pgls: Penalized Generalized Lomb-Scargle

In multiband: Period Estimation for Multiple Bands

Description

pgls estimates periods for a collection of lightcurves sampled over multiple bands. It borrows strength across multiple bands via shrinkage penalties on amplitudes an phases.

Usage

pgls(lclist, period_min = NULL, period_max = NULL, periods = NULL,
  gamma1 = 0, gamma2 = 20, at = rep(1, length(lclist)), LS_flag = TRUE,
  sol_ls = NULL, BCD_flag = TRUE, fast_BCD_flag = TRUE, max_iter = 100,
  tol = 1e-04, mm_iter = 5, verbose = FALSE)

Arguments

lclist	list of lightcurve data frames
period_min	minimum period
period_max	maximum period
periods	grid of periods
gamma1	vector of Amplitude regularization parameter
gamma2	vector of Phase regularization parameter
at	amplitude prior parameter
LS_flag	boolean whether to run Lomb-Scargle algorithm
sol_ls	Lomb-Scargle solution, used if LS_flag=FALSE
BCD_flag	boolean whether to run bcd algorithm
fast_BCD_flag	boolean whether to run BCD on relevant subset of periods
max_iter	maximum number of outer iterations - passed to bcd_inexact
tol	tolerance on relative change in loss - passed to bcd_inexact
mm_iter	number of MM iterations for rho update - passed to bcd_inexact
verbose	boolean whether to print progress

Examples

period_min <- 3.5
period_max <- 3.65
out <- pgls(cepii,period_min=period_min,period_max=period_max)
out$best_fitBCD

Example output

      param
band      beta0        amp      rho
  [1,] 17.74903 0.02970725 2.338195
  [2,] 18.27665 0.24931230 2.744296

multiband documentation built on May 2, 2019, 3:30 a.m.