import toy
from matplotlib import pyplot as plt
fig = plt.figure('savenoise2')
fig.clf()
axs = fig.subplots(2, 1)
axs[0].set_title('LNGS 1 GSa/s')
axs[1].set_title('LNGS 125 MSa/s')
sources = list(sorted(glob.glob('darksidehd/LF_TILE15_77K_*VoV_1.wav')))
for source in sources:
suffix = '.wav'
dest = 'noises/' + os.path.split(source)[1][:-len(suffix)] + '-noise.npz'
print(f'saving {source} to {dest}...')
lngs1GSas = toy.DataCycleNoise(timebase=1)
lngs1GSas.load_LNGS_wav(source, 1100)
lngs1GSas.save(dest)
lngs1GSas = toy.DataCycleNoise(timebase=1)
lngs1GSas.load(dest)
axs[0].plot(lngs1GSas.generate(1, 1000)[0])
lngs125MSas = toy.DataCycleNoise(timebase=8)
lngs125MSas.load(dest)
axs[1].plot(lngs125MSas.generate(1, 1000)[0])
fig.tight_layout()
fig.show()
of the script set the various parameters.
"""
import numpy as np
import toy
import template as _template
tau = np.array([4, 8, 16, 24, 32, 40, 48, 64, 96, 128, 192, 256, 384])
snr = np.linspace(1.8, 6, 15)
wlen = np.array(64 * np.array([1.5, 2, 2.5, 3, 4, 5, 6, 7]), int) - 32
wlmargin = np.full_like(wlen, 64 - 32)
wlmargin[0] = 16
noise_proto0 = toy.DataCycleNoise()
noise_proto0.load('noises/merged_000886-adc_W201_Ch00.npz')
noise_LNGS = toy.DataCycleNoise(allow_break=True)
noise_LNGS.load('noises/nuvhd_lf_3x_tile57_77K_64V_6VoV_1-noise.npz')
noise_white = toy.WhiteNoise()
noise_obj = [noise_proto0, noise_LNGS, noise_white]
noise_name = ['proto0', 'LNGS', 'white']
template = _template.Template.load(
'templates/nuvhd_lf_3x_tile57_77K_64V_6VoV_1-template.npz')
toys = []
for name, noise in zip(noise_name, noise_obj):
t = toy.Toy(template, tau, snr, noise, upsampling=False)
# upsampling=True reduces speed dramatically
print(f'running with {name} noise...')
for i in range(len(timebase)):
filename = f'toy1gvs125m/toy1gvs125m-{prefix}-{timebase[i]}-{noise_labels[inoise]}.npz'
print(f'load {filename}...')
t = toy.Toy.load(filename)
toys[inoise].append(t)
noise = [None] * 2
noise_ratio = [None] * 2
noise[0] = [toy.WhiteNoise(timebase=t) for t in timebase]
noise_ratio[0] = [np.sqrt(timebase[0] / tb) for tb in timebase]
noise_file = f'noises/{prefix}-noise.npz'
noise[1] = []
for i in range(len(timebase)):
n = toy.DataCycleNoise(allow_break=True, timebase=timebase[i])
n.load(noise_file)
noise[1].append(n)
noise_ratio[1] = [
np.std(toy.downsample(noise[1][0].noise_array, tb // timebase[0]))
for i, tb in enumerate(timebase)
]
for inoise in range(len(noise_labels)):
print(
f'noise_ratio ({noise_labels[inoise]}) =',
', '.join(f'{noise_ratio[inoise][i]:.3f} ({tb})'
for i, tb in enumerate(timebase)))
tau = toys[0][0].tau * toys[0][0].timebase
if not all(
os.path.exists(simfile(n, tb)) for n in timebase
for tb in timebase[n]):
template = _template.Template.load(f'templates/{prefix}-template.npz')
for n, timebases in timebase.items():
timebases = np.sort(timebases)
noises = {}
noise_ratio = {}
for itb, tb in enumerate(timebases):
if n == 'proto0':
noise = toy.DataCycleNoise(maxcycles=2,
chunk_skip=1000,
timebase=tb)
noise.load(noisefile)
elif n == 'lngs':
noise = toy.DataCycleNoise(maxcycles=2, timebase=tb)
noise.load(f'noises/{prefix}-noise.npz')
elif n == 'white':
noise = toy.WhiteNoise(timebase=tb)
noises[tb] = noise
if itb == 0:
basenoise = noise
basetb = tb
noise_ratio[tb] = np.std(toy.downsample(
basenoise.generate(20, 5000), tb // basetb),
axis=None)
import numpy as np from matplotlib import pyplot as plt import figlatex import toy import num2si import template as _template templfile = 'templates/nuvhd_lf_3x_tile57_77K_64V_6VoV_1-template.npz' noisefile = 'noises/merged_000886-adc_W201_Ch00.npz' tau = 256 snr = 5 ########################### noise = toy.DataCycleNoise(maxcycles=1, chunk_skip=1000) noise.load(noisefile) template = _template.Template.load(templfile) sim = toy.Toy(template, [tau], [snr], noise) sim.run(1000, pbar=10, seed=202102172153) fig, axs = plt.subplots(2, 2, num='figlochist', clear=True, figsize=[7, 4.3]) sim.plot_loc(0, 0, axs=axs, center=True) fig.tight_layout() fig.show() figlatex.save(fig)
import figlatex
import toy
import num2si
import template as _template
prefix = 'nuvhd_lf_3x_tile57_77K_64V_6VoV_1'
snr = 5
mftau = 256
locfilter = 2
loctau = 16
wlen = 64
wlmargin = 16
###########################
noise = toy.DataCycleNoise(allow_break=True)
noise.load(f'noises/{prefix}-noise.npz')
template = _template.Template.load(f'templates/{prefix}-template.npz')
sim = toy.Toy(template, [loctau, mftau], [snr], noise)
sim.run(100, seed=202102190959)
wcenter = sim.window_center([locfilter], [0], [0])
sim.run_window([wlen], [wlmargin], wcenter)
fig, axs = plt.subplots(1,
2,
num='figwindowevent',
clear=True,
sharex=True,
sharey=True,
figsize=[8.94, 4.41])
"""
Run a simulation at 1 GSa/s with white and LNGS noise, saving the result to
file.
"""
import numpy as np
import toy
import template as _template
tau = np.array([4, 8, 16, 24, 32, 40, 48, 64, 96, 128, 192, 256, 384]) * 8
snr = np.linspace(1.8, 6, 15)
noise_LNGS = toy.DataCycleNoise(allow_break=True, timebase=1)
noise_LNGS.load('noises/nuvhd_lf_3x_tile57_77K_64V_6VoV_1-noise.npz')
noise_white = toy.WhiteNoise(timebase=1)
noise_obj = [noise_LNGS, noise_white]
noise_name = ['LNGS', 'white']
template = _template.Template.load(
'templates/nuvhd_lf_3x_tile57_77K_64V_6VoV_1-template.npz')
toys = []
for name, noise in zip(noise_name, noise_obj):
t = toy.Toy(template, tau, snr, noise, timebase=1)
print(f'running with {name} noise...')
t.run(1000, pbar=10, seed=202012081136)
toys.append(t)
for inoise in range(len(toys)):
filename = f'toy1gsa-{noise_name[inoise]}.npz'
def isomething(a, x, strict=True):
i = min(np.searchsorted(a, x), len(a) - 1)
assert not strict or a[i] == x
return i
itau = lambda t: isomething(tau, t)
isnr = lambda s: isomething(snr, s, False)
names = ['proto0', 'lngs']
nicenames = dict(proto0='Proto0', lngs='LNGS')
simfile = lambda n: f'figthesis/figwindowtempres-{n}.npz'
if not all(os.path.exists(simfile(n)) for n in names):
noise_proto0 = toy.DataCycleNoise(maxcycles=2, chunk_skip=1000)
noise_proto0.load(noisefile)
noise_lngs = toy.DataCycleNoise(maxcycles=2)
noise_lngs.load(f'noises/{prefix}-noise.npz')
noise = dict(proto0=noise_proto0, lngs=noise_lngs)
template = _template.Template.load(f'templates/{prefix}-template.npz')
for n in names:
if os.path.exists(simfile(n)):
continue
sim = toy.Toy(template, tau, snr, noise[n])
sim.run(1000, pbar=10, seed=202102191139)
res = sim.templocres() # shape == (nfilter, ntau, nsnr)
i = isnr(2.4)
template = _template.Template.load(templfile)
savedir = 'toy1gvs125m'
os.makedirs(savedir, exist_ok=True)
for whitenoise in [True, False]:
if whitenoise:
noise = [toy.WhiteNoise(timebase=t) for t in timebase]
noise_name = 'white'
else:
noise_file = f'noises/{prefix}-noise.npz'
print(f'read {noise_file}...')
noise = []
for i in range(len(timebase)):
n = toy.DataCycleNoise(maxcycles=2, timebase=timebase[i])
n.load(noise_file)
noise.append(n)
noise_name = 'lngs'
for i in range(len(timebase)):
tb = timebase[i]
noise_ratio = np.std(toy.downsample(noise[0].generate(100, 100), tb),
axis=None)
snr = snr_1GSa / noise_ratio
t = toy.Toy(template, tau // tb, snr, noise[i], timebase=tb)
print(
f'running with timebase={tb} (SNR *= {1/noise_ratio:.2g}), {noise_name} noise...'
)
t.run(1000, pbar=10, seed=202012111417)
filename = f'{savedir}/toy1gvs125m-{prefix}-{timebase[i]}-{noise_name}.npz'
}
###########################
fig, axs = plt.subplots(1,
2,
num='fignoise',
clear=True,
figsize=[9, 3.3],
sharex=True,
sharey=True)
for ax, (label, noisefile, N, timebase) in zip(axs, config):
basetb = np.gcd.reduce(timebase)
noise = toy.DataCycleNoise(timebase=basetb)
noise.load(noisefile)
ax.set_xlabel(f'Time [ns]')
n = noise.generate(1, 3 * N)[0]
for tb in timebase:
nr = toy.downsample(n, tb // basetb)
x = (tb - 1) / 2 + tb * np.arange(len(nr)) - N
kwargs = dict(
label=num2si.num2si(1e9 / tb, format='%.3g') + 'Sa/s',
marker='.',
)
kwargs.update(plotkw[tb])
ax.plot(x, nr, **kwargs)
