def main():
progname = 'find_dist_pbdot_mc.py'
args = get_opt(progname)
ra_deg = 360.0 * (args.ra[0] + args.ra[1]/60. + args.ra[2]/3600.)/24.0
if(args.raerr != None):
raerr = (args.raerr/3600.)*360./24.0 # Convert to degrees
raerr_str = ' +/- '+str(raerr)
else:
raerr_str = ''
if(args.dec[0] < 0):
sgn = -1.
else:
sgn = 1.
dec_deg = sgn * (np.abs(args.dec[0]) + args.dec[1]/60. + args.dec[2]/3600.)
if(args.decerr != None):
decerr = args.decerr/3600. # Convert to degrees
decerr_str = ' +/- '+ str(decerr)
else:
decerr_str = ''
print 'RA = ', ra_deg, raerr_str, ' deg '
print 'Dec = ', dec_deg, decerr_str, ' deg '
# Convert nominal values of ra and dec to l and b here just for show
c = coord.ICRSCoordinates(ra=ra_deg, dec=dec_deg, unit=(u.degree, u.degree))
l = c.galactic.l.degrees
b = c.galactic.b.degrees
print 'l = ', c.galactic.l.degrees, ' deg = ', c.galactic.l.radians, ' rad'
print 'b = ', c.galactic.b.degrees, ' deg = ', c.galactic.b.radians, ' rad'
if(args.xpbdoterr != None):
xpbdoterr = ' +/- '+str(args.xpbdoterr)
else:
xpbdoterr = ''
print 'Xpbdot = ', args.xpbdot, xpbdoterr
if(args.pmerr != None):
pmerr = ' +/- '+str(args.pmerr)
else:
pmerr = ''
print 'PM = ', args.pm, pmerr, ' mas/yr'
# convert pb to seconds
args.pb = u.day.to(u.s, args.pb)
# convert velocity/ velocity error to kpc/s
# No need to convert R0, which is already in R0
args.v0 = u.km.to(u.kpc, args.v0)
args.v0err = u.km.to(u.kpc, args.v0err)
# convert pm to rad/s
args.pm = (u.mas.to(u.radian, args.pm)) / u.year.to(u.s, 1.0)
if(args.pmerr != None):
args.pmerr = (u.mas.to(u.radian, args.pmerr)) / u.year.to(u.s, 1.0)
# Check whether all error bars required are given for distance and proper
# motion. If they are, go ahead with the random number generation thing.
# If not, just do a quick straight calculation of the pbdot correction.
if((args.pmerr != None) & (args.xpbdoterr != None) & (args.raerr != None) & (args.decerr != None)):
# For each of the varying parameters (dist, pm, v0, R0), generate
# args.niter gaussian deviates, with mean = value and sigma = error
xpbdot = np.random.normal(args.xpbdot, args.xpbdoterr, args.niter)
pm = np.random.normal(args.pm, args.pmerr, args.niter)
v0 = np.random.normal(args.v0, args.v0err, args.niter)
R0 = np.random.normal(args.R0, args.R0err, args.niter)
ra = np.random.normal(ra_deg, raerr, args.niter)
dec = np.random.normal(dec_deg, decerr, args.niter)
# Plot distributions for each parameter:
fig_param = plt.figure()
# xpbdot
ax_param = fig_param.add_subplot(321)
hist, bin_val = np.histogram(xpbdot, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
A, mu, sig = utils.fitgauss(bin_val, hist)
#ax_param.plot(bin_val, utils.gaussian(bin_val, A, mu, sig), color='red')
ax_param.set_xlabel('Difference in orbital decay ($10^{14}$ s/s)')
# proper motion
ax_param = fig_param.add_subplot(322)
hist, bin_val = np.histogram(pm, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
ax_param.set_xlabel('Proper motion (mas/yr)')
# v0
ax_param = fig_param.add_subplot(323)
hist, bin_val = np.histogram(v0, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
ax_param.set_xlabel('Solar velocity (km/s)')
# R0
ax_param = fig_param.add_subplot(324)
hist, bin_val = np.histogram(R0, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
ax_param.set_xlabel('Galactocentric distance (kpc)')
# RA
ax_param = fig_param.add_subplot(325)
hist, bin_val = np.histogram(ra, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
ax_param.set_xlabel('Right ascension (deg)')
# Dec
ax_param = fig_param.add_subplot(326)
hist, bin_val = np.histogram(dec, bins=32)
bin_size = (bin_val[1]-bin_val[0])
bin_val = bin_val[0:len(bin_val)-1] + bin_size/2.
ax_param.plot(bin_val, hist, linestyle='steps-mid', linewidth=1.4,
color='blue')
ax_param.set_xlabel('Declination (deg)')
plt.show()
# Run calcuation of pbdot correction through each iteration and build
# up array of values for pbdot correction
#for i_iter in np.arange(args.niter):
sys.stdout.write('\n {0:<7s} {1:<7s} {2:5s} {3:<10s}\n'.format('Iter', '% done', 'n_bad', 'n_negative'))
dist_array = []
n_bad = 0
n_negative = 0
n_step_worked = 0
for i_iter in np.arange(args.niter):
c = coord.ICRSCoordinates(ra=ra[i_iter], dec=dec[i_iter], unit=(u.degree, u.degree))
l = c.galactic.l.radians
b = c.galactic.b.radians
n_tries = 0
sgn_guess = 1.0
dist_guess = args.distguess
dist_trial = -1.0
try:
# while(dist_trial < 0.001 and n_tries <= 25):
dist_trial = newton(pc.dist_func, dist_guess,
args=(l, b,
v0[i_iter], R0[i_iter],
pm[i_iter], args.pb,
xpbdot[i_iter]))
# n_tries += 1
# sgn_guess *= -1.0
# dist_guess = args.distguess + sgn_guess*np.ceil(float(n_tries)/2.0)*args.stepguess
except RuntimeError:
n_bad+=1
else: # If there are no errors, then append to array
# if(dist_trial > 0.000001 and n_tries <= 25):
if(dist_trial > 0.001):
dist_array.append(dist_trial)
# if(n_tries > 1):
# n_step_worked += 1
else:
n_negative += 1
# dist_array.append(pc.get_dist_pbdot(ra[i_iter], dec[i_iter],
# pm[i_iter], xpbdot[i_iter],
# args.pb,
# v0=v0[i_iter],
# R0=R0[i_iter],
# dist_guess=args.distguess))
if(np.fmod(i_iter, 16)==0):
display_status(i_iter, args.niter, n_bad, n_negative)
dist_array = np.array(dist_array)
print '\nNumber of Runtime Errors = ', n_bad
print 'Number of times stepping guess value worked = ', n_step_worked
# Create histogram of pbdot correction, and fit a gsaussian to it.
# Extract median (mean?) and sigma as our final value and error.
if(args.distlim == None):
distlim = (np.amin(dist_array)-0.1*np.abs(np.amin(dist_array)),
np.amax(dist_array)+0.1*np.abs(np.amax(dist_array)))
else:
distlim = (args.distlim[0], args.distlim[1])
dist_pdf, bin_val = np.histogram(dist_array,
range=distlim, bins=96, density=True)
bin_size = bin_val[1]-bin_val[0]
dist_x = bin_val[0:len(bin_val)-1] + 0.5*bin_size
#fig_pbdot = plt.figure()
#ax_pbdot = fig_pbdot.add_axes([0.12, 0.1, 0.8, 0.85])
#ax_pbdot.plot(pbdot_x, pbdot_pdf, linestyle='steps-mid', linewidth=1.4,
# color='blue')
# Get upper limit from distance distribution
# dist_med, dist_min, dist_max = \
dist_upper = get_pdf_prob(dist_x, dist_pdf, prob_intervals,
norm=True, upper=True) #, \
plot_pdf(dist_x, dist_pdf, \
### weights=alpha_weights, \
xlabel='Distance (kpc)', \
ylabel='Probability density',\
prob_lines=dist_upper,\
# prob_lines=np.append(dist_min, dist_max),\
prob_linestyle=['dashed','dashdot','dotted'] #, \
# 'dashed','dashdot','dotted'], \
)
print " "
print "DISTANCE (kpc): "
print " 68%: < ", dist_upper[0]
print " 95%: < ", dist_upper[1]
print " 99%: < ", dist_upper[2]
print " "
# plt.axvline(dist_med)
# dist_err_high = np.abs(dist_max-dist_med)
# dist_err_low = np.abs(dist_min-dist_med)
#A, pbdot_corr, pbdot_corr_err = utils.fitgauss(pbdot_x, pbdot_pdf)
#ax_pbdot.plot(pbdot_x, utils.gaussian(pbdot_x, A, pbdot_corr, pbdot_corr_err),
#linestyle='solid', linewidth=1.4, color='red')
plt.savefig('dist_pbdot.pdf')
#print 'Corrected Pbdot = ', pbdot_corr, ' +/- ', pbdot_corr_err
# print 'Distance (kpc) = ', dist_med, ' -', dist_err_low[0], \
# ' +', dist_err_high[0]
# print 'Distance peak = ', dist_x[np.argmax(dist_pdf)]
# pbdot_new = args.pbdot + pbdot_corr_med
#pbdot_corr_err_mean = np.mean(np.array([pbdot_corr_err_low[0],pbdot_corr_err_high[0]]))
#pbdot_new_err = np.sqrt(args.pbdoterr**2.0 + pbdot_corr_err_mean**2.0)
# pbdot_new_err_low = np.sqrt(args.pbdoterr**2.0 + pbdot_corr_err_low**2.0)
# pbdot_new_err_high = np.sqrt(args.pbdoterr**2.0 + pbdot_corr_err_high**2.0)
# print 'Pbdot new (median and mean error) = ', pbdot_new, ' +/- ', pbdot_new_err
# print 'Pbdot new range (68%) = ', pbdot_corr_min[0], ' , ', pbdot_corr_max[0]
# pbdot_new_mid = np.mean(np.array([pbdot_new - pbdot_new_err_low[0], pbdot_new + pbdot_new_err_high[0]]))
# print 'Pbdot new (using midpoint between ranges) = ', pbdot_new_mid, ' - ', pbdot_new_mid - (pbdot_new-pbdot_new_err_low[0]), ' + ', (pbdot_new+pbdot_new_err_high[0])-pbdot_new_mid
# print 'Pbdot new (median corr and asymmetric error) = ', pbdot_new, ' - ', pbdot_new_err_low[0], ' + ', pbdot_new_err_high[0]
else:
c = coord.ICRSCoordinates(ra=ra_deg, dec=dec_deg,
unit=(u.degree, u.degree))
l = c.galactic.l.radians
b = c.galactic.b.radians
dist_pbdot = newton(pc.dist_func, args.distguess,
args=(l, b, args.v0, args.R0,
args.pm, args.pb, args.xpbdot))
#pbdot_new = args.pbdot - pbdot_corr
print '\nWill not calculate final uncertainties...\n'
print 'Distance (kpc) = ', dist_pbdot
return
def main():
progname = "find_pbdot_corr_mc.py"
args = get_opt(progname)
ra_deg = 360.0 * (args.ra[0] + args.ra[1] / 60.0 + args.ra[2] / 3600.0) / 24.0
if args.raerr != None:
raerr = (args.raerr / 3600.0) * 360.0 / 24.0 # Convert to degrees
raerr_str = " +/- " + str(raerr)
else:
raerr_str = ""
if args.dec[0] < 0:
sgn = -1.0
else:
sgn = 1.0
dec_deg = sgn * (np.abs(args.dec[0]) + args.dec[1] / 60.0 + args.dec[2] / 3600.0)
if args.decerr != None:
decerr = args.decerr / 3600.0 # Convert to degrees
decerr_str = " +/- " + str(decerr)
else:
decerr_str = ""
print "RA = ", ra_deg, raerr_str, " deg "
print "Dec = ", dec_deg, decerr_str, " deg "
# Convert nominal values of ra and dec to l and b here just for show
c = coord.ICRSCoordinates(ra=ra_deg, dec=dec_deg, unit=(u.degree, u.degree))
l = c.galactic.l.degrees
b = c.galactic.b.degrees
print "l = ", l, " deg"
print "b = ", b, " deg"
if args.disterr != None:
derr = " +/- " + str(args.disterr)
else:
derr = ""
print "dist = ", args.dist, derr, " kpc"
if args.pmerr != None:
pmerr = " +/- " + str(args.pmerr)
else:
pmerr = ""
print "PM = ", args.pm, pmerr, " mas/yr"
print "Pbdot = ", args.pbdot
if args.pbdoterr != None:
print "Pbdot error = ", args.pbdoterr
# convert pb to seconds
args.pb = u.day.to(u.s, args.pb)
# Check whether all error bars required are given for distance and proper
# motion. If they are, go ahead with the random number generation thing.
# If not, just do a quick straight calculation of the pbdot correction.
if (
(args.disterr != None)
& (args.pmerr != None)
& (args.pbdoterr != None)
& (args.raerr != None)
& (args.decerr != None)
):
# For each of the varying parameters (dist, pm, v0, R0), generate
# args.niter gaussian deviates, with mean = value and sigma = error
dist = np.random.normal(args.dist, args.disterr, args.niter)
pm = np.random.normal(args.pm, args.pmerr, args.niter)
v0 = np.random.normal(args.v0, args.v0err, args.niter)
R0 = np.random.normal(args.R0, args.R0err, args.niter)
ra = np.random.normal(ra_deg, raerr, args.niter)
dec = np.random.normal(dec_deg, decerr, args.niter)
# Plot distributions for each parameter:
fig_param = plt.figure()
# distance
ax_param = fig_param.add_subplot(321)
hist, bin_val = np.histogram(dist, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
A, mu, sig = utils.fitgauss(bin_val, hist)
# ax_param.plot(bin_val, utils.gaussian(bin_val, A, mu, sig), color='red')
ax_param.set_xlabel("Distance (kpc)")
# proper motion
ax_param = fig_param.add_subplot(322)
hist, bin_val = np.histogram(pm, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
ax_param.set_xlabel("Proper motion (mas/yr)")
# v0
ax_param = fig_param.add_subplot(323)
hist, bin_val = np.histogram(v0, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
ax_param.set_xlabel("Solar velocity (km/s)")
# R0
ax_param = fig_param.add_subplot(324)
hist, bin_val = np.histogram(R0, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
ax_param.set_xlabel("Galactocentric distance (kpc)")
# RA
ax_param = fig_param.add_subplot(325)
hist, bin_val = np.histogram(ra, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
ax_param.set_xlabel("Right ascension (deg)")
# Dec
ax_param = fig_param.add_subplot(326)
hist, bin_val = np.histogram(dec, bins=32)
bin_size = bin_val[1] - bin_val[0]
bin_val = bin_val[0 : len(bin_val) - 1] + bin_size / 2.0
ax_param.plot(bin_val, hist, linestyle="steps-mid", linewidth=1.4, color="blue")
ax_param.set_xlabel("Declination (deg)")
plt.show()
# Run calcuation of pbdot correction through each iteration and build
# up array of values for pbdot correction
# for i_iter in np.arange(args.niter):
sys.stdout.write("\n {0:<7s} {1:<7s} \n".format("Iter", "% done"))
pbdot_corr_array = []
for i_iter in np.arange(args.niter):
pbdot_corr_array.append(
pc.get_pbdot_corr(
ra[i_iter], dec[i_iter], pm[i_iter], dist[i_iter], args.pb, v0=v0[i_iter], R0=R0[i_iter]
)
)
if np.fmod(i_iter, 16) == 0:
display_status(i_iter, args.niter)
pbdot_corr_array = np.array(pbdot_corr_array)
# Create histogram of pbdot correction, and fit a gsaussian to it.
# Extract median (mean?) and sigma as our final value and error.
if args.corrlim == None:
corrlim = (
np.amin(pbdot_corr_array) - 0.1 * np.abs(np.amin(pbdot_corr_array)),
np.amax(pbdot_corr_array) + 0.1 * np.abs(np.amax(pbdot_corr_array)),
)
else:
corrlim = (args.corrlim[0] * 10.0 ** (-12), args.corrlim[1] * 10 ** (-12))
pbdot_pdf, bin_val = np.histogram(pbdot_corr_array, range=corrlim, bins=192, density=True)
bin_size = bin_val[1] - bin_val[0]
pbdot_x = bin_val[0 : len(bin_val) - 1] + 0.5 * bin_size
# fig_pbdot = plt.figure()
# ax_pbdot = fig_pbdot.add_axes([0.12, 0.1, 0.8, 0.85])
# ax_pbdot.plot(pbdot_x, pbdot_pdf, linestyle='steps-mid', linewidth=1.4,
# color='blue')
pbdot_corr_med, pbdot_corr_min, pbdot_corr_max = get_pdf_prob(
pbdot_x, pbdot_pdf, prob_intervals, norm=True
) # , \
plot_pdf(
pbdot_x,
pbdot_pdf,
### weights=alpha_weights, \
xlabel="$\\dot{P}_{b, corr}}$",
ylabel="Probability density",
prob_lines=np.append(pbdot_corr_min, pbdot_corr_max),
prob_linestyle=["dashed", "dashdot", "dotted", "dashed", "dashdot", "dotted"],
)
plt.axvline(pbdot_corr_med)
pbdot_corr_err_high = np.abs(pbdot_corr_max - pbdot_corr_med)
pbdot_corr_err_low = np.abs(pbdot_corr_min - pbdot_corr_med)
# A, pbdot_corr, pbdot_corr_err = utils.fitgauss(pbdot_x, pbdot_pdf)
# ax_pbdot.plot(pbdot_x, utils.gaussian(pbdot_x, A, pbdot_corr, pbdot_corr_err),
# linestyle='solid', linewidth=1.4, color='red')
plt.savefig("pbdot_new.png")
print ""
print "Original Pbdot = ", args.pbdot
print ""
# print 'Corrected Pbdot = ', pbdot_corr, ' +/- ', pbdot_corr_err
print "Pbdot correction = ", pbdot_corr_med, " +", pbdot_corr_err_high[0], " -", pbdot_corr_err_low[0]
print "Pbdot correction peak = ", pbdot_x[np.argmax(pbdot_pdf)]
pbdot_new = args.pbdot - pbdot_corr_med
# pbdot_corr_err_mean = np.mean(np.array([pbdot_corr_err_low[0],pbdot_corr_err_high[0]]))
# pbdot_new_err = np.sqrt(args.pbdoterr**2.0 + pbdot_corr_err_mean**2.0)
pbdot_new_err_low = np.sqrt(args.pbdoterr ** 2.0 + pbdot_corr_err_low ** 2.0)
pbdot_new_err_high = np.sqrt(args.pbdoterr ** 2.0 + pbdot_corr_err_high ** 2.0)
# print 'Pbdot new (median and mean error) = ', pbdot_new, ' +/- ', pbdot_new_err
# print 'Pbdot new range (68%) = ', pbdot_corr_min[0], ' , ', pbdot_corr_max[0]
pbdot_new_mid = np.mean(np.array([pbdot_new - pbdot_new_err_low[0], pbdot_new + pbdot_new_err_high[0]]))
print "Pbdot new (using midpoint between ranges) = ", pbdot_new_mid, " - ", pbdot_new_mid - (
pbdot_new - pbdot_new_err_low[0]
), " + ", (pbdot_new + pbdot_new_err_high[0]) - pbdot_new_mid
print "Pbdot new (median corr and asymmetric error) = ", pbdot_new, " - ", pbdot_new_err_low[
0
], " + ", pbdot_new_err_high[0]
else:
pbdot_corr = pc.get_pbdot_corr(
args.ra, args.dec, args.pm, args.dist, args.pb, v0=args.v0, R0=args.R0, verbose=True
)
pbdot_new = args.pbdot - pbdot_corr
print "\nWill not calculate final uncertainties...\n"
print "Original Pbdot = ", args.pbdot
print "Corrected Pbdot = ", pbdot_new
return
