            sample_stride=10,
            backend="numpy")        # or "cupy"

population = np.abs(psi_t)**2
print(population.shape)            # (Nt, dim)

---

Quick start : batch runner

1. Create a parameter file (params_CO2.py)

# description is used in results/<timestamp>_<description>/
description = "CO2_antisymm_stretch"

# --- time axis (fs) ---------------------------------------------
t_start, t_end, dt = -200.0, 200.0, 0.1       # Unit is fs

# --- electric-field scan ----------------------------------------
duration       = [50.0, 80.0]                 # Gaussian FWHM (fs)
polarization   = [[1,0], [1/2**0.5,1j/2**0.5]]
t_center       = [0.0, 100.0]

carrier_freq   = 2349*2*np.pi*1e12*1e-15      # rad/fs
amplitude      = 1.0e9                        # V/m

# --- molecular constants ----------------------------------------
V_max, J_max   = 2, 4
omega_rad_phz  = carrier_freq * 2 * np.pi
mu0_Cm         = 0.3 * 3.33564e-30            # 0.3 D

2. Run

python -m rovibrational_excitation.simulation.runner \
       examples/params_CO2.py     -j 4      # 4 processes

• Creates results/YYYY-MM-DD_hh-mm-ss_CO2_antisymm_stretch/…
• For each case a folder with result.npz, parameters.json
• Top-level summary.csv (final populations etc.)

Add --dry-run to just list cases without running.

---

Directory layout (after refactor)

rovibrational_excitation/
  __init__.py                # public re-export
  core/                      # low-level numerics
    basis.py, propagator.py, ...
  dipole/
    linmol/                  # high-level dipole API
      builder.py, cache.py
    rot/                     # rotational TDM formulae
      jm.py, j.py
    vib/
      harmonic.py, morse.py
  plots/                     # helper scripts (matplotlib)
  simulation/                # batch manager, CLI

---

Development

git clone https://github.com/yourname/rovibrational-excitation.git
cd rovibrational-excitation
python -m venv .venv && source .venv/bin/activate
pip install -r requirements-dev.txt
pytest -v

Black + Ruff + MyPy configs are in pyproject.toml.

---

License

MIT

© 2025 Hiroki Tsusaka. All rights reserved.