def out_backplot(self, fit):
res = np.median(fit['phi'], axis=0)
res[0:6] = 10**res[0:6]
model = self._get_background(f, *res)
pg = lk.Periodogram(f * u.microhertz, p * (cds.ppm**2 / u.microhertz))
ax = pg.plot(alpha=.25, label='Data', scale='log')
ax.plot(f, model, label='Model')
ax.plot(f,
self._harvey(f, res[0], res[1], 4.),
label='Harvey 1',
ls=':')
ax.plot(f,
self._harvey(f, res[2], res[3], 4.),
label='Harvey 2',
ls=':')
ax.plot(f,
self._harvey(f, res[4], res[5], 2.),
label='Harvey 3',
ls=':')
ax.plot(f, self._get_apodization(f, f[-1]), label='Apod', ls='--')
ax.plot(f, res[-4] * np.ones_like(f), label='white', ls='-.')
plt.legend(fontsize=10)
plt.savefig(self.dir + 'modelplot.png')
plt.close('all')
def out_modelplot(self, fit):
model = np.ones(len(self.data['f']))
nus = np.median(fit['nus'])
for mode in range(len(self.data['ids'])):
l = self.data['ids'][mode]
for m in range(-l, l + 1):
loc = np.median(fit['locs'].T[mode])
H = np.median(fit['H'].T[mode])
w = np.median(fit['w'].T[mode])
model += self._lorentzian(f, l, m, loc, i, H, w, nus)
fitlocs = np.median(fit['locs'], axis=0)
pg = lk.Periodogram(data['f'] * u.microhertz,
data['p'] * (cds.ppm**2 / u.microhertz))
ax = pg.plot(alpha=.5, label='Data')
plt.scatter(fitlocs, [15] * len(fitlocs),
c='k',
s=25,
label='fit locs')
plt.scatter(data['pr_locs'], [15] * len(data['pr_locs']),
c='r',
s=5,
label='true locs')
plt.plot(data['f'], model, linewidth=1, label='Model')
plt.legend()
plt.savefig(self.dir + 'modelplot.png')
plt.close('all')
def full_plot(mod, p, res, kic, idx):
pg = lk.Periodogram(mod.f * u.microhertz,
p * (u.cds.ppm**2 / u.microhertz))
ax = pg.plot(alpha=.5, label='Data')
plt.plot(mod.f, mod.model(res, theano=False), lw=3, label='Model')
ax.set_title(f'KIC {kic}, ID {idx}')
plt.savefig(
f'/home/oliver/PhD/mnt/RDS/malatium/peakbag/byeye/{kic}_full.png')
plt.close()
def out_modelplot(self):
labels=['f0','f1','f2',
'g0','g1','g2',
'h0','h1','h2',
'split','i',
'phi']
res_m = np.array([np.median(self.trace[label],axis=0) for label in labels])
pg = lk.Periodogram(self.mod.f*u.microhertz, self.p*(cds.ppm**2/u.microhertz))
ax = pg.plot(alpha=.5, label='Data')
plt.plot(self.mod.f, self.mod.model(res_m, theano=False), lw=3, label='Model')
plt.legend()
plt.savefig(self.dir+'modelplot.png')
plt.close('all')
def out_modelplot(self, fit):
labels=['loga','logb','logc','logd','logj','logk','white','numax','scale','nyq']
res = np.array([np.median(fit[label]) for label in labels])
res[0:6] = 10**res[0:6]
model = self.get_background(f, *res)
pg = lk.Periodogram(f*u.microhertz, p*(cds.ppm**2/u.microhertz))
ax = pg.plot(alpha=.25, label='Data', scale='log')
ax.plot(f, model, label='Model')
ax.plot(f, self.harvey(f, res[0],res[1], 4.), label='Harvey 1', ls=':')
ax.plot(f, self.harvey(f, res[2],res[3], 4.), label='Harvey 2', ls=':')
ax.plot(f, self.harvey(f, res[4],res[5], 2.), label='Harvey 3', ls=':')
ax.plot(f, self.get_apodization(f, f[-1]), label='Apod', ls='--')
ax.plot(f, res[-4]*np.ones_like(f), label='white',ls='-.')
plt.legend(fontsize=10)
plt.savefig(self.dir+'modelplot.png')
plt.close('all')
f2_, # l2 modes
np.ones(len(f0_)) * 2.0, # l0 widths
np.ones(len(f1_)) * 2.0, # l1 widths
np.ones(len(f2_)) * 2.0, # l2 widths
np.ones(len(f0_)) * 15. * 2.0 / np.pi / 2.0, # l0 amps
np.ones(len(f1_)) * 15. * 2.0 / np.pi / 2.0, # l1 amps
np.ones(len(f2_)) * 15. * 2.0 / np.pi / 2.0, # l2 amps
1.0 , # projected splitting
np.pi/2., # inclination angle
phi_ # background terms
]
mod = model(f, n0_, n1_, n2_, deltanu_)
#Plot the data
pg = lk.Periodogram(f*u.microhertz, p*(cds.ppm**2/u.microhertz))
ax = pg.plot(alpha=.5)
ax.scatter(locs, [15]*len(locs),c=modeids, s=20, edgecolor='k')
ax.plot(f, mod.model(init, theano=False), lw=2)
plt.savefig(dir+'dataplot.png')
# plt.show()
plt.close()
# Run stan
print('About to go into Pymc3')
run = run_pymc3(mod, p, nf_, kic, phi_, phi_cholesky, dir)
run()
#Print a big label
print('#################################\n')
print(f'DONE RUNNING KIC {str(kic)} | IDX {str(idx)} \n')
def job(r, save=True):
ID = int(r['ID'])
if CATALOGUE == "TIC":
file = f"{PATH}/{ID}_psd_tot_nw.txt"
elif CATALOGUE == "EPIC":
file = f"{PATH}/{ID}/{ID}_psd_tot.txt"
Δν = float(r['Δν'])
ν_max = float(r['ν_max'])
if Δν < 0:
return
# Read in power spectrum
try:
file = f"DIAMONDS/data/{CATALOGUE}{ID}.txt"
ν, ps = np.loadtxt(file).T
except (FileNotFoundError, OSError):
# tqdm.write(f"PS for {CATALOGUE} {ID} not found")
return
except ValueError:
print(f"PS for {CATALOGUE} {ID} malformed")
return
# Create periodogram object
pg = lk.Periodogram(ν * u.uHz, u.Quantity(ps), label=f'{CATALOGUE} {ID}')
# Make echelle diagram
plt.close('all')
ss = pg.to_seismology()
ss.numax = ν_max
ss.deltanu = Δν
ss.plot_echelle(smooth_filter_width=Δν / 100)
# get manual mode ID
if isfile(f'Seed/{CATALOGUE}{ID}.pkl'):
seed = pd.read_pickle(f'Seed/{CATALOGUE}{ID}.pkl')
modes = seed.nu_med.values
plt.scatter(modes % Δν, (modes // Δν) * Δν, c='green')
# get Appourchaux frequencies
if CATALOGUE == "KIC" and ID in app_KIC:
modes = app[app_KIC_all == ID].Freq.astype('float')
plt.plot(modes % Δν, (modes // Δν) * Δν,
'v',
c='red',
fillstyle='none')
# get MLE estimate
if isfile(f'mle/{CATALOGUE}{ID}.npy'):
params = np.load(f'mle/{CATALOGUE}{ID}.npy', allow_pickle=True)
cov = np.array([(lambda J: np.linalg.inv(J.T @ J))(j['jac'])
for j in params])
ν0 = np.array([j['x'][0] for j in params])
e_ν0 = np.array([np.sqrt(C[0, 0]) for C in cov])
plt.errorbar(ν0 % Δν, (ν0 // Δν) * Δν,
fmt='^',
color='orange',
fillstyle='none')
# Get DIAMONDS results
def get_results(n):
result = f"DIAMONDS/results/{CATALOGUE}{ID}/pb/{n}/peakbagging_parameterSummary.txt"
if not isfile(result):
return None
res = np.loadtxt(result)
return res.reshape((len(res) // 3, 3, -1))
for i in range(3):
res = get_results(i)
if res is None:
break
for mode in res:
ν = mode[0, 1]
eν1 = mode[0, 4]
eν2 = mode[0, 5]
err = [[ν - eν1], [eν2 - ν]]
plt.errorbar([ν % Δν], [(ν // Δν) * Δν],
xerr=err,
yerr=err,
fmt='.',
c='red')
print(ν)
if save:
plt.savefig(f"DIAMONDS/preview/{CATALOGUE}{ID}.png")
else:
plt.show()