periodicity 0.1.0b2

Useful tools for analysis of periodicities in time series data

Periodicity

Useful tools for analysis of periodicities in time series data.

Includes:

• Auto-Correlation Function
• Fourier methods:
  • Lomb-Scargle periodogram
  • Wavelet Transform (in progress)
• Phase-folding methods:
  • String Length
  • Analysis of Variance (in progress)
• Gaussian Processes:
  • george implementation
  • celerite implementation
  • pymc3 implementation (in progress)

Quick start

Installing current release from pypi (v0.1.0-alpha)

$ pip install periodicity

Installing current development version

$ git clone https://github.com/dioph/periodicity.git
$ cd periodicity
$ python setup.py install

Example using GP with astronomical data

from periodicity.gp import *
from lightkurve import search_lightcurvefile

lcs = search_lightcurvefile(target=9895037, quarter=[4,5]).download_all()
lc = lcs[0].PDCSAP_FLUX.normalize().append(lcs[1].PDCSAP_FLUX.normalize())
lc = lc.remove_nans().remove_outliers().bin(binsize=4)

t, x = lc.time, lc.flux
x = x - x.mean()

model = FastGPModeler(t, x)
model.prior = make_gaussian_prior(t, x, pmin=2)
model.minimize()
samples = model.mcmc(nwalkers=32, nsteps=5000, burn=500)

print('Median period: {:.2f}'.format(np.exp(np.median(samples[:, 4]))))

Visualization of this example: