def prop1d(mode,
dielectric,
max_harmonic,
max_power,
has_main_phase=False,
optimize_z=True):
propagator1d = prop.ParaxialPerpendicularPropagator1D(
p1d,
dielectric,
mode,
direction=-1,
max_harmonic=max_harmonic,
max_power=max_power)
E = propagator1d.propagate(omega,
x_start,
x_end,
nx,
E_start,
Y1D,
Z1D,
mute=False,
debug_mode=True,
include_main_phase=has_main_phase,
optimize_z=optimize_z)
return (E, propagator1d)
def error_1d(nx_power_min,
nx_power_max,
nx_power_step,
mode='O',
dielectric=dt.ColdElectronColdIon,
max_harmonic=2,
max_power=2,
has_main_phase=False):
"""
nx_step: the step size in nx power
"""
nx_power_array = np.arange(nx_power_min, nx_power_max + 1, nx_power_step)
nx_array = 2**nx_power_array
x_stepsize = np.float(np.abs(x_end - x_start)) / nx_array
p1 = prop.ParaxialPerpendicularPropagator1D(p1d,
dielectric,
mode,
direction=-1,
max_harmonic=max_harmonic,
max_power=max_power)
max_abs_err = np.zeros_like(nx_array, dtype='float')
max_rel_err = np.zeros_like(max_abs_err)
tstart = time.clock()
E_mark = p1.propagate(omega,
x_start,
x_end,
nx_array[-1],
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
for i, ni in enumerate(nx_array[:-1]):
Ei = p1.propagate(omega,
x_start,
x_end,
ni,
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
step = nx_array[-1] / nx_array[i]
E_marki = E_mark[..., ::step]
abs_err = np.abs(Ei - E_marki)
max_abs_err[i] = np.max(abs_err)
rel_idx = np.abs(E_marki) > 1e-3 * np.max(np.abs(E_marki))
max_rel_err[i] = np.max(np.abs(abs_err[rel_idx] / E_marki[rel_idx]))
tend = time.clock()
print('Total time used: {:.4}s'.format(tend - tstart))
return max_abs_err, max_rel_err, nx_array, x_stepsize
def compare_1d2d(mode,
dielectric,
max_harmonic,
max_power,
has_main_phase=False):
propagator1d = prop.ParaxialPerpendicularPropagator1D(
p1d,
dielectric,
mode,
direction=-1,
max_harmonic=max_harmonic,
max_power=max_power)
E1 = propagator1d.propagate(omega,
x_start,
x_end,
nx,
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
propagator2d_uni = prop.ParaxialPerpendicularPropagator2D(
p2d_uni,
dielectric,
mode,
direction=-1,
ray_y=0,
max_harmonic=max_harmonic,
max_power=max_power)
E2d_uni = propagator2d_uni.propagate(omega,
x_start,
x_end,
nx,
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
return (E1, propagator1d, E2d_uni, propagator2d_uni)
def benchmark_1d2d(nx_power_min,
nx_power_max,
nx_step,
mode='O',
dielectric=dt.ColdElectronColdIon,
max_harmonic=2,
max_power=2,
has_main_phase=False):
"""calculate error convergence against x step size
:param float nx_min: log2(nx) minimum, start point of log mesh of total x
steps.
:param float nx_max: log2(nx) maximum, end point of log mesh of total x
steps.
:param int nx_step: step size in power array
:param string mode: polarization mode
:param dielectric: dielectric tensor type
:type dielectric: :py:class:`sdp.Plamsa.DielectricTensor.Dielectric`
:param int max_harmonic: highest harmonic resonance included
:param int max_power: highest FLR effect power included
"""
tstart = time.clock()
nx_power = np.arange(nx_power_min, nx_power_max + 1, nx_step)
nx_array = 2**nx_power
x_stepsize = np.float(np.abs(x_end - x_start)) / nx_array
p1 = prop.ParaxialPerpendicularPropagator1D(p1d,
dielectric,
mode,
direction=-1,
max_harmonic=max_harmonic,
max_power=max_power)
p2 = prop.ParaxialPerpendicularPropagator2D(p2d_uni,
dielectric,
mode,
direction=-1,
ray_y=0,
max_harmonic=max_harmonic,
max_power=max_power)
max_abs_err = np.empty_like(nx_array, dtype='float')
max_rel_err = np.empty_like(max_abs_err)
for i, ni in enumerate(nx_array):
E1 = p1.propagate(omega,
x_start,
x_end,
ni,
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
E2 = p2.propagate(omega,
x_start,
x_end,
ni,
E_start,
Y1D,
Z1D,
include_main_phase=has_main_phase)
abs_err = np.abs(E2 - E1)
max_abs_err[i] = np.max(abs_err)
rel_idx = np.abs(E1) > 1e-3 * np.max(np.abs(E1))
max_rel_err[i] = np.max(np.abs(abs_err[rel_idx] / E1[rel_idx]))
print('Total time: {:.4}s'.format(time.clock() - tstart))
return max_abs_err, max_rel_err, nx_array, x_stepsize