def bound_mass_evolution(prop, output, max_time=None):
plt.clf()
times = nbody_tools.get_times(prop['simFileGRAVOUTPUT'])
if (max_time):
times = times[times < max_time]
mass_dm = np.ones(len(times))
mass_st = np.ones(len(times))
for i, t in enumerate(times):
data, centre = grab_snapshot(prop, t, with_centre=True)
data = data_to_pandas(data)
mass_dm[i] = np.sum(data.mass[data.dm == 1.])
mass_st[i] = np.sum(data.mass[data.dm == 0.])
plt.plot(times, mass_dm / mass_dm[0], ls='dotted', label='DM')
plt.plot(times, mass_st / mass_st[0], ls='dashed', label='Stars')
plt.plot(times, (mass_st + mass_dm) / (mass_dm[0] + mass_st[0]),
color='k',
label='Total')
plt.xlabel(r'$t/\mathrm{Gyr}$')
plt.ylabel(r'$M/M_\mathrm{init}$')
plt.legend()
plt.semilogy()
plt.savefig(output, bbox_inches='tight')
def shape_evolution(prop, output, max_time=None):
plt.clf()
times = nbody_tools.get_times(prop['simFileGRAVOUTPUT'])
if (max_time):
times = times[times < max_time]
ba_dm = np.ones(len(times))
ca_dm = np.ones_like(ba_dm)
align_dm = np.ones_like(ba_dm)
ba_st = np.ones(len(times))
ca_st = np.ones_like(ba_st)
align_st = np.ones_like(ba_st)
for i, t in enumerate(times):
data, centre = grab_snapshot(prop, t, with_centre=True)
if not len(data):
continue
data = data_to_pandas(data)
ba_st[i], ca_st[i], pr = moment_of_inertia_diagonalization(
data[data.dm == 0.],
# no offset
0.,
0.,
0.,
reduced=True)[:3]
align_st[i] = np.abs(
np.dot(pr, centre[:3]) / np.sqrt(np.sum(centre[:3]**2)))
ba_dm[i], ca_dm[i], pr = moment_of_inertia_diagonalization(
data[data.dm == 1.],
# no offset
0.,
0.,
0.,
reduced=True)[:3]
align_dm[i] = np.abs(
np.dot(pr, centre[:3]) / np.sqrt(np.sum(centre[:3]**2)))
# times/=nbody_tools.kms2kpcGyr
return shape_evolution_plot(output, ba_dm, ca_dm, align_dm, ba_st, ca_st,
align_st, times)
def all_evolution(prop, max_time=None, prune_radius=None):
plt.clf()
times = nbody_tools.get_times(prop['simFileGRAVOUTPUT'])
if (max_time):
times = times[times < max_time]
mass_dm = np.ones(len(times))
mass_st = np.ones(len(times))
mass_dm_core = np.ones(len(times))
mass_st_core = np.ones(len(times))
mass_dm_core_init = np.ones(len(times))
mass_st_core_init = np.ones(len(times))
mass_dm_core_tilde = np.ones(len(times))
mass_st_core_tilde = np.ones(len(times))
ba_dm = np.ones(len(times))
ca_dm = np.ones_like(ba_dm)
align_dm = np.ones_like(ba_dm)
ba_st = np.ones(len(times))
ca_st = np.ones_like(ba_st)
align_st = np.ones_like(ba_st)
r_half = np.zeros_like(ba_st)
eps = np.ones_like(ba_st)
r_core = np.zeros_like(ba_st)
eps_king = np.ones_like(ba_st)
c_king = np.ones_like(ba_st)
sig_maj = np.zeros_like(ba_st)
sig_int = np.zeros_like(ba_st)
sig_min = np.zeros_like(ba_st)
sig_proj = np.zeros_like(ba_st)
sig_proj_flat = np.zeros_like(ba_st)
centre_x = np.zeros_like(ba_st)
centre_y = np.zeros_like(ba_st)
centre_z = np.zeros_like(ba_st)
centre_vx = np.zeros_like(ba_st)
centre_vy = np.zeros_like(ba_st)
centre_vz = np.zeros_like(ba_st)
eq_centre_s = np.zeros_like(ba_st)
eq_centre_vlos = np.zeros_like(ba_st)
for i, t in enumerate(times):
data, centre = grab_snapshot(prop, t, with_centre=True)
if not len(data):
continue
data = data_to_pandas(data)
ba_dm[i], ca_dm[
i], pmaj, pint, pmin = moment_of_inertia_diagonalization(
data[data.dm == 1.],
# no offset
0.,
0.,
0.,
reduced=True)
align_dm[i] = np.abs(
np.dot(pmaj, centre[:3]) / np.sqrt(np.sum(centre[:3]**2)))
ba_st[i], ca_st[
i], pmaj, pint, pmin = moment_of_inertia_diagonalization(
data[data.dm == 0.],
# no offset
0.,
0.,
0.,
reduced=True)
align_st[i] = np.abs(
np.dot(pmaj, centre[:3]) / np.sqrt(np.sum(centre[:3]**2)))
xtil = np.sum(pmaj * data.values.T[1:4].T, axis=1)
ytil = np.sum(pint * data.values.T[1:4].T, axis=1)
ztil = np.sum(pmin * data.values.T[1:4].T, axis=1)
if np.count_nonzero(data.dm == 0.):
vmaj = np.sum(pmaj * data[data.dm == 0.].values.T[4:7].T, axis=1)
vint = np.sum(pint * data[data.dm == 0.].values.T[4:7].T, axis=1)
vmin = np.sum(pmin * data[data.dm == 0.].values.T[4:7].T, axis=1)
vmaj -= np.sum(data.mass[data.dm == 0.] * vmaj) / np.sum(
data.mass[data.dm == 0.])
vint -= np.sum(data.mass[data.dm == 0.] * vint) / np.sum(
data.mass[data.dm == 0.])
vmin -= np.sum(data.mass[data.dm == 0.] * vmin) / np.sum(
data.mass[data.dm == 0.])
sig_maj[i] = np.sqrt(
np.sum(data.mass[data.dm == 0.] * vmaj**2) /
np.sum(data.mass[data.dm == 0.]))
sig_int[i] = np.sqrt(
np.sum(data.mass[data.dm == 0.] * vint**2) /
np.sum(data.mass[data.dm == 0.]))
sig_min[i] = np.sqrt(
np.sum(data.mass[data.dm == 0.] * vmin**2) /
np.sum(data.mass[data.dm == 0.]))
mass_dm[i] = np.sum(data.mass[data.dm == 1.])
mass_st[i] = np.sum(data.mass[data.dm == 0.])
radii = np.sqrt(data.x**2 + data.y**2 + data.z**2)
radii_tilde = np.sqrt(xtil**2 + (ytil / ba_st[i])**2 +
(ztil / ca_st[i])**2)
mass_dm_core_init[i] = np.sum(data.mass[(data.dm == 1.)
& (radii < prop['rs'])])
mass_st_core_init[i] = np.sum(data.mass[(data.dm == 0.)
& (radii < prop['rs'])])
# data,centre = grab_snapshot(prop,t,with_centre=True,recentre_sim=False)
# data = data_to_pandas(data)
data_recentre = recentre(data.values[:, :7], centre=-np.copy(centre))
for ii, nn in enumerate(['x', 'y', 'z', 'vx', 'vy', 'vz']):
data[nn] = data_recentre[:, ii + 1]
if np.count_nonzero(data.dm == 0.):
results, xtmp, ytmp = nbody_tools.fit_surface_density(
data[data.dm == 0.].values, centre=np.copy(centre))
r_half[i] = results[0]
eps[i] = results[1]
results, xtmp, ytmp = nbody_tools.fit_surface_density(
data[data.dm == 0.].values,
centre=np.copy(centre),
profile="King")
r_core[i] = results[0]
eps_king[i] = results[1]
c_king[i] = results[-1]
print '#######################'
print 'King fit ##############'
print r_core[i], c_king[i]
print '#######################'
mass_dm_core[i] = np.sum(data.mass[(data.dm == 1.)
& (radii < r_half[i])])
mass_st_core[i] = np.sum(data.mass[(data.dm == 0.)
& (radii < r_half[i])])
mass_dm_core_tilde[i] = np.sum(data.mass[(data.dm == 1.)
& (radii_tilde < r_half[i])])
mass_st_core_tilde[i] = np.sum(data.mass[(data.dm == 0.)
& (radii_tilde < r_half[i])])
## Solar velocity doesn't matter here
eq = np.array([
aa_py.CartesianToGalactic(
ii[1:7], np.array([8.2, 0., 0., 11.1, 240., 7.25]))
for ii in data[data.dm == 0.].values
])
eq_centre = aa_py.CartesianToGalactic(
centre[:6], np.array([8.2, 0., 0., 11.1, 240., 7.25]))
from scipy.optimize import minimize
from scipy.misc import logsumexp
def logl(p, x):
if p[1] < 0. or p[3] < 0. or p[-1] < 0. or p[-1] > 1.:
return np.inf
return -np.sum(
np.logaddexp(
-(x - p[0])**2 / 2. / p[1]**2 - .5 * np.log(p[1]**2) +
np.log(1. - p[-1]),
(-(x - p[2])**2 / 2. / p[3]**2 - .5 * np.log(p[3]**2) +
np.log(p[-1]))))
if np.count_nonzero(data.dm == 0.):
## Changed this to see if it makes a difference
vproj = eq.T[3] - eq_centre[3]
vproj -= np.sum(data.mass[data.dm == 0.] * vproj) / np.sum(
data.mass[data.dm == 0.])
Rvec = centre[:3] - np.array([8.2, 0., 0.])
Rvec /= np.sqrt(np.sum(Rvec**2))
vprojR = data.vx * Rvec[0] + data.vy * Rvec[1] + data.vz * Rvec[2]
vprojR = vprojR[data.dm == 0]
vprojR -= np.sum(data.mass[data.dm == 0.] * vprojR) / np.sum(
data.mass[data.dm == 0.])
sig_proj[i] = minimize(
logl, [0.05, sig_maj[i], 0.05, 2. * sig_maj[i], 0.2],
args=(vproj, )).x[1]
sig_proj_flat[i] = minimize(
logl, [0.05, sig_maj[i], 0.05, 2. * sig_maj[i], 0.2],
args=(vprojR, )).x[1]
# sig_proj[i]=np.sqrt(np.sum(data.mass[data.dm==0.]*vproj**2)/np.sum(data.mass[data.dm==0.]))
# print i, sig_maj[i], sig_min[i], sig_proj[i], centre, eq_centre, np.count_nonzero(data.dm==0.)
print '#####################'
print t, r_half[i], sig_proj[i]
print '#####################'
centre_x[i] = centre[0]
centre_y[i] = centre[1]
centre_z[i] = centre[2]
centre_vx[i] = centre[3]
centre_vy[i] = centre[4]
centre_vz[i] = centre[5]
eq_centre_s[i] = eq_centre[2]
eq_centre_vlos[i] = eq_centre[3]
# times/=nbody_tools.kms2kpcGyr
names = [
'times', 'mass_dm', 'mass_st', 'ba_dm', 'ca_dm', 'align_dm', 'ba_st',
'ca_st', 'align_st', 'r_half', 'eps', 'r_core', 'eps_king', 'c_king',
'sig_maj', 'sig_int', 'sig_min', 'sig_proj', 'sig_proj_flat',
'mass_dm_core', 'mass_st_core', 'mass_dm_core_tilde',
'mass_st_core_tilde', 'mass_dm_core_init', 'mass_st_core_init', 'x',
'y', 'z', 'vx', 'vy', 'vz', 's', 'vlos'
]
outputs = np.vstack(
(times, mass_dm, mass_st, ba_dm, ca_dm, align_dm, ba_st, ca_st,
align_st, r_half, eps, r_core, eps_king, c_king, sig_maj, sig_int,
sig_min, sig_proj, sig_proj_flat, mass_dm_core, mass_st_core,
mass_dm_core_tilde, mass_st_core_tilde, mass_dm_core_init,
mass_st_core_init, centre_x, centre_y, centre_z, centre_vx, centre_vy,
centre_vz, eq_centre_s, eq_centre_vlos))
return pd.DataFrame(outputs.T, columns=names)
propermotions = [caldwell_propermotion, [0., 0.], [0., 0.17]]
rot = [[84., 208., 0.]]
veldisp = 10.
if __name__ == '__main__':
for p in list(itertools.product(*[segregation, propermotions, shapes])):
print p
simproperties = generate_simproperties(SegregationParameter=p[0],
propermotion=p[1],
flattening=p[2],
fixed_veldisp=veldisp)
generate_plum_in_nfw(simproperties, nmax=20, int_step_frac=50.)
times = nbody_tools.get_times(simproperties['simFileGRAVOUTPUT'])
# compare_1d_profiles_2(simproperties,
# simproperties['sim_name'] + '1dprof.pdf',
# times=[0., times[-1]])
# compare_1d_sigmar_profiles_2(simproperties,
# simproperties['sim_name'] +
# '1dsrprof.pdf',
# times=[0., times[-1]], sigmayaxis=100.)
# shape_evolution(simproperties,
# simproperties['sim_name'] + 'shape.pdf',
# max_time=times[-1])
# bound_mass_evolution(simproperties,
# simproperties['sim_name'] + 'mass.pdf',
# max_time=times[-1])