def test_serial_operator(verbose=True, plot=False, *args, **kwargs):
import matplotlib.pyplot as plt
s = load_spec(getTestFile('CO_Tgas1500K_mole_fraction0.01.spec'))
s.update()
s1 = s.rescale_path_length(10, inplace=False) # make non optically thin
s2 = s.rescale_path_length(20, inplace=False) # make non optically thin
s3 = s.rescale_path_length(30, inplace=False) # make non optically thin
if plot:
# s.plot('radiance_noslit', lw=2, nfig='Line of sight (SerialSlabs): s > s > s')
plt.figure('test_serial_operator').clear()
SerialSlabs(s1, s2, s3).apply_slit(1).plot(nfig='same')
((s1 > s2) > s3).apply_slit(1).plot(nfig='same')
assert SerialSlabs(s1, s2, s3).compare_with((s1 > s2) > s3,
spectra_only=True,
plot=False)
assert (s1 > (s2 > s3)) == ((s1 > s2) > s3)
# Forbidden syntax:
with pytest.raises(ArithmeticError):
assert SerialSlabs(s1, s2, s3).compare_with(s1 > s2 > s3,
spectra_only=True,
plot=False)
def test_compare_torch_CO2(verbose=True,
plot=False,
save=False,
warnings=True,
use_cache=True,
*args,
**kwargs):
"""Reproduce the plasma torch experiment of Packan 2003 in in atmospheric air
Notes
-----
Thresholds parameters are reduced for a faster calculation time. For instance:
- broadening_max_width should be 50 rather than 20
- cutoff should be 1e-27 rather than 1e-25
- wstep should be 0.01 or even 0.008 rather than 0.1
Performance:
- neq==0.9.20: Finished test_compare_torch_CO2 in 758.640324s
- neq==0.9.21: Finished test_compare_torch_CO2 in 672s
- neq==0.9.21*: (with ParallelFactory) Finished test_compare_torch_CO2 in 298s
- neq==0.9.22: (Parallel + continuum) Finished in 65s
RADIS 0.9.9 == neq 0.9.24
- RADIS 0.9.26: Finished test_compare_torch_CO2 in 57s (DLM, no continuum)
Reference
--------
Packan et al 2003 "Measurement and Modeling of OH, NO, and CO Infrared Radiation
at 3400 K", JTHT
"""
if plot: # Make sure matplotlib is interactive so that test are not stuck in pytest
plt.ion()
t0 = time()
# %% Conditions
wlmin = 4100
wlmax = 4900
wmin = nm2cm(wlmax)
wmax = nm2cm(wlmin)
wstep = 0.1
# Load concentration profile
conds = pd.read_csv(
getValidationCase(
"test_compare_torch_CO2_data/test_compare_torch_CO2_conditions_JTHT2003.dat"
),
comment="#",
delim_whitespace=True,
)
slab_width = 0.05 # cm, WARNING. Hardcoded (look up the table)
# %% Calculate slabs
# CO2
sf = ParallelFactory(
wavenum_min=wmin,
wavenum_max=wmax,
mole_fraction=None,
path_length=None,
molecule="CO2",
isotope="1,2",
parallel=False,
wstep=wstep,
db_use_cached=use_cache,
export_lines=False, # saves some memory
export_populations=False, # saves some memory
cutoff=1e-25,
Nprocs=cpu_count() - 1, # warning with memory if too many procs
broadening_max_width=20,
# pseudo_continuum_threshold=0.01, # use pseudo-continuum, no DLM. Note : 56s on 20/08.
# optimization=None,
pseudo_continuum_threshold=
0, # use DLM, no pseudo-continuum. Note : 84s on 20/08
optimization="min-RMS",
verbose=False,
)
sf.warnings["MissingSelfBroadeningWarning"] = "ignore"
sf.warnings["NegativeEnergiesWarning"] = "ignore"
# # Init database: only if we want to save all spectra being calculated
# (discarded for the moment)
# if use_cache:
# sf.init_database('test_compare_torch_CO2_SpectrumDatabase_HITEMP_1e25',
# add_info=['Tgas'])
sf.init_databank("HITEMP-CO2-DUNHAM",
load_energies=False) # saves some memory
# sf.init_databank('CDSD-4000')
# CO
sfco = ParallelFactory(
wavenum_min=wmin,
wavenum_max=wmax,
mole_fraction=None,
path_length=None,
molecule="CO",
isotope="1,2",
parallel=False,
wstep=wstep,
db_use_cached=use_cache,
export_lines=False, # saves some memory
export_populations=False, # saves some memory
cutoff=1e-25,
Nprocs=cpu_count() - 1,
broadening_max_width=20,
optimization=None,
verbose=False,
)
sfco.warnings["MissingSelfBroadeningWarning"] = "ignore"
# Init database: only if we want to save all spectra being calculated
# (discarded for the moment)
# if use_cache:
# sfco.init_database(
# 'test_compare_torch_CO_SpectrumDatabase_HITEMP_1e25', add_info=['Tgas'])
sfco.init_databank("HITEMP-CO-DUNHAM")
# Calculate all slabs
# .. CO2 at equilibrium
slabsco2 = sf.eq_spectrum(list(conds.T_K),
mole_fraction=list(conds.CO2),
path_length=slab_width)
# .. CO at equilibrium, but with non_eq to calculate PartitionFunctions instead of using
# ... tabulated ones (which are limited to 3000 K in HAPI)
slabsco = sfco.non_eq_spectrum(
list(conds.T_K),
list(conds.T_K),
mole_fraction=list(conds.CO),
path_length=slab_width,
)
# Room absorption
with pytest.warns(
UserWarning
): # we expect a "using ParallelFactory for single case" warning
s0 = sf.eq_spectrum(Tgas=300, mole_fraction=330e-6, path_length=600)[0]
# Warning: This slab includes the CO2 emission from the 300 K air
# within the spectrometer. But experimentally it is substracted
# by there the chopper (this is corrected below)
# ... see RADIS Article for more details
# Add radiance for everyone if not calculated (ex: loaded from database)
for s in slabsco2 + slabsco + [s0]:
s.update()
# Merge CO + CO2 for each slab
slabstot = [MergeSlabs(s, sco) for (s, sco) in zip(slabsco2, slabsco)]
# %% Line-of-sight
# Solve ETR along line of sight
# --------
# two semi profile, + room absortion
line_of_sight = slabstot[1:][::-1] + slabstot + [s0]
stot = SerialSlabs(*line_of_sight)
# stot = SerialSlabs(*slabstot[1:][::-1], *slabstot, s0) # Python 3 syntax only
# Also calculate the contribution of pure CO
# ---------
s0tr = s0.copy() # transmittance only
s0tr.conditions["thermal_equilibrium"] = False
s0tr._q["radiance_noslit"] *= 0 # hack to remove CO2 emission
s0tr._q["emisscoeff"] *= 0 # hack to remove CO2 emission
line_of_sight = slabsco[1:][::-1] + slabsco + [s0tr]
sco = SerialSlabs(*line_of_sight)
# sco = SerialSlabs(*slabsco[1:][::-1], *slabsco, s0tr) # Python 3 syntax only
# Generate Slit
# ------
disp = 4 # dispersion conversion function (nm/mm)
s_in = 1 # entrance slit (mm)
s_out = 2.8 # exit slit (mm)
top = (s_out - s_in) * disp
base = (s_out + s_in) * disp
slit_function = (top, base) # (nm)
# Final plot unit
unit = "µW/cm2/sr"
norm_by = "max" # should be 'max' if unit ~ 'µW/cm2/sr',
# 'area' if unit ~ 'µW/cm2/sr/nm'
# Convolve with Slit
# -------
sco.apply_slit(slit_function,
unit="nm",
norm_by=norm_by,
shape="trapezoidal")
stot.apply_slit(slit_function,
unit="nm",
norm_by=norm_by,
shape="trapezoidal")
# Remove emission from within the spectrometer (substracted
# by the chopper experimentaly)
# -------
s0spectro = s0.copy()
s0spectro.rescale_path_length(75 * 4) # spectrometer length
s0spectro.apply_slit(slit_function,
unit="nm",
norm_by=norm_by,
shape="trapezoidal")
_, I0spectro = s0spectro.get("radiance", Iunit=unit)
wtot, Itot = stot.get("radiance", Iunit=unit)
stot_corr = experimental_spectrum(
wtot,
Itot - I0spectro, # hack to remove
# emission from within
# spectrometer
Iunit=unit,
conditions={"medium": "air"},
)
# %% Compare with experiment data
# Plot experimental data
# ------
exp = pd.read_csv(
getValidationCase(
"test_compare_torch_CO2_data/test_compare_torch_CO2_spectrum_JTHT2003.dat"
),
skiprows=5,
delim_whitespace=True,
)
exp.w /= 10 # Angstrom to nm
exp = exp[(exp.w > wlmin) & (exp.w < wlmax)]
sexp = experimental_spectrum(exp.w,
exp.I,
Iunit="mW/cm2/sr",
conditions={"medium": "air"})
if plot:
sexp.plot("radiance",
wunit="nm",
Iunit=unit,
lw=2,
label="Packan 2003")
# Plot calculated
# ----------
stot.plot(
"radiance",
wunit="nm",
Iunit=unit,
nfig="same",
color="r",
ls="--",
label="All: CO$_2$+CO",
)
# Plot internal spectrometer emission
s0spectro.plot(
"radiance",
wunit="nm",
Iunit=unit,
nfig="same",
color="b",
label="CO$_2$ in Spectrometer",
)
stot_corr.plot(
"radiance",
wunit="nm",
Iunit=unit,
nfig="same",
color="r",
lw=2,
label="All-CO$_2$ in Spectro",
zorder=10,
)
sco.plot(
"radiance",
wunit="nm",
Iunit=unit,
nfig="same",
color="g",
ls="-",
label="CO",
)
plt.ylim((0, 9))
plt.legend(loc="upper right")
if save:
plt.savefig("test_compare_torch_CO2_brd20.png")
plt.savefig("test_compare_torch_CO2_brd20k.pdf")
assert get_residual(stot_corr, sexp, "radiance") < 0.02
if verbose:
print(("Finished test_compare_torch_CO2 in {0:.0f}s".format(time() -
t0)))