09_generate_training_data.py

""" 
Python3 - UTF-8

Ekaterina Ilin 
MIT License (2022)

Script to generate training data. Takes the time stamp of the planned
run as a first argument. The second argument is the number of light curves
to generate. The third argument is the path to the output file. The fourth
argument is either "train" or "validate" to indicate whether the data is
for training or validation.

"""

import numpy as np
import pandas as pd

from altaipony.flarelc import FlareLightCurve

from flares.flares import get_flares

from flares.__init__ import LOG_DATA_OVERVIEW_PATH                     

import sys

if __name__ == "__main__":

    # ---------------- COMMAND LINE INPUT PARAMETERS ---------------------------

    # timestamp for this dataset
    tstamp = sys.argv[1]
    
    # number of lcs in batch
    n_lcs = int(sys.argv[2])

    # table of flares to store the results in
    outpath = sys.argv[3]

    # either train or validate
    typ = sys.argv[4]

    # ---------------- COMMAND LINE INPUT PARAMETERS END -----------------------


    # -------------------- LOG FILE INPUT PARAMETERS ---------------------------

    # read log file with input parameters
    df = pd.read_csv(LOG_DATA_OVERVIEW_PATH)

    # pick the right row and make sure all identifiers fit!
    # only n_lcs does not fit because we are dealing with batches here
    row = df[(df.typ == typ) &
             (df.tstamp == tstamp) &
             (df.outpath == outpath)].iloc[0]

    # ------------------ LOG FILE INPUT PARAMETERS END -------------------------


    # --------------------- DERIVED INPUT PARAMETERS ---------------------------

    # time series in rad
    t = np.arange(0, 2 * np.pi, 2 * np.pi / row.size_lc)

    # define flare light curve
    flc = FlareLightCurve(time=t)

    # error series kept constant to save computational time
    # is correct per definition, even somewhat more correct
    # than doable in practice
    err = np.full_like(t, row.errval)
    flc.detrended_flux_err = err

    # the quiescent median is one 
    itmed = np.ones_like(t)
    flc.it_med = itmed

    # generate only one lc per iteration to make code less complicated
    n_inclinations = 1

    # ------------------- DERIVED INPUT PARAMETERS END -------------------------


    # ---------------------- RUN LOOP WITH INPUTS ------------------------------

    inputs = ((row.u_ld_0, row.u_ld_1), flc, row.emin, row.emax, row.errval,
              row.spot_radius, n_inclinations,row.alphamin, row.alphamax,
              row.betamin, row.betamax, row.n_spots_min, row.n_spots_max, 
              row.midlat, row.latwidth, row.decomposeed, outpath)

    for i in range(n_lcs):
        get_flares(*inputs)

    # --------------------- RUN LOOP WITH INPUTS END ---------------------------