atom_classical_mc

Classical Monte Carlo multiphysics simulator for cold atoms: an ensemble of classical point atoms evolved under composable physics modules — optical tweezers (static, moving, astigmatic, gridded), magnetic (Zeeman) potentials, optical dipole beams, and near-resonant photon scattering with internal-state dynamics (a MOT is one configuration of it, not a special case).

The core library lives in the atommc/ package. In the spirit of a COMSOL-style multiphysics model, a simulation is an AtomSystem — a species plus a list of physics modules — handed to the simulate driver:

from atommc import (
    AtomSystem, SimulationConfig, simulate, RB85_D2,
    ZeemanPotential, LightScattering, LightMatterSystem,
    QuadrupoleMagneticField, six_beam_mot, gauss_per_cm, ms,
)

quad = QuadrupoleMagneticField(gradient_T_per_m=gauss_per_cm(10.0))
system = AtomSystem(species=RB85_D2, modules=[
    ZeemanPotential.for_sublevel(quad, g_f=1/3, m_f=3),          # magnetic force
    LightScattering(LightMatterSystem(                            # radiation pressure
        species=RB85_D2, beams=six_beam_mot(detuning_hz=-9.1e6,
        saturation=2.0), magnetic_fields=[quad])),
])
result = simulate(system, SimulationConfig(
    initial_temperature_uK=3000.0, initial_cloud_sigma_m=1e-3,
    timestep_s=5e-8, duration_s=ms(5.0), ensemble_size=400,
))

Modules come in two kinds (see atommc/physics/base.py): ConservativeForce (potential/force/Hessian — traps, Zeeman potentials, dipole beams) and StochasticProcess (per-step velocity kicks with opaque per-atom state — photon scattering, future collisions). The driver never special-cases any physics; new modules plug in without touching the loop. See doc/plan.md for the tweezer model, doc/light_matter_plan.md for the scattering model, and doc/multiphysics_plan.md for the module architecture; atommc/README.md is a one-line index of every function in the package.

Examples

Optical tweezer transfer (SLM to moving AOD trap):

python3 example/aod/slm_to_aod_transfer.py

The example writes separate suffixed figures into example/aod/render/: _traj for trajectory/ramp geometry, _energy for heating/loss/motional occupation distributions, and _3d for the atom trajectories with the AOD center path. (example/aod/transfer_animation.py renders the same transfer as a GIF.) It also prints harmonic radial/axial trap frequencies and motional occupation estimates for atoms decomposed in the initial SLM trap and final AOD trap basis.

By default, the simulator rejects and resamples atoms that are already unbound in the initial trap, so reported loss is conditioned on successful initial loading.

example/aod/ramp_compare.py (--gif) sweeps the position-ramp shape of a moving-AOD drag and reports how the profile trades off peak velocity against heating and survival:

Near-resonant light forces (MOT, molasses, probe beams):

python3 example/mot/rb85_mot.py --save-plot

This writes the summary figure to example/mot/render/rb85_mot_summary.png (add --gif for an animated cooling-cloud GIF alongside it). example/mot/mmwave_mot.py loads a tri-sector grating MOT from a 600 K effusive beam; example/tweezer_probe_heating.py shows the mixed case — recoil heating of a tweezer-trapped atom.

Magnetic and dipole potentials (the newest modules):

python3 example/magnetic_dipole/quadrupole_and_dipole.py

This holds an Rb87 |F=2, m_F=2> cloud in a bare quadrupole magnetic trap (and expels the anti-trapped sublevel), then builds an 850 nm dipole tweezer from power/waist/wavelength and cross-checks it against an equivalent hand-tuned GaussianTrap.