Using Exponential Moving Averages to detect outburst in ZTF

Below are some objects from the Outburst catalog, with a timespan of a possible outburst and the magnitude limits for plotting. They are crossmatched from: "Outburst catalogue of cataclysmic variables" (Coppejans+, 2016) in Vizier as J/MNRAS/456/4441/catalog

In [1]:
outbursts = [
    {'objectId':'ZTF18aazvajt', 'Tmin':58620, 'Tmax':58650, 'magmin':19, 'magmax':15},
    {'objectId':'ZTF18aabenub', 'Tmin':58700, 'Tmax':58720, 'magmin':18, 'magmax':12},
    {'objectId':'ZTF18abaxueg', 'Tmin':58580, 'Tmax':58680, 'magmin':19, 'magmax':15},
    {'objectId':'ZTF18aanhfnc', 'Tmin':58680, 'Tmax':58700, 'magmin':20, 'magmax':14},   
In [2]:
import numpy as np
import math
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline
import json
import wget

Fetch data from Lasair

This function fetches the object data from Lasair as a JSON file, and makes lists of the apparent magnitude, its three EMAs at 2 days, 8 days, and 28 days, for the two filters, g and r.

In [3]:
def from_database(objectId):
    file = '%s.json' % objectId
        o = json.loads(open(file).read())
        url = '' % objectId, out=file)
        o = json.loads(open(file).read())
    print (o['objectId'])
    clist = o['candidates']

    tlist = {'g':[], 'r':[]}
    mags  = {'g':[], 'r':[]}
    f02   = {'g':[], 'r':[]}
    f08   = {'g':[], 'r':[]}
    f28   = {'g':[], 'r':[]}
    for c in clist:
        if 'candid' in c:
            if c['fid'] == 1: f = 'g'
            else:             f = 'r'
    return {'objectId':objectId, 'tlist':tlist, 'mags':mags, 'f02':f02, 'f08':f08, 'f28':f28}

Plot the outburst

Here we plot the apparent magnitude and the three EMAs for one of the two filters, with big red markers for candidates that re in outburst, according to the criterion E02 > E08 > E28 + 0.3. The number 0.3 hers is about 3 times the error estimate in magnitudes from ZTF.

In [4]:
def plot_outburst(d, f, Tmin=0, Tmax=0, min=0, max=0):

    tlist = np.array(d['tlist'][f])
    mags  = np.array(d['mags'][f])
    f02   = np.array(d['f02'][f])
    f08   = np.array(d['f08'][f])
    f28   = np.array(d['f28'][f])
    plt.plot(tlist, mags, marker='o', color=(0.0, 0.0, 0.0))
    plt.plot(tlist, f02, color=(0.5, 0.5, 0.5))
    plt.plot(tlist, f08, color=(0.7, 0.7, 0.7))
    plt.plot(tlist, f28, color=(0.8, 0.8, 0.8))

    for i in range(len(d['tlist'][f])):
        p = -mags[i]
        q = -f02[i]
        r = -f08[i]
        s = -f28[i]
        if q > r and r > s + 0.3:
            t = tlist[i]
            m = mags[i]
            plt.plot([t], [m], marker='o', markersize=10, color="red")
    plt.ylabel('%s magnitude' % f)
    plt.title('%s, filter %s' % (d['objectId'], f))
    if min > 0:
        plt.axis(ymin=max, ymax=min)
    if Tmin > 0:
        plt.axis(xmin=Tmin, xmax=Tmax)

Now plot the four suspected outbursts, with g filter left and r filter right.

In [5]:
for ob in outbursts:
    d = from_database(ob['objectId'])
    plt.subplot(1, 2, 1)
    plot_outburst(d, 'g', ob['Tmin'], ob['Tmax'], ob['magmin'], ob['magmax'])
    plt.subplot(1, 2, 2)
    plot_outburst(d, 'r', ob['Tmin'], ob['Tmax'], ob['magmin'], ob['magmax'])