PySDM_examples.Pyrcel.simulation

  1import numpy as np
  2from PySDM_examples.utils import BasicSimulation
  3
  4from PySDM import Builder
  5from PySDM.backends import CPU
  6from PySDM.backends.impl_numba.test_helpers import scipy_ode_condensation_solver
  7from PySDM.dynamics import AmbientThermodynamics, Condensation
  8from PySDM.environments import Parcel
  9from PySDM.initialisation import equilibrate_wet_radii
 10from PySDM.initialisation.sampling.spectral_sampling import ConstantMultiplicity
 11from PySDM.physics import si
 12
 13
 14class Simulation(BasicSimulation):
 15    def __init__(
 16        self, settings, products=None, scipy_solver=False, rtol_thd=1e-10, rtol_x=1e-10
 17    ):
 18        n_sd = sum(settings.n_sd_per_mode)
 19        builder = Builder(
 20            n_sd=n_sd,
 21            backend=CPU(
 22                formulae=settings.formulae, override_jit_flags={"parallel": False}
 23            ),
 24            environment=Parcel(
 25                dt=settings.timestep,
 26                p0=settings.initial_pressure,
 27                initial_water_vapour_mixing_ratio=settings.initial_vapour_mixing_ratio,
 28                T0=settings.initial_temperature,
 29                w=settings.vertical_velocity,
 30                mass_of_dry_air=44 * si.kg,
 31            ),
 32        )
 33        builder.add_dynamic(AmbientThermodynamics())
 34        builder.add_dynamic(Condensation(rtol_thd=rtol_thd, rtol_x=rtol_x))
 35
 36        volume = (
 37            builder.particulator.environment.mass_of_dry_air
 38            / settings.initial_air_density
 39        )
 40        attributes = {
 41            k: np.empty(0)
 42            for k in ("dry volume", "kappa times dry volume", "multiplicity")
 43        }
 44        for i, (kappa, spectrum) in enumerate(settings.aerosol_modes_by_kappa.items()):
 45            sampling = ConstantMultiplicity(spectrum)
 46            r_dry, n_per_volume = sampling.sample(settings.n_sd_per_mode[i])
 47            v_dry = settings.formulae.trivia.volume(radius=r_dry)
 48            attributes["multiplicity"] = np.append(
 49                attributes["multiplicity"], n_per_volume * volume
 50            )
 51            attributes["dry volume"] = np.append(attributes["dry volume"], v_dry)
 52            attributes["kappa times dry volume"] = np.append(
 53                attributes["kappa times dry volume"], v_dry * kappa
 54            )
 55        r_wet = equilibrate_wet_radii(
 56            r_dry=settings.formulae.trivia.radius(volume=attributes["dry volume"]),
 57            environment=builder.particulator.environment,
 58            kappa_times_dry_volume=attributes["kappa times dry volume"],
 59        )
 60        attributes["volume"] = settings.formulae.trivia.volume(radius=r_wet)
 61
 62        super().__init__(
 63            particulator=builder.build(attributes=attributes, products=products)
 64        )
 65        if scipy_solver:
 66            scipy_ode_condensation_solver.patch_particulator(self.particulator)
 67
 68        self.output_attributes = {
 69            "volume": tuple([] for _ in range(self.particulator.n_sd))
 70        }
 71        self.settings = settings
 72
 73        self.__sanity_checks(attributes, volume)
 74
 75    def __sanity_checks(self, attributes, volume):
 76        for attribute in attributes.values():
 77            assert attribute.shape[0] == self.particulator.n_sd
 78        np.testing.assert_approx_equal(
 79            sum(attributes["multiplicity"]) / volume,
 80            sum(
 81                mode.norm_factor
 82                for mode in self.settings.aerosol_modes_by_kappa.values()
 83            ),
 84            significant=4,
 85        )
 86
 87    def _save(self, output):
 88        for key, attr in self.output_attributes.items():
 89            attr_data = self.particulator.attributes[key].to_ndarray()
 90            for drop_id in range(self.particulator.n_sd):
 91                attr[drop_id].append(attr_data[drop_id])
 92        super()._save(output)
 93
 94    def run(self, observers=()):
 95        for observer in observers:
 96            self.particulator.observers.append(observer)
 97        output_products = super()._run(
 98            self.settings.nt, self.settings.steps_per_output_interval
 99        )
100        return {"products": output_products, "attributes": self.output_attributes}