def ringfunc(taun, *args):
'cost function for ring fit to photometric data'
(t, f, f_err, rad_ri, re, k, dst) = args
# convolve and make smoothed ring photometry
strip, dummy, g = exorings.ellipse_strip(rad_ri, \
exorings.y_to_tau(taun), re[0], re[1], re[2], re[3], k, dst)
# interpolate the smoothed curve....
ring_model = interp1d(g[0], g[1], kind='linear')
# ... to the times of the photometry
ring_model_phot = ring_model(t)
# calculate the residuals and the chi squared
diff = f - ring_model_phot
chisq = np.sum(np.power(diff / f_err, 2))
red_chisq = chisq / diff.size
return red_chisq
def ringfunc(taun, *args):
'cost function for ring fit to photometric data'
(t, f, f_err, rad_ri, re, k, dst) = args
# convolve and make smoothed ring photometry
strip, dummy, g = exorings.ellipse_strip(rad_ri, \
exorings.y_to_tau(taun), re[0], re[1], re[2], re[3], k, dst)
# interpolate the smoothed curve....
ring_model = interp1d(g[0], g[1], kind='linear')
# ... to the times of the photometry
ring_model_phot = ring_model(t)
# calculate the residuals and the chi squared
diff = f - ring_model_phot
chisq = np.sum(np.power(diff/f_err, 2))
red_chisq = chisq / diff.size
return red_chisq
def calc_ring_stats(taun, t, f, f_err, rad_ri, re, k, dst, tmin, tmax):
'full statistics function for ring fit to photometric data'
# convolve and make smoothed ring photometry
strip, dummy, g = exorings.ellipse_strip(rad_ri, \
exorings.y_to_tau(taun), re[0], re[1], re[2], re[3], k, dst)
# select points within the rings_tmin/tmax range
mask = (t > tmin) * (t < tmax)
t_sel = t[mask]
f_sel = f[mask]
f_err_sel = f_err[mask]
print '%d points in time range %.2f to %.2f' % (t_sel.size, tmin, tmax)
# interpolate the smoothed curve....
ring_model = interp1d(g[0], g[1], kind='linear')
# ... to the times of the photometry
ring_model_phot = ring_model(t_sel)
# calculate the residuals and the chi squared
diff = f_sel - ring_model_phot
chisq = np.sum(np.power(diff/f_err_sel, 2))
# degrees of freedom = number of photometry points - number of ring edges - 1
dof = diff.size - taun.size - 1
red_chisq = chisq / dof
print 'number of photometric = %d ' % diff.size
print 'number of ring edges = %d ' % taun.size
print 'number of DOF = %d ' % dof
print 'chi squared = %.2f' % chisq
print ' reduced chi squared = %.2f' % red_chisq
# http://en.wikipedia.org/wiki/Bayesian_information_criterion
# n - number of points in data
# k - number of free parameters
# BIC = chisquared + k . ln(n) + C
# C is a constant which does not change between candidate models but is
# dependent on the data points
BIC = chisq + (taun.size) * np.log(diff.size)
print ' BIC = %.2f' % BIC
return red_chisq
def calc_ring_stats(taun, t, f, f_err, rad_ri, re, k, dst, tmin, tmax):
'full statistics function for ring fit to photometric data'
# convolve and make smoothed ring photometry
strip, dummy, g = exorings.ellipse_strip(rad_ri, \
exorings.y_to_tau(taun), re[0], re[1], re[2], re[3], k, dst)
# select points within the rings_tmin/tmax range
mask = (t > tmin) * (t < tmax)
t_sel = t[mask]
f_sel = f[mask]
f_err_sel = f_err[mask]
print '%d points in time range %.2f to %.2f' % (t_sel.size, tmin, tmax)
# interpolate the smoothed curve....
ring_model = interp1d(g[0], g[1], kind='linear')
# ... to the times of the photometry
ring_model_phot = ring_model(t_sel)
# calculate the residuals and the chi squared
diff = f_sel - ring_model_phot
chisq = np.sum(np.power(diff / f_err_sel, 2))
# degrees of freedom = number of photometry points - number of ring edges - 1
dof = diff.size - taun.size - 1
red_chisq = chisq / dof
print 'number of photometric = %d ' % diff.size
print 'number of ring edges = %d ' % taun.size
print 'number of DOF = %d ' % dof
print 'chi squared = %.2f' % chisq
print ' reduced chi squared = %.2f' % red_chisq
# http://en.wikipedia.org/wiki/Bayesian_information_criterion
# n - number of points in data
# k - number of free parameters
# BIC = chisquared + k . ln(n) + C
# C is a constant which does not change between candidate models but is
# dependent on the data points
BIC = chisq + (taun.size) * np.log(diff.size)
print ' BIC = %.2f' % BIC
return red_chisq
def onclick(event):
global rad_rings # naughty, I know, I know...
global taun_rings
global badring
for case in switch(event.key):
newt = event.xdata
newtau = event.ydata
print 'newtau is %f' % newtau
if newtau > 1.0:
newtau = 1.0
if newtau < 0.0:
newtau = 0.000001
newr = hjd_to_ring(newt + hjd_minr)
if case('r'):
exorings.plot_tilt_faceon(rad_rings, taun_rings, res, hjd_minr, dstar)
break
if case('v'):
exorings.plot_tilt_faceon_vect(rad_rings, taun_rings, res, \
hjd_minr, rstart, rend, dstar)
break
if case('a'):
print 'a pressed, adding a new ring'
# assume that r is ordered
print ' inserting new ring date %d and radius %d' % (newt, newr)
bigr = np.append(rad_rings, newr)
bigtau = np.append(taun_rings, exorings.tau_to_y(newtau))
# index sort r, rearrange both r and tau to this
sortedr = np.argsort(bigr)
rad_rings = bigr[sortedr]
taun_rings = bigtau[sortedr]
break
if case('d'):
if rad_rings.size > 1:
# find closest ring distance
rsel_idx = ind_ring(rad_rings, newr)
rad_rings = np.delete(rad_rings, rsel_idx)
taun_rings = np.delete(taun_rings, rsel_idx)
break
if case('m'):
rsel_idx = ind_ring(rad_rings, newr)
rad_rings[rsel_idx] = newr
taun_rings[rsel_idx] = exorings.tau_to_y(newtau)
break
if case('b'):
badring *= -1
print badring
break
if case('o'):
taun_rings = fmin(ringfunc, taun_rings, maxiter=1000, \
args=(time, flux, flux_err, rad_rings, res, kern, dstar))
break
if case('q'):
print 'q is for quitters'
break
if case():
print 'not a recognised keypress'
# print stats of fit
calc_ring_stats(taun_rings, time, flux, flux_err, rad_rings, res, \
kern, dstar, rings_tmin, rings_tmax)
# regenerate drawing
strip, dummy, g = exorings.ellipse_strip(rad_rings, exorings.y_to_tau(taun_rings), \
res[0], res[1], res[2], res[3], kern, dstar)
gt_abs = np.abs(g[0]-hjd_minr)
g1 = g[1]
gt_ingr = gt_abs[(g[0] <= hjd_minr)]
gt_egr = gt_abs[(g[0] > hjd_minr)]
g1_ingr = g1[(g[0] <= hjd_minr)]
g1_egr = g1[(g[0] > hjd_minr)]
# save the current axis ranges
x1, x2 = h1.get_xlim()
y1, y2 = h1.get_ylim()
h1.cla()
plot_folded_phot(fig_fold)
h1.plot(event.xdata, event.ydata, color='green')
h1.plot(np.abs(g[0]-hjd_minr), g[1])
h1.plot(gt_ingr, g1_ingr, color='blue')
h1.plot(gt_egr, g1_egr, color='red')
h1.plot(np.abs(g[0]-hjd_minr), g[2], color='orange')
h1.set_xlabel('Time from eclipse midpoint [days]')
h1.set_ylabel('Transmission')
# zoom back the the last view
h1.set_xlim([x1, x2])
h1.set_ylim([y1, y2])
plt.draw()
# save tmp version
# erase old version otherwise write_ring_fits throws a fit
if os.path.isfile('tmp.fits'):
os.remove('tmp.fits')
exorings.write_ring_fits('tmp.fits', res, taun_rings, rad_rings, dstar)
dstar = (Rstar * rsol * 2 / v) / 86400.
print 'Primary diameter = %5.2f days' % dstar
if read_in_ring_parameters:
print 'Reading in rings from %s' % fitsin_ring
(resxx, taun_rings, rad_rings, xxxdstar) = exorings.read_ring_fits(fitsin_ring)
else:
print "Starting with new rings...."
rad_rings = np.array([59.0])
taun_rings = np.array([0.0])
rad_rings = np.append(rad_rings, (100.))
taun_rings = np.append(taun_rings, (1000.))
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
if read_in_disk_parameters:
print 'Reading in disk parameters from %s' % fitsin_disk
(res, taun_ringsxx, rad_ringsxx, dstarxx) = exorings.read_ring_fits(fitsin_disk)
else:
# run minimizer to find best guess values
print 'No disk gradient parameters read in - refitting new ones....'
res = fmin(costfunc, np.array([y, dt, i_deg, phi_deg]), maxiter=5000, \
args=(grad_time, grad_mag_norm, tx))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit, grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2], res[3])
for opt, arg in opts:
if opt == '-h':
print help
sys.exit()
elif opt in ("-r", "--rfile"):
fitsin = arg
elif opt in ("-o", "--ofile"):
plotout = arg
elif opt in ("-s", "--vstar"):
v = np.array(float(arg))
print 'Reading in ring and disk parameters from %s' % fitsin
(res, taun_rings, rad_rings, dstar) = exorings.read_ring_fits(fitsin)
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit,
grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2],
res[3])
# produce fine grained gradient and ring values
samp_t = np.arange(-100, 100, 0.001) + 54222.
(samp_r, samp_g) = exorings.ring_grad_line(samp_t, res[0], res[1], res[2],
res[3])
hjd_minr = samp_t[np.argmin(samp_g)]
for opt, arg in opts:
if opt == '-h':
print help
sys.exit()
elif opt in ("-r", "--rfile"):
fitsin = arg
elif opt in ("-o", "--ofile"):
plotout = arg
elif opt in ("-s", "--vstar"):
v = np.array(float(arg))
print 'Reading in ring and disk parameters from %s' % fitsin
(res, taun_rings, rad_rings, dstar) = exorings.read_ring_fits(fitsin)
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit,
grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2],
res[3])
# produce fine grained gradient and ring values
samp_t = np.arange(-100, 100, 0.001) + 54222.
(samp_r, samp_g) = exorings.ring_grad_line(samp_t, res[0], res[1], res[2],
res[3])
hjd_minr = samp_t[np.argmin(samp_g)]
for opt, arg in opts:
if opt == '-h':
print help
sys.exit()
elif opt in ("-r", "--rfile"):
fitsin = arg
elif opt in ("-o", "--ofile"):
plotout = arg
elif opt in ("-s", "--vstar"):
v = np.array(float(arg))
print 'Reading in ring and disk parameters from %s' % fitsin
(res, taun_rings, rad_rings, dstar) = exorings.read_ring_fits(fitsin)
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit, grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2], res[3])
# produce fine grained gradient and ring values
samp_t = np.arange(-100, 100, 0.001) + 54222.
(samp_r, samp_g) = exorings.ring_grad_line(samp_t, res[0], res[1], res[2], res[3])
hjd_minr = samp_t[np.argmin(samp_g)]
# times when there is no photometry
(rstart, rend) = exorings.ring_mask_no_photometry(kern, dstar, time, res[0], res[1], res[2], res[3])
def onclick(event):
global rad_rings # naughty, I know, I know...
global taun_rings
global badring
for case in switch(event.key):
newt = event.xdata
newtau = event.ydata
print 'newtau is %f' % newtau
if newtau > 1.0:
newtau = 1.0
if newtau < 0.0:
newtau = 0.000001
newr = hjd_to_ring(newt + hjd_minr)
if case('r'):
exorings.plot_tilt_faceon(rad_rings, taun_rings, res, hjd_minr,
dstar)
break
if case('v'):
exorings.plot_tilt_faceon_vect(rad_rings, taun_rings, res, \
hjd_minr, rstart, rend, dstar)
break
if case('a'):
print 'a pressed, adding a new ring'
# assume that r is ordered
print ' inserting new ring date %d and radius %d' % (newt, newr)
bigr = np.append(rad_rings, newr)
bigtau = np.append(taun_rings, exorings.tau_to_y(newtau))
# index sort r, rearrange both r and tau to this
sortedr = np.argsort(bigr)
rad_rings = bigr[sortedr]
taun_rings = bigtau[sortedr]
break
if case('d'):
if rad_rings.size > 1:
# find closest ring distance
rsel_idx = ind_ring(rad_rings, newr)
rad_rings = np.delete(rad_rings, rsel_idx)
taun_rings = np.delete(taun_rings, rsel_idx)
break
if case('m'):
rsel_idx = ind_ring(rad_rings, newr)
rad_rings[rsel_idx] = newr
taun_rings[rsel_idx] = exorings.tau_to_y(newtau)
break
if case('b'):
badring *= -1
print badring
break
if case('o'):
taun_rings = fmin(ringfunc, taun_rings, maxiter=1000, \
args=(time, flux, flux_err, rad_rings, res, kern, dstar))
break
if case('q'):
print 'q is for quitters'
break
if case():
print 'not a recognised keypress'
# print stats of fit
calc_ring_stats(taun_rings, time, flux, flux_err, rad_rings, res, \
kern, dstar, rings_tmin, rings_tmax)
# regenerate drawing
strip, dummy, g = exorings.ellipse_strip(rad_rings, exorings.y_to_tau(taun_rings), \
res[0], res[1], res[2], res[3], kern, dstar)
gt_abs = np.abs(g[0] - hjd_minr)
g1 = g[1]
gt_ingr = gt_abs[(g[0] <= hjd_minr)]
gt_egr = gt_abs[(g[0] > hjd_minr)]
g1_ingr = g1[(g[0] <= hjd_minr)]
g1_egr = g1[(g[0] > hjd_minr)]
# save the current axis ranges
x1, x2 = h1.get_xlim()
y1, y2 = h1.get_ylim()
h1.cla()
plot_folded_phot(fig_fold)
h1.plot(event.xdata, event.ydata, color='green')
h1.plot(np.abs(g[0] - hjd_minr), g[1])
h1.plot(gt_ingr, g1_ingr, color='blue')
h1.plot(gt_egr, g1_egr, color='red')
h1.plot(np.abs(g[0] - hjd_minr), g[2], color='orange')
h1.set_xlabel('Time from eclipse midpoint [days]')
h1.set_ylabel('Transmission')
# zoom back the the last view
h1.set_xlim([x1, x2])
h1.set_ylim([y1, y2])
plt.draw()
# save tmp version
# erase old version otherwise write_ring_fits throws a fit
if os.path.isfile('tmp.fits'):
os.remove('tmp.fits')
exorings.write_ring_fits('tmp.fits', res, taun_rings, rad_rings, dstar)
dstar = (Rstar * rsol * 2 / v) / 86400.
print 'Primary diameter = %5.2f days' % dstar
if read_in_ring_parameters:
print 'Reading in rings from %s' % fitsin_ring
(resxx, taun_rings, rad_rings,
xxxdstar) = exorings.read_ring_fits(fitsin_ring)
else:
print "Starting with new rings...."
rad_rings = np.array([59.0])
taun_rings = np.array([0.0])
rad_rings = np.append(rad_rings, (100.))
taun_rings = np.append(taun_rings, (1000.))
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
if read_in_disk_parameters:
print 'Reading in disk parameters from %s' % fitsin_disk
(res, taun_ringsxx, rad_ringsxx,
dstarxx) = exorings.read_ring_fits(fitsin_disk)
else:
# run minimizer to find best guess values
print 'No disk gradient parameters read in - refitting new ones....'
res = fmin(costfunc, np.array([y, dt, i_deg, phi_deg]), maxiter=5000, \
args=(grad_time, grad_mag_norm, tx))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
for opt, arg in opts:
if opt == '-h':
print help
sys.exit()
elif opt in ("-r", "--rfile"):
fitsin = arg
elif opt in ("-o", "--ofile"):
plotout = arg
elif opt in ("-s", "--vstar"):
v = np.array(float(arg))
print 'Reading in ring and disk parameters from %s' % fitsin
(res, taun_rings, rad_rings, dstar) = exorings.read_ring_fits(fitsin)
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit, grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2], res[3])
# produce fine grained gradient and ring values
samp_t = np.arange(-100, 100, 0.001) + 54222.
(samp_r, samp_g) = exorings.ring_grad_line(samp_t, res[0], res[1], res[2], res[3])
hjd_minr = samp_t[np.argmin(samp_g)]
# times when there is no photometry
(rstart, rend) = exorings.ring_mask_no_photometry(kern, dstar, time, res[0], res[1], res[2], res[3])
rmask = (rstart < 40)
for opt, arg in opts:
if opt == '-h':
print help
sys.exit()
elif opt in ("-r", "--rfile"):
fitsin = arg
elif opt in ("-o", "--ofile"):
plotout = arg
elif opt in ("-s", "--vstar"):
v = np.array(float(arg))
print 'Reading in ring and disk parameters from %s' % fitsin
(res, taun_rings, rad_rings, dstar) = exorings.read_ring_fits(fitsin)
exorings.print_ring_tau(rad_rings, exorings.y_to_tau(taun_rings))
# set up stellar disk
kern = exorings.make_star_limbd(21, 0.8)
# make the radius and projected gradient for the measured gradient points
(ring_disk_fit, grad_disk_fit) = exorings.ring_grad_line(grad_time, res[0], res[1], res[2], res[3])
# produce fine grained gradient and ring values
samp_t = np.arange(-100, 100, 0.001) + 54222.
(samp_r, samp_g) = exorings.ring_grad_line(samp_t, res[0], res[1], res[2], res[3])
hjd_minr = samp_t[np.argmin(samp_g)]
(rstart, rend) = exorings.ring_mask_no_photometry(kern, dstar, time, res[0], res[1], res[2], res[3])
rmask = (rstart < 40)
