def psearch(time, flux): freq,power,nout,jmax,prob = fasper(time, flux, 6, 0.5) period = 1./freq m = (period > 0.05) & (period < 35) period, freq, power = period[m], freq[m], power[m] j = np.argmax(power) expy = np.exp(-power) effm = 2.*nout/6 fap = expy*effm mfap = fap > 0.01 fap[mfap] = 1.0-(1.0-expy[mfap])**effm return period, power, fap, j
def psearch(time, flux, min_p, max_p):
'''
Search for a statistically significant period using a Lomb-Scargle periodogram.
'''
freq, power, nout, jmax, prob = fasper(time, flux, 6, 0.5)
period = 1 / freq
m = (period > min_p) & (period < max_p)
period, power = period[m], power[m]
j = argmax(power)
expy = mt.exp(-power[j])
effm = 2 * nout / 6
fap = expy * effm
if fap > 0.01:
fap = 1.0 - (1.0 - expy)**effm
return period[j], fap
ax[2].plot(time, (cflux-1)/sigma)
ax[2].grid(axis='y')
ax[3].plot(time[mask], N(flux[mask] - tp[mask], flux))
ax[3].plot(time[~mask], N(flux[~mask] - tp[~mask], flux), '.')
setp(ax[2], ylabel='Normalised flux [$\sigma$]')
setp(ax[3], ylabel='Normalised flux')
nflux = flux - tp + nanmedian(tp)
for split in dt.gp.splits:
i = argmin(abs(time - split))
nflux[i:] += nanmedian(nflux[i-10:i]) - nanmedian(nflux[i:i+10])
ax[4].plot(time[mask], N(nflux[mask]))
freq,power,nout,jmax,prob = fasper(time[mask], nflux[mask], 6, 0.5)
period = 1./freq
m = (period > 0.25) & (period < 25)
period, freq, power = period[m], freq[m], power[m]
j = argmax(power)
expy = exp(-power)
effm = 2.*nout/6
fap = expy*effm
mfap = fap > 0.01
fap[mfap] = 1.0-(1.0-expy[mfap])**effm
ax[5].semilogy(period, fap)
setp(ax[5], ylabel='FAP', xlabel='Period [d]', xlim=(0,25))
print period[j], power[j], fap[j]