def plot_bott_lpparam_glatparam(self, lgbott=None, lpparam='delta', glatparam='ABDelta',
outname=None, xlabel=None, ylabel=None, title=None, check_omegacs=False):
"""Plot phase diagram with lattice param varying on x axis and glatparam varying on y axis
Parameters
----------
lgbott : N x 3 float array or None
lattice params array, glatparam array, bott indices array
glatparam : str
outname : str or None
"""
if lgbott is None:
lpV, glatV, botts, omegacs = self.collect_botts_lpparam_glatparam(glatparam=glatparam)
# lgbott is the [lattice param, glat param, botts] vector
lgbott = np.dstack((lpV, glatV, botts, omegacs))[0]
print 'lpV, glatV, botts:'
print 'lgbott = ', lgbott
ngrid = len(lgbott) + 7
fig, ax, cbar_ax = leplt.initialize_1panel_cbar_cent()
leplt.plot_pcolormesh(lgbott[:, 0], lgbott[:, 1], lgbott[:, 2], ngrid, ax=ax, cax=cbar_ax, method='nearest',
make_cbar=True, cmap=lecmaps.diverging_cmap(250, 10, l=30),
vmin=-1., vmax=1., title=None, xlabel=None, ylabel=None, ylabel_right=True,
ylabel_rot=90, cax_label=r'Bott index, $B$',
cbar_labelpad=-30, cbar_orientation='horizontal',
ticks=[-1, 0, 1], fontsize=8, title_axX=None, title_axY=None, alpha=1.0)
if title is not None:
cbar_ax.text(0.5, 0.95, title, ha='center', va='center', transform=fig.transFigure)
if xlabel is None:
xlabel = leplt.param2description(glatparam)
if ylabel is None:
ylabel = leplt.param2description(lpparam)
ax.set_xlabel(xlabel)
ax.set_ylabel(ylabel)
if outname is not None:
print 'saving figure to ' + outname
plt.savefig(outname)
if check_omegacs:
fig2, ax2 = leplt.initialize_1panel_centered_fig(wsfrac=0.5, tspace=4)
leplt.plot_pcolormesh(lgbott[:, 0], lgbott[:, 1], omegacs[:, 3], ngrid, ax=ax2, cax=cbar_ax, method='nearest',
make_cbar=True, cmap=lecmaps.diverging_cmap(250, 10, l=30),
vmin=-1., vmax=1., title=None, xlabel=None, ylabel=None, ylabel_right=True,
ylabel_rot=90, cax_label=r'Bott index, $B$',
cbar_labelpad=-30, cbar_orientation='horizontal',
ticks=[-1, 0, 1], fontsize=8, title_axX=None, title_axY=None, alpha=1.0)
return fig, ax
def plot_gapbounds(abv, gapbounds, outname, lp):
"""Plot the gap closing as a function of Delta_AB for magnetic system.
Axis parameters are fixed so that output can be made into a movie
Parameters
----------
abv : n x 1 float array
The Delta_AB values for each spectrum
gapbounds: n x 2(#bands) float array
The bottom and top of each band for each value of abv
"""
plt.close('all')
fig, ax = leplt.initialize_1panel_centered_fig(wsfrac=0.5, tspace=2)
# ax.plot(abv, gapbounds[:, 0], '.-', color=lecmaps.violet())
# ax.plot(abv, gapbounds[:, 1], '.-', color=lecmaps.green())
nbands = int(0.5 * np.shape(gapbounds)[1])
colors = [
lecmaps.green(),
lecmaps.violet(),
lecmaps.orange(),
lecmaps.blue(),
lecmaps.red(), '#000000'
]
for ii in range(nbands):
ax.fill_between(abv,
gapbounds[:, 1 + 2 * ii],
gapbounds[:, 0 + 2 * ii],
color=colors[ii])
ax.text(0.5,
1.12,
'Bands for magnetic gyros (' + r'$a/\ell=$' +
'{0:0.2f}'.format(lp['aoverl']) + ', '
r'$\Omega_k/\Omega_g=$' +
'{0:0.3f}'.format(lp['Omk'] / lp['Omg']) + ')',
ha='center',
va='center',
transform=ax.transAxes)
ax.set_xlabel(r'Inversion symmetry breaking $\Delta_{AB}$')
ax.set_ylabel(r'Frequency, $\omega$')
ax.set_ylim(0, 3.5)
print 'saving figure: ', outname
plt.savefig(outname)
plt.close('all')
def compare_dispersion_to_dos(omegas, kx, ky, mlat, outdir=None):
"""Compare the projection of the dispersion onto the omega axis with the DOS of the MagneticGyroLattice
Parameters
----------
omegas
kx
ky
mlat
outdir
Returns
-------
"""
# Save DOS from projection
if outdir is None:
outdir = dio.prepdir(mlat.lp['meshfn'])
else:
outdir = dio.prepdir(outdir)
name = outdir + 'dispersion_gyro' + mlat.lp['meshfn_exten'] + '_nx' + str(len(kx)) + '_ny' + str(len(ky))
name += '_maxkx{0:0.3f}'.format(np.max(np.abs(kx))).replace('.', 'p')
name += '_maxky{0:0.3f}'.format(np.max(np.abs(ky))).replace('.', 'p')
# initialize figure
fig, ax = leplt.initialize_1panel_centered_fig()
ax2 = ax.twinx()
ax.hist(omegas.ravel(), bins=1000)
# Compare the histograms of omegas to the dos and save the figure
eigval = np.imag(mlat.get_eigval())
print 'eigval = ', eigval
ax2.hist(eigval[eigval > 0], bins=50, color=lecmap.green(), alpha=0.2)
ax.set_title('DOS from dispersion')
xlims = ax.get_xlim()
ax.set_xlim(0, xlims[1])
plt.savefig(name + '_dos.png', dpi=300)
def infinite_dispersion_matk(matk,
lat,
kx=None,
ky=None,
nkxvals=50,
nkyvals=20,
save=True,
save_plot=True,
title='matrix dispersion relation',
outdir=None,
name=None,
ax=None,
lwscale=1.,
verbose=False):
"""
Parameters
----------
glat
kx
ky
nkxvals
nkyvals
save
save_plot
title
outdir
name
ax
lwscale
verbose
Returns
-------
"""
name, kx, ky = prepare_generalized_dispersion_params(lat,
kx=kx,
ky=ky,
nkxvals=nkxvals,
nkyvals=nkyvals,
outdir=outdir,
name=name)
print('checking for file: ' + name + '.pkl')
if glob.glob(name + '.pkl'):
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
omegas = np.zeros((len(kx), len(ky), len(matk([0, 0]))))
ii = 0
for kxi in kx:
if ii % 50 == 0:
print 'glatkspace_fns: infinite_dispersion(): ii = ', ii
jj = 0
for kyj in ky:
# print 'jj = ', jj
matrix = matk([kxi, kyj])
# print 'glatkspace_fns: diagonalizing...'
eigval, eigvect = np.linalg.eig(matrix)
si = np.argsort(np.imag(eigval))
omegas[ii, jj, :] = np.imag(eigval[si])
jj += 1
ii += 1
if save_plot or ax is not None:
if ax is None:
fig, ax = leplt.initialize_1panel_centered_fig(Hfig=90, wsfrac=0.6)
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx,
omegas[:, jj, kk],
'k-',
lw=lwscale * max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k$ $[\langle \ell \rangle ^{-1}]$')
ax.set_ylabel(r'$\omega$')
ylims = ax.get_ylim()
ax.set_ylim(0, ylims[1])
# Save the plot
if save_plot:
print 'saving ' + name + '.png'
plt.savefig(name + '.png', dpi=200)
plt.close('all')
if save:
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
return omegas, kx, ky
def plot_cherns_vary_param(self, param_type='glat', sz_param_nu=None,
reverse=False, param='percolation_density',
title='Chern index calculation', xlabel=None):
"""Plot chern indices with x axis being the parameter which is varying between networks.
If there are multiple chern calculations of a particular lattice, average nu over them all.
Parameters
----------
param_type : str ('glat' or 'lat')
Whether we are varying a lattice parameter (different lattices) or a gyrolattice parameter (same lattice,
different physics)
sz_param_nu : None or 'glat' or ('lat' or 'lp' -- last two do same thing)
string specifier for how to vary params
reverse : bool
Compute cherns for GyroLattice instances in self.mgyro_collection in reverse order
param : string
string specifier for GyroLattice parameter to vary between networks; key for mglat.lp dict
title : str
title of the plot
xlabel : str
xlabel, if desired to be other than default for param (ie, param.replace('_', ' '))
"""
if sz_param_nu is None:
if param_type == 'lat' or param_type == 'lp':
param_nu = self.collect_cherns_vary_lpparam(param=param, reverse=reverse)
elif param_type == 'glat':
param_nu = self.collect_cherns_vary_glatparam(param=param, reverse=reverse)
else:
raise RuntimeError("param_type argument passed is not 'glat' or 'lat/lp'")
# Plot it as colormap
plt.close('all')
paramV = param_nu[:, 0]
nu = param_nu[:, 1]
# Make figure
import lepm.plotting.plotting as leplt
fig, ax = leplt.initialize_1panel_centered_fig(wsfrac=0.5, tspace=4)
if xlabel is None:
xlabel = param.replace('_', ' ')
# Plot the curve
# first sort the paramV values
if isinstance(paramV[0], float):
si = np.argsort(paramV)
else:
si = np.arange(len(paramV), dtype=int)
ax.plot(paramV[si], nu[si], '.-')
ax.set_xlabel(xlabel)
# Add title
ax.text(0.5, 0.95, title, transform=fig.transFigure, ha='center', va='center')
# Save the plot
if param_type == 'glat':
outdir = rootdir + 'kspace_cherns_mgyro/chern_glatparam/' + param + '/'
dio.ensure_dir(outdir)
# Add meshfn name to the output filename
outbase = self.cherns[self.cherns.items()[0][0]][0].mgyro_lattice.lp['meshfn']
if outbase[-1] == '/':
outbase = outbase[:-1]
outbase = outbase.split('/')[-1]
outd_ex = self.cherns[self.cherns.items()[0][0]][0].mgyro_lattice.lp['meshfn_exten']
# If the parameter name is part of the meshfn_exten, replace its value with XXX in
# the meshfnexten part of outdir.
mfestr = glat_param2meshfnexten_name(param)
if mfestr in outd_ex:
'param is in meshfn_exten, splitting...'
# split the outdir by the param string
od_split = outd_ex.split(mfestr)
# split the second part by the value of the param string and the rest
od2val_rest = od_split[1].split('_')
odrest = od_split[1].split(od2val_rest[0])[1]
print 'odrest = ', odrest
print 'od2val_rest = ', od2val_rest
outd_ex = od_split[0] + param + 'XXX'
outd_ex += odrest
print 'outd_ex = ', outd_ex
else:
outd_ex += '_' + param + 'XXX'
elif param_type == 'lat':
outdir = rootdir + 'kspace_cherns_mgyro/chern_lpparam/' + param + '/'
outbase = self.cherns[self.cherns.items()[0][0]][0].mgyro_lattice.lp['meshfn']
# Take apart outbase to parse out the parameter that is varying
mfestr = lat_param2meshfn_name(param)
if mfestr in outbase :
'param is in meshfn_exten, splitting...'
# split the outdir by the param string
od_split = outbase.split(mfestr)
# split the second part by the value of the param string and the rest
od2val_rest = od_split[1].split('_')
odrest = od_split[1].split(od2val_rest[0])[1]
print 'odrest = ', odrest
print 'od2val_rest = ', od2val_rest
outd_ex = od_split[0] + param + 'XXX'
outd_ex += odrest
print 'outd_ex = ', outd_ex
else:
outbase += '_' + param + 'XXX'
outd_ex = self.cherns[self.cherns.items()[0][0]][0].mgyro_lattice.lp['meshfn_exten']
dio.ensure_dir(outdir)
fname = outdir + outbase
fname += '_chern_' + param + '_Ncoll' + '{0:03d}'.format(len(self.mgyro_collection.mgyro_lattices))
fname += outd_ex
print 'saving to ' + fname + '.png'
plt.savefig(fname + '.png')
plt.clf()
def infinite_dispersion_unstructured(glat,
kxy,
load=False,
save=True,
save_plot=True,
title='gyro dispersion relation',
outdir=None,
name=None,
ax=None,
lwscale=1.,
verbose=False,
overwrite=False,
return_eigvects=False):
""" Do not assume grid structure of kxy for this function. Compute the spectrum evaluated at the points given
by the 2d array kxy.
Parameters
----------
glat
kxy
save
save_plot
title
outdir
name
ax
lwscale
verbose
Returns
-------
omegas : tuple (omegas, eigvects) if return_eigvects is True, otherwise len(kxy) x 2 * len(xy) float array
The eigenvalues, and possibly the corresponding eigenvectors if kwarg return_eigvects is True, returned as
float and complex arrays, respectively. omegas is a len(kxy) x 2 * len(xy) float array, while eigvects, if
returned, is a len(kxy) x 2 * len(xy) x 2 * len(xy) complex array
"""
name, kx, ky = prepare_dispersion_params(glat,
kx=kxy[:, 0],
ky=kxy[:, 1],
outdir=outdir,
name=name)
name += '_unstructured'
if glob.glob(name + '.pkl') and not overwrite and load:
print('checking for file: ' + name + '.pkl')
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
omegas = np.zeros((len(kxy), len(glat.lattice.xy) * 2))
if return_eigvects:
eigvects = np.zeros(
(len(kxy), len(glat.lattice.xy) * 2, len(glat.lattice.xy) * 2),
dtype=complex)
matk = lambda k: dynamical_matrix_kspace(k, glat, eps=1e-10)
ii = 0
for pt in kxy:
if ii % 50 == 1:
print 'glatkspace_fns: infinite_dispersion(): ii = ', ii
matrix = matk([pt[0], pt[1]])
eigval, eigvect = np.linalg.eig(matrix)
si = np.argsort(np.imag(eigval))
omegas[ii, :] = np.imag(eigval[si])
if return_eigvects:
eigvect = np.array(eigvect)
eigvect_out = eigvect.T[si]
# print 'eigvect_out = ', eigvect_out
# if ortho_eigvect:
# eigvect_out = gdh.orthonormal_eigvect(eigvect)
# print 'eigvect_out = ', eigvect_out
eigvects[ii] = eigvect_out
ii += 1
if save_plot or ax is not None:
if ax is None:
fig, ax = leplt.initialize_1panel_centered_fig()
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx,
omegas[:, jj, kk],
'k-',
lw=lwscale * max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k$ $[\langle \ell \rangle ^{-1}]$')
ax.set_ylabel(r'$\omega$')
ylims = ax.get_ylim()
ax.set_ylim(0, ylims[1])
# Save the plot
if save_plot:
plt.savefig(name + '.png', dpi=300)
plt.close('all')
if save:
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
if return_eigvects:
omegas = (omegas, eigvects)
return omegas, kx, ky
def infinite_dispersion(glat,
kx=None,
ky=None,
nkxvals=50,
nkyvals=20,
save=True,
save_plot=True,
title='gyro dispersion relation',
outdir=None,
name=None,
ax=None,
lwscale=1.,
verbose=False):
"""Compute the imaginary part of the eigvalues of the dynamical matrix for a grid (or unstructued set)
of wavevectors kx, ky.
See also calc_bands() and calc_band_gaps()
Parameters
----------
glat : GyroLattice class instance
the gyro network whose dispersion we compute
kx : n x 1 float array
the x components of the wavevectors over which to diagonalize the dynamical matrix
ky : m x 1 float array
the y components of the wavevectors over which to diagonalize the dynamical matrix
nkxvals : int
If kx is unspecified, then nkxvals determines how many kvectors are sampled in x dimension.
nkyvals : int
If ky is unspecified and if network is not a periodic_strip, then nkyvals determines how
many kvectors are sampled in y dimension.
save : bool
Save the omega vs k information in a pickle
save_plot : bool
Save the omega vs k info as a matplotlib figure png
title : str
title for the plot to save
outdir : str or None
The directory in which to output the image and pickle of the results if save == True
Returns
-------
"""
name, kx, ky = prepare_dispersion_params(glat,
kx=kx,
ky=ky,
nkxvals=nkxvals,
nkyvals=nkyvals,
outdir=outdir,
name=name)
print('checking for file: ' + name + '.pkl')
if glob.glob(name + '.pkl'):
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
omegas = np.zeros((len(kx), len(ky), len(glat.lattice.xy) * 2))
matk = lambda k: dynamical_matrix_kspace(k, glat, eps=1e-10)
ii = 0
for kxi in kx:
if ii % 50 == 0:
print 'glatkspace_fns: infinite_dispersion(): ii = ', ii
jj = 0
for kyj in ky:
# print 'jj = ', jj
matrix = matk([kxi, kyj])
# print 'glatkspace_fns: diagonalizing...'
eigval, eigvect = np.linalg.eig(matrix)
si = np.argsort(np.imag(eigval))
omegas[ii, jj, :] = np.imag(eigval[si])
# print 'eigvals = ', eigval
# print 'omegas --> ', omegas[ii, jj]
jj += 1
ii += 1
if save_plot or ax is not None:
if ax is None:
fig, ax = leplt.initialize_1panel_centered_fig()
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx,
omegas[:, jj, kk],
'k-',
lw=lwscale * max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k$ $[\langle \ell \rangle ^{-1}]$')
ax.set_ylabel(r'$\omega$')
ylims = ax.get_ylim()
ax.set_ylim(0, ylims[1])
# Save the plot
if save_plot:
print 'saving ' + name + '.png'
plt.savefig(name + '.png', dpi=300)
plt.close('all')
# Plot in 3D
# fig = plt.gcf()
# ax = fig.add_subplot(projection='3d') # 111,
# # rows will be kx, cols wll be ky
# kyv = np.array([[ky[i].tolist()] * len(kx) for i in range(len(ky))]).ravel()
# kxv = np.array([[kx.tolist()] * len(ky)]).ravel()
# print 'kyv = ', np.shape(kyv)
# print 'kxv = ', np.shape(kxv)
# for kk in range(len(omegas[0, 0, :])):
# ax.plot_trisurf(kxv, kyv, omegas[:, :, kk].ravel())
#
# ax.view_init(elev=0, azim=0.)
# ax.set_title(title)
# ax.set_xlabel(r'$k_x$ $[1/\langle l \rangle]$')
# ax.set_ylabel(r'$k_y$ $[1/\langle l \rangle]$')
# plt.savefig(name + '_3d.png')
if save:
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
return omegas, kx, ky
result[kk, 0] = ntest
result[kk, 1] = bott
results[nsize] = result
# from scipy.linalg import logm
# print 'total bott = ', np.trace(logm(np.dot(vv, np.dot(uu, np.dot(vv.conj().T, uu.conj().T)))))
# bott = np.imag(np.trace(logm(np.dot(vv, np.dot(uu, np.dot(vv.conj().T, uu.conj().T))))))
######################
kk += 1
plt.close('all')
# print 'result = ', result
# sys.exit()
fig, ax = leplt.initialize_1panel_centered_fig()
markers = ['o', '.', '^']
lines = ['-', '--', '-.']
kk = 0
for nsize in results:
ax.plot(results[nsize][:, 0],
results[nsize][:, 1],
marker=markers[kk],
linestyle=lines[kk],
markersize=1,
markeredgecolor=None,
label=str(nsize))
kk += 1
ax.legend(loc='lower right')
# check system size
def infinite_dispersion(mglat, kx=None, ky=None, nkxvals=50, nkyvals=20, save=True, save_plot=True,
title='Dispersion relation', outdir=None):
"""
Parameters
----------
mglat :
kx :
ky :
save :
title :
outdir :
Returns
-------
omegas, kx, ky
"""
if not mglat.lp['periodicBC']:
raise RuntimeError('Cannot compute dispersion for open BC system')
elif mglat.lp['periodic_strip']:
print 'Evaluating infinite-system dispersion for strip: setting ky=constant=0.'
ky = [0.]
if ky is None or kx is None:
bboxx = max(mglat.lattice.lp['BBox'][:, 0]) - min(mglat.lattice.lp['BBox'][:, 0])
bboxy = max(mglat.lattice.lp['BBox'][:, 1]) - min(mglat.lattice.lp['BBox'][:, 1])
if kx is None:
tmp = np.linspace(-1. / bboxx, 1. / bboxx, nkxvals - 1)
step = np.diff(tmp)[0]
kx = 2. * np.pi * np.linspace(-1. / bboxx, 1. / bboxx + step, nkxvals)
# kx = np.linspace(-5., 5., 40)
if ky is None:
tmp = np.linspace(-1. / bboxy, 1. / bboxy, nkyvals - 1)
step = np.diff(tmp)[0]
ky = 2. * np.pi * np.linspace(-1. / bboxy, 1. / bboxy + step, nkyvals)
# ky = np.linspace(-5., 5., 4)
# First check for saved dispersion
if outdir is None:
outdir = dio.prepdir(mglat.lp['meshfn'])
else:
outdir = dio.prepdir(outdir)
name = outdir + 'dispersion' + mglat.lp['meshfn_exten'] + '_nx' + str(len(kx)) + '_ny' + str(len(ky))
name += '_maxkx{0:0.3f}'.format(np.max(np.abs(kx))).replace('.', 'p')
name += '_maxky{0:0.3f}'.format(np.max(np.abs(ky))).replace('.', 'p')
print('checking for file: ' + name + '.pkl')
if glob.glob(name + '.pkl'):
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
# Use PVx and PVy to multiply exp(i*np.dot(k, PV[0,:])) to periodic vectors in x, similar in y
omegas = np.zeros((len(kx), len(ky), len(mglat.lattice.xy) * 2))
matk = lambda k: dynamical_matrix_kspace(k, mglat, eps=1e-10)
ii = 0
for kxi in kx:
if ii % 25 == 0:
print 'infinite_dispersion(): ii = ', ii
jj = 0
for kyj in ky:
# print 'jj = ', jj
matrix = matk([kxi, kyj])
eigval, eigvect = np.linalg.eig(matrix)
si = np.argsort(np.imag(eigval))
omegas[ii, jj, :] = np.imag(eigval[si])
# print 'omegas = ', omegas
jj += 1
ii += 1
if save_plot:
fig, ax = leplt.initialize_1panel_centered_fig()
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx, omegas[:, jj, kk], 'k-', lw=max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k$ $[1/\langle l \rangle]$')
ax.set_ylabel(r'$\omega$')
print 'magnetic_gyro_kspace_functions: saving image to ' + name + '.png'
plt.savefig(name + '.png')
plt.close('all')
# Plot in 3D
# fig = plt.gcf()
# ax = fig.add_subplot(projection='3d') # 111,
# # rows will be kx, cols wll be ky
# kyv = np.array([[ky[i].tolist()] * len(kx) for i in range(len(ky))]).ravel()
# kxv = np.array([[kx.tolist()] * len(ky)]).ravel()
# print 'kyv = ', np.shape(kyv)
# print 'kxv = ', np.shape(kxv)
# for kk in range(len(omegas[0, 0, :])):
# ax.plot_trisurf(kxv, kyv, omegas[:, :, kk].ravel())
#
# ax.view_init(elev=0, azim=0.)
# ax.set_title(title)
# ax.set_xlabel(r'$k_x$ $[1/\langle l \rangle]$')
# ax.set_ylabel(r'$k_y$ $[1/\langle l \rangle]$')
# plt.savefig(name + '_3d.png')
if save:
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
return omegas, kx, ky
def plot_cherns_vary_param(kcgcoll,
param_type='glat',
sz_param_nu=None,
reverse=False,
param='percolation_density',
title='Chern index calculation',
xlabel=None):
"""
Parameters
----------
kcgcoll :
param_type :
sz_param_nu :
reverse :
param :
title :
xlabel :
Returns
-------
"""
if sz_param_nu is None:
if param_type == 'lat' or param_type == 'lp':
param_nu = self.collect_cherns_vary_lpparam(param=param,
reverse=reverse)
elif param_type == 'glat':
param_nu = self.collect_cherns_vary_glatparam(param=param,
reverse=reverse)
else:
raise RuntimeError(
"param_type argument passed is not 'glat' or 'lat/lp'")
# Plot it as colormap
plt.close('all')
paramV = param_nu[:, 0]
nu = param_nu[:, 1]
# Make figure
import lepm.plotting.plotting as leplt
fig, ax = leplt.initialize_1panel_centered_fig(wsfrac=0.5, tspace=4)
if xlabel is None:
xlabel = param.replace('_', ' ')
# Plot the curve
# first sort the paramV values
if isinstance(paramV[0], float):
si = np.argsort(paramV)
else:
si = np.arange(len(paramV), dtype=int)
ax.plot(paramV[si], nu[si], '.-')
ax.set_xlabel(xlabel)
# Add title
ax.text(0.5,
0.95,
title,
transform=fig.transFigure,
ha='center',
va='center')
# Save the plot
if param_type == 'glat':
outdir = rootdir + 'kspace_cherns_gyro/chern_glatparam/' + param + '/'
dio.ensure_dir(outdir)
# Add meshfn name to the output filename
outbase = self.cherns[self.cherns.items()[0]
[0]][0].gyro_lattice.lp['meshfn']
if outbase[-1] == '/':
outbase = outbase[:-1]
outbase = outbase.split('/')[-1]
outd_ex = self.cherns[self.cherns.items()[0]
[0]][0].gyro_lattice.lp['meshfn_exten']
# If the parameter name is part of the meshfn_exten, replace its value with XXX in
# the meshfnexten part of outdir.
mfestr = glatfns.param2meshfnexten_name(param)
if mfestr in outd_ex:
'param is in meshfn_exten, splitting...'
# split the outdir by the param string
od_split = outd_ex.split(mfestr)
# split the second part by the value of the param string and the rest
od2val_rest = od_split[1].split('_')
odrest = od_split[1].split(od2val_rest[0])[1]
print 'odrest = ', odrest
print 'od2val_rest = ', od2val_rest
outd_ex = od_split[0] + param + 'XXX'
outd_ex += odrest
print 'outd_ex = ', outd_ex
else:
outd_ex += '_' + param + 'XXX'
elif param_type == 'lat':
outdir = rootdir + 'kspace_cherns_gyro/chern_lpparam/' + param + '/'
outbase = self.cherns[self.cherns.items()[0]
[0]][0].gyro_lattice.lp['meshfn']
# Take apart outbase to parse out the parameter that is varying
mfestr = latfns.param2meshfnexten_name(param)
if mfestr in outbase:
'param is in meshfn_exten, splitting...'
# split the outdir by the param string
od_split = outbase.split(mfestr)
# split the second part by the value of the param string and the rest
od2val_rest = od_split[1].split('_')
odrest = od_split[1].split(od2val_rest[0])[1]
print 'odrest = ', odrest
print 'od2val_rest = ', od2val_rest
outd_ex = od_split[0] + param + 'XXX'
outd_ex += odrest
print 'outd_ex = ', outd_ex
else:
outbase += '_' + param + 'XXX'
outd_ex = self.cherns[self.cherns.items()[0]
[0]][0].gyro_lattice.lp['meshfn_exten']
dio.ensure_dir(outdir)
fname = outdir + outbase
fname += '_chern_' + param + '_Ncoll' + '{0:03d}'.format(
len(self.gyro_collection.gyro_lattices))
fname += outd_ex
print 'saving to ' + fname + '.png'
plt.savefig(fname + '.png')
plt.clf()
def infinite_dispersion(mlat, kx=None, ky=None, nkxvals=50, nkyvals=20, save=True, save_plot=True,
title='twisty dispersion relation', outdir=None, name=None, ax=None):
"""Compute the imaginary part of the eigvalues of the dynamical matrix for a grid of wavevectors kx, ky
Parameters
----------
mlat : MassLattice class instance
the twisty network whose dispersion we compute
kx : n x 1 float array
the x components of the wavevectors over which to diagonalize the dynamical matrix
ky : m x 1 float array
the y components of the wavevectors over which to diagonalize the dynamical matrix
nkxvals : int
If kx is unspecified, then nkxvals determines how many kvectors are sampled in x dimension.
nkyvals : int
If ky is unspecified and if network is not a periodic_strip, then nkyvals determines how
many kvectors are sampled in y dimension.
save : bool
Save the omega vs k information in a pickle
save_plot : bool
Save the omega vs k info as a matplotlib figure png
title : str
title for the plot to save
outdir : str or None
The directory in which to output the image and pickle of the results if save == True
Returns
-------
"""
if not mlat.lp['periodicBC']:
raise RuntimeError('Cannot compute dispersion for open BC system')
elif mlat.lp['periodic_strip']:
print 'Evaluating infinite-system dispersion for strip: setting ky=constant=0.'
ky = [0.]
bboxx = max(mlat.lattice.lp['BBox'][:, 0]) - min(mlat.lattice.lp['BBox'][:, 0])
elif ky is None or kx is None:
bboxx = max(mlat.lattice.lp['BBox'][:, 0]) - min(mlat.lattice.lp['BBox'][:, 0])
bboxy = max(mlat.lattice.lp['BBox'][:, 1]) - min(mlat.lattice.lp['BBox'][:, 1])
if kx is None:
if nkxvals == 0:
kx = np.array([0.])
else:
tmp = np.linspace(-1. / bboxx, 1. / bboxx, nkxvals - 1)
step = np.diff(tmp)[0]
kx = 2. * np.pi * np.linspace(-1. / bboxx, 1. / bboxx + step, nkxvals)
# kx = np.linspace(-5., 5., 40)
if ky is None:
if nkyvals == 0:
ky = np.array([0.])
else:
tmp = np.linspace(-1. / bboxy, 1. / bboxy, nkyvals - 1)
step = np.diff(tmp)[0]
ky = 2. * np.pi * np.linspace(-1. / bboxy, 1. / bboxy + step, nkyvals)
# ky = np.linspace(-5., 5., 4)
# First check for saved dispersion
if outdir is None:
outdir = dio.prepdir(mlat.lp['meshfn'])
else:
outdir = dio.prepdir(outdir)
if name is None:
name = 'dispersion' + mlat.lp['meshfn_exten'] + '_nx' + str(len(kx)) + '_ny' + str(len(ky))
name += '_maxkx{0:0.3f}'.format(np.max(np.abs(kx))).replace('.', 'p')
name += '_maxky{0:0.3f}'.format(np.max(np.abs(ky))).replace('.', 'p')
name = outdir + name
print('checking for file: ' + name + '.pkl')
if glob.glob(name + '.pkl'):
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
omegas = np.zeros((len(kx), len(ky), len(mlat.lattice.xy) * 2))
matk = lambda k: dynamical_matrix_kspace(k, mlat, eps=1e-10)
ii = 0
for kxi in kx:
print 'mlatkspace_fns: infinite_dispersion(): ii = ', ii
jj = 0
for kyj in ky:
# print 'jj = ', jj
matrix = matk([kxi, kyj])
print 'mlatkspace_fns: diagonalizing...'
eigval, eigvect = np.linalg.eig(matrix)
si = np.argsort(np.real(eigval))
omegas[ii, jj, :] = np.real(np.sqrt(-eigval[si]))
# print 'eigvals = ', eigval
# print 'omegas --> ', omegas[ii, jj]
jj += 1
ii += 1
if save_plot:
if ax is None:
fig, ax = leplt.initialize_1panel_centered_fig()
axsupplied = False
else:
axsupplied = True
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx, omegas[:, jj, kk], 'k-', lw=max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k_x$ $[\langle \ell \rangle ^{-1}]$')
ax.set_ylabel(r'$\omega$')
ylims = ax.get_ylim()
ylim0 = min(ylims[0], -0.1 * ylims[1])
ax.set_ylim(ylim0, ylims[1])
# Save the plot
plt.savefig(name + '.png', dpi=300)
ax.set_ylim(max(ylim0, -0.05 * ylims[1]), 0.05 * ylims[1])
# Save the plot
plt.savefig(name + '_zoom.png', dpi=300)
# Fixed zoom
ax.set_ylim(-0.3, 0.6)
plt.savefig(name + '_zoom2.png', dpi=300)
plt.close('all')
# save plot of ky if no axis supplied
if not axsupplied:
fig, ax = leplt.initialize_1panel_centered_fig()
for jj in range(len(kx)):
for kk in range(len(omegas[jj, 0, :])):
ax.plot(ky, omegas[jj, :, kk], 'k-', lw=max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k_y$ $[\langle \ell \rangle ^{-1}]$')
ax.set_ylabel(r'$\omega$')
ylims = ax.get_ylim()
ylim0 = min(ylims[0], -0.1 * ylims[1])
ax.set_ylim(ylim0, ylims[1])
# Save the plot
plt.savefig(name + '_ky.png', dpi=300)
ax.set_ylim(max(ylim0, -0.05 * ylims[1]), 0.05 * ylims[1])
# Save the plot
plt.savefig(name + '_zoom_ky.png', dpi=300)
# Fixed zoom
ax.set_ylim(-0.3, 0.6)
plt.savefig(name + '_zoom2_ky.png', dpi=300)
plt.close('all')
# Plot in 3D
# fig = plt.gcf()
# ax = fig.add_subplot(projection='3d') # 111,
# # rows will be kx, cols wll be ky
# kyv = np.array([[ky[i].tolist()] * len(kx) for i in range(len(ky))]).ravel()
# kxv = np.array([[kx.tolist()] * len(ky)]).ravel()
# print 'kyv = ', np.shape(kyv)
# print 'kxv = ', np.shape(kxv)
# for kk in range(len(omegas[0, 0, :])):
# ax.plot_trisurf(kxv, kyv, omegas[:, :, kk].ravel())
#
# ax.view_init(elev=0, azim=0.)
# ax.set_title(title)
# ax.set_xlabel(r'$k_x$ $[1/\langle l \rangle]$')
# ax.set_ylabel(r'$k_y$ $[1/\langle l \rangle]$')
# plt.savefig(name + '_3d.png')
if save:
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
return omegas, kx, ky
def infinite_dispersion(hlat,
kx=None,
ky=None,
save=True,
title='Dispersion relation',
outdir=None):
"""Compute energy versus wavenumber for a grid of kx ky values for a haldane model network.
If the network is a periodic strip, then ky is set to zero and we look at E(kx).
Parameters
----------
hlat : HaldaneLattice instance
the tight-binding network over which to compute the dispersion relation
kx : N x 1 float array
The x component of the wavenumbers to evaluate
ky : M x 1 float array
The y component of the wavenumbers to evaluate
save : bool
Whether to save the dispersion to disk
title : str
title of the plot to make if save is True
outdir : str
path to the place to save the plot if save is True
Returns
-------
"""
if not hlat.lp['periodicBC']:
raise RuntimeError('Cannot compute dispersion for open BC system')
elif hlat.lp['periodic_strip']:
print 'Evaluating infinite-system dispersion for strip: setting ky=constant=0.'
ky = [0.]
bboxx = max(hlat.lattice.lp['BBox'][:, 0]) - min(
hlat.lattice.lp['BBox'][:, 0])
print 'bboxx = ', bboxx
elif ky is None or kx is None:
bboxx = max(hlat.lattice.lp['BBox'][:, 0]) - min(
hlat.lattice.lp['BBox'][:, 0])
bboxy = max(hlat.lattice.lp['BBox'][:, 1]) - min(
hlat.lattice.lp['BBox'][:, 1])
if kx is None:
kx = np.linspace(-2. * np.pi / bboxx, 2. * np.pi / bboxx, 50)
# kx = np.linspace(-5., 5., 40)
if ky is None:
ky = np.linspace(-2. * np.pi / bboxy, 2. * np.pi / bboxy, 5)
# ky = np.linspace(-5., 5., 4)
# First check for saved dispersion
if outdir is None:
outdir = dio.prepdir(hlat.lp['meshfn'])
else:
outdir = dio.prepdir(outdir)
name = outdir + 'dispersion' + hlat.lp['meshfn_exten'] + '_nx' + str(
len(kx)) + '_ny' + str(len(ky))
name += '_maxkx{0:0.3f}'.format(np.max(np.abs(kx))).replace('.', 'p')
name += '_maxky{0:0.3f}'.format(np.max(np.abs(ky))).replace('.', 'p')
print('checking for file: ' + name + '.pkl')
if glob.glob(name + '.pkl'):
saved = True
with open(name + '.pkl', "rb") as fn:
res = pkl.load(fn)
omegas = res['omegas']
kx = res['kx']
ky = res['ky']
else:
# dispersion is not saved, compute it!
saved = False
# Use PVx and PVy to multiply exp(i*np.dot(k, PV[0,:])) to periodic vectors in x, similar in y
# First convert PVx and PVy to matrices that are the same shape as hlat.matrix
# np.exp()
omegas = np.zeros((len(kx), len(ky), len(hlat.lattice.xy)))
matk = lambda k: dynamical_matrix_kspace(k, hlat, eps=1e-8)
timer = Timer()
ii = 0
for kxi in kx:
print 'infinite_dispersion(): ii = ', ii
jj = 0
for kyj in ky:
# print 'jj = ', jj
timer.restart()
matrix = matk([kxi, kyj])
test = timer.get_time_ms()
print 'test = ', test
print('constructed matrix in: ' + timer.get_time_ms())
print 'diagonalizing...'
eigval, eigvect = np.linalg.eig(matrix)
timer.restart()
print('diagonalized matrix in: ' + timer.get_time_ms())
si = np.argsort(np.real(eigval))
omegas[ii, jj, :] = eigval[si]
# print 'omegas = ', omegas
jj += 1
ii += 1
if save:
fig, ax = leplt.initialize_1panel_centered_fig()
for jj in range(len(ky)):
for kk in range(len(omegas[0, jj, :])):
ax.plot(kx,
omegas[:, jj, kk],
'k-',
lw=max(0.03, 5. / (len(kx) * len(ky))))
ax.set_title(title)
ax.set_xlabel(r'$k$ $[1/\langle \ell \rangle]$')
ax.set_ylabel(r'$\omega$')
plt.savefig(name + '.png')
plt.close('all')
# Plot in 3D
# fig = plt.gcf()
# ax = fig.add_subplot(projection='3d') # 111,
# # rows will be kx, cols wll be ky
# kyv = np.array([[ky[i].tolist()] * len(kx) for i in range(len(ky))]).ravel()
# kxv = np.array([[kx.tolist()] * len(ky)]).ravel()
# print 'kyv = ', np.shape(kyv)
# print 'kxv = ', np.shape(kxv)
# for kk in range(len(omegas[0, 0, :])):
# ax.plot_trisurf(kxv, kyv, omegas[:, :, kk].ravel())
#
# ax.view_init(elev=0, azim=0.)
# ax.set_title(title)
# ax.set_xlabel(r'$k_x$ $[1/\langle l \rangle]$')
# ax.set_ylabel(r'$k_y$ $[1/\langle l \rangle]$')
# plt.savefig(name + '_3d.png')
if not saved:
res = {'omegas': omegas, 'kx': kx, 'ky': ky}
with open(name + '.pkl', "wb") as fn:
pkl.dump(res, fn)
return omegas, kx, ky
def movie_varyDeltaAB(t1,
t2,
dab_arr,
nx,
ny,
maindir,
outdir,
fontsize=20,
tick_fontsize=16,
saveims=True,
dual_panel=False,
color_berry=False,
cmap='coolwarm',
vmin=0.49,
vmax=0.51,
plot3d=False):
"""Create a movie of band structure with varying DeltaAB
Parameters
----------
t1 : float
magnitude of real NN hopping
t2 : float
magnitude of complex NNN hopping
dab_arr : n x 1 float array
the DeltaAB values to use for each frame
nn : int
number of kx values (same as # ky values)
maindir : str
where to save the movie
outdir : str
where to save the images
saveims : bool
overwrite the existing images
dual_panel : bool
plot the 2d dispersion with calibration of gap opening as function of DeltaAB
vmin : float
the minimimum color value from 0 to 1, if color_berry=True
vmax : float
the maximum color value from 0 to 1, if color_berry = True
plot3d : bool
whether to plot the dispersion in 3d or not
"""
if isinstance(cmap, str):
cmap = lecmaps.ensure_cmap(cmap)
kk = 0
gaps = []
dabs = []
for mm in dab_arr:
if color_berry:
km, kv, energy1, energy2, berry1, berry2 = haldane_dispersion_berry(
nx, ny, t1, t2, mm)
else:
km, kv, energy1, energy2 = haldane_dispersion(nx, ny, t1, t2, mm)
gap = np.min(energy1) * 2.
print 'gap = ', gap
gaps.append(gap)
dabs.append(mm)
if saveims:
if not plot3d:
if dual_panel:
fig, ax = leplt.initialize_2panel_3o4ar_cent(
fontsize=fontsize, x0frac=0.085)
ax, ax2 = ax[0], ax[1]
# Make second figure
ax2.plot()
ax2.set_xlim(0, 0.2)
ax2.set_ylim(0, 0.7)
ax2.plot([np.min(t2arr), np.max(t2arr)],
[0.0, 2 * np.max(t2arr)],
'-',
color='#a1bfdb',
lw=3)
ax2.scatter(dabs, gaps, color='k', zorder=9999999999)
title = 'Gap size: ' + r'$\Delta\omega/t_1 \approx 2 \, \Delta_{AB}$'
ax2.text(0.5,
1.1,
title,
ha='center',
va='center',
fontsize=fontsize,
transform=ax2.transAxes)
for tick in ax2.xaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
for tick in ax2.yaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax2.set_xlabel(r'$t_2 = \Omega_k^2/8\Omega_g$',
fontsize=fontsize,
labelpad=15)
ax2.set_ylabel(
r'$\Delta\omega/t_1 = 2 \Delta\omega/\Omega_k$',
fontsize=fontsize,
rotation=90,
labelpad=35)
else:
fig, ax = leplt.initialize_1panel_centered_fig(
Wfig=180, Hfig=180, wsfrac=0.7, fontsize=fontsize)
if not color_berry:
# check if topological or not
print 't2 = ', t2
print 'mm = ', mm * (3. * np.sqrt(3.))
if t2 > mm / np.sqrt(3.):
# yes, topological
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=band1color)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=band2color)
elif t2 < -mm / np.sqrt(3.):
# flipped topology
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=band2color)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=band1color)
else:
# not topological
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=graycolor)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=graycolor)
else:
b1color = berry1.reshape(np.shape(
km[0])) / (2. * np.pi) + 0.5
b2color = berry2.reshape(np.shape(
km[0])) / (2. * np.pi) + 0.5
energy1 = energy1.reshape(np.shape(km[0]))
energy2 = energy2.reshape(np.shape(km[0]))
# ax.scatter([0., 0.5, 1.0], [0., 0.5, 1.0], c=[0., 0.5, 1.0], edgecolor='none', cmap=cmap, alpha=1)
for ii in range(len(km[0])):
ax.scatter(km[1][ii],
energy1[ii],
c=b1color[ii],
edgecolor='none',
cmap=cmap,
alpha=0.1,
vmin=vmin,
vmax=vmax)
ax.scatter(km[1][ii],
energy2[ii],
c=b2color[ii],
edgecolor='none',
cmap=cmap,
alpha=0.1,
vmin=vmin,
vmax=vmax)
plt.show()
sys.exit()
ax.set_ylabel(r'$\omega$',
labelpad=15,
fontsize=fontsize,
rotation=0)
# title = r'$\Delta \omega =$' + '{0:0.3f}'.format(gap) + r' $t_1$'
title = r'$t_2 =$' + '{0:0.3f}'.format(t2) + r' $t_1$, ' + \
r'$\Delta_{AB} =$' + '{0:0.3f}'.format(mm) + r' $t_1$'
ax.text(0.5,
1.1,
title,
ha='center',
va='center',
fontsize=fontsize,
transform=ax.transAxes)
else:
# Plot it in 3d
fig, ax = leplt.initialize_1panel_centered_fig(
Wfig=180, Hfig=180, wsfrac=0.7, fontsize=fontsize)
ax = fig.gca(projection='3d')
# Reshape colors, energies
b1color = cmap((berry1.reshape(np.shape(km[0])) /
(2. * np.pi)) / (vmax - vmin) + 0.5)
b2color = cmap((berry2.reshape(np.shape(km[0])) /
(2. * np.pi)) / (vmax - vmin) + 0.5)
energy1 = energy1.reshape(np.shape(km[0]))
energy2 = energy2.reshape(np.shape(km[0]))
print 'b1color = ', b1color
# Plot surfaces
surf = ax.plot_surface(km[0],
km[1],
energy1,
facecolors=b1color,
rstride=1,
cstride=1,
vmin=vmin,
vmax=vmax,
cmap=cmap,
linewidth=1,
antialiased=False)
surf = ax.plot_surface(km[0],
km[1],
energy2,
facecolors=b2color,
rstride=1,
cstride=1,
vmin=vmin,
vmax=vmax,
cmap=cmap,
linewidth=1,
antialiased=False)
ax.set_ylabel(r'$k_y$', labelpad=15, fontsize=fontsize)
ax.yaxis.set_ticks([-np.pi, 0, np.pi])
ax.yaxis.set_ticklabels([r'-$\pi$', 0, r'$\pi$'],
fontsize=tick_fontsize)
for tick in ax.xaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax.set_zlabel(r'$\omega$',
labelpad=15,
fontsize=fontsize,
rotation=90)
# ax.zaxis.set_ticks([-np.pi, 0, np.pi])
# ax.zaxis.set_ticklabels([r'-$\pi$', 0, r'$\pi$'], fontsize=tick_fontsize)
for tick in ax.zaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax.axis('scaled')
ax.set_zlim(-np.pi, np.pi)
ax.set_xlabel(r'$k_x$', labelpad=15, fontsize=fontsize)
ax.xaxis.set_ticks([-np.pi, 0, np.pi])
ax.xaxis.set_ticklabels([r'-$\pi$', 0, r'$\pi$'],
fontsize=tick_fontsize)
for tick in ax.yaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax.axis('scaled')
ax.set_xlim(-np.pi, np.pi)
ax.set_ylim(-np.pi, np.pi)
# Save it
plt.savefig(outdir + 'dispersion_{0:04d}'.format(kk) +
'.png'.format(t2),
dpi=140)
plt.close('all')
kk += 1
print 'dirac = ', 2. * np.pi / 3., ', ', -2. * np.pi / (
3. * np.sqrt(3.))
ind = np.argmin(energy1)
dirac = kv[ind]
print 'dirac where = ', dirac
# print the fitting
print 'dabs = ', dabs
print 'gaps = ', gaps
zz = np.polyfit(np.array(dabs), np.array(gaps), 1)
pp = np.poly1d(zz)
print 'pp = ', pp
print 'zz[0] = ', zz[0]
print 'zz[1] = ', zz[1]
imgname = outdir + 'dispersion_'
movname = maindir + 'dispersion_varyDeltaAB_{0:0.3f}'.format(t2).replace(
'.', 'p')
movname += '_nkx{0:06d}'.format(nx) + '_nky{0:06d}'.format(ny)
lemov.make_movie(imgname, movname, indexsz='04', framerate=15)
def movie_varyt2(t1,
t2arr,
mm,
nx,
ny,
maindir,
outdir,
fontsize=20,
tick_fontsize=16,
saveims=True,
dual_panel=False,
color_berry=False,
cmap='bbr0',
vmin=0.499,
vmax=0.501,
plot3d=False):
"""Plot the band structure of the haldane model as we vary the NNN hopping
Parameters
----------
t1 : float
magnitude of real NN hopping
t2 : float
magnitude of complex NNN hopping
dab_arr : n x 1 float array
the DeltaAB values to use for each frame
nn : int
number of kx values (same as # ky values)
maindir : str
where to save the movie
outdir : str
where to save the images
saveims : bool
overwrite the existing images
dual_panel : bool
plot the 2d dispersion with calibration of gap opening as function of DeltaAB
vmin : float
the minimimum color value from 0 to 1, if color_berry=True
vmax : float
the maximum color value from 0 to 1, if color_berry = True
plot3d : bool
whether to plot the dispersion in 3d or not
"""
kk = 0
gaps = []
t2s = []
if isinstance(cmap, str):
cmap = lecmaps.ensure_cmap(cmap)
for t2 in t2arr:
if color_berry:
km, kv, energy1, energy2, berry1, berry2 = haldane_dispersion_berry(
nx, t1, t2, mm)
else:
km, kv, energy1, energy2 = haldane_dispersion(nx, ny, t1, t2, mm)
gap = np.min(energy1) * 2.
print 'gap = ', gap
gaps.append(gap)
t2s.append(t2)
if saveims:
if dual_panel:
fig, ax = leplt.initialize_2panel_3o4ar_cent(fontsize=fontsize,
x0frac=0.085)
ax, ax2 = ax[0], ax[1]
else:
fig, ax = leplt.initialize_1panel_centered_fig(
Wfig=180, Hfig=180, wsfrac=0.7, fontsize=fontsize)
if not color_berry:
if t2 > mm / np.sqrt(3.):
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=band1color)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=band2color)
elif t2 < -mm / np.sqrt(3.):
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=band2color)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=band1color)
else:
ax.plot(km[1],
energy1.reshape(np.shape(km[0])),
'-',
color=graycolor)
ax.plot(km[1],
energy2.reshape(np.shape(km[0])),
'-',
color=graycolor)
else:
b1color = berry1.reshape(np.shape(km[0])) / (2. * np.pi) + 0.5
b2color = berry2.reshape(np.shape(km[0])) / (2. * np.pi) + 0.5
print 'b1color=', b1color
# sys.exit()
energy1 = energy1.reshape(np.shape(km[0]))
energy2 = energy2.reshape(np.shape(km[0]))
# ax.scatter([0., 0.5, 1.0], [0., 0.5, 1.0], c=[0., 0.5, 1.0], edgecolor='none', cmap=cmap, alpha=1)
for ii in range(len(km[0])):
ax.scatter(km[1][ii],
energy1[ii],
c=b1color,
edgecolor='none',
cmap=cmap,
alpha=1.,
vmin=vmin,
vmax=vmax)
ax.scatter(km[1][ii],
energy2[ii],
c=b2color,
edgecolor='none',
cmap=cmap,
alpha=1.,
vmin=vmin,
vmax=vmax)
# plt.show()
# sys.exit()
ax.set_xlabel(r'$k_x$', labelpad=15, fontsize=fontsize)
ax.set_ylabel(r'$\omega$',
labelpad=15,
fontsize=fontsize,
rotation=0)
ax.xaxis.set_ticks([-np.pi, 0, np.pi])
ax.xaxis.set_ticklabels([r'-$\pi$', 0, r'$\pi$'],
fontsize=tick_fontsize)
for tick in ax.yaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax.axis('scaled')
ax.set_xlim(-np.pi, np.pi)
ax.set_ylim(-np.pi, np.pi)
# title = r'$\Delta \omega =$' + '{0:0.3f}'.format(gap) + r' $t_1$'
title = r'$t_2 =$' + '{0:0.3f}'.format(t2) + r' $t_1$, ' + \
r'$\Delta_{AB} =$' + '{0:0.3f}'.format(mm) + r' $t_1$'
ax.text(0.5,
1.1,
title,
ha='center',
va='center',
fontsize=fontsize,
transform=ax.transAxes)
if dual_panel:
# Make second figure
ax2.plot()
ax2.set_xlim(0, 0.2)
ax2.set_ylim(0, 0.7)
ax2.plot([np.min(t2arr), np.max(t2arr)],
[0.0, 3.36864398885 * np.max(t2arr)],
'-',
color='#a1bfdb',
lw=3)
ax2.scatter(t2s, gaps, color='k', zorder=9999999999)
title = 'Gap size: ' + r'$\Delta\omega/t_1 \approx 3.369\, t_2$'
ax2.text(0.5,
1.1,
title,
ha='center',
va='center',
fontsize=fontsize,
transform=ax2.transAxes)
for tick in ax2.xaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
for tick in ax2.yaxis.get_major_ticks():
tick.label.set_fontsize(tick_fontsize)
ax2.set_xlabel(r'$t_2 = \Omega_k^2/8\Omega_g$',
fontsize=fontsize,
labelpad=15)
ax2.set_ylabel(r'$\Delta\omega/t_1 = 2 \Delta\omega/\Omega_k$',
fontsize=fontsize,
rotation=90,
labelpad=35)
# Save the image
plt.savefig(outdir + 'dispersion_{0:04d}'.format(kk) + '.png',
dpi=140)
plt.close('all')
kk += 1
print 'dirac = ', 2. * np.pi / 3., ', ', -2. * np.pi / (
3. * np.sqrt(3.))
ind = np.argmin(energy1)
dirac = kv[ind]
print 'dirac where = ', dirac
print 't2s = ', t2s
print 'gaps = ', gaps
zz = np.polyfit(np.array(t2s), np.array(gaps), 1)
pp = np.poly1d(zz)
print pp
print zz[0]
print zz[1]
imgname = outdir + 'dispersion_'
movname = maindir + 'dispersion_varyt2'
movname += '_nkx{0:06d}'.format(nx) + '_nky{0:06d}'.format(ny)
lemov.make_movie(imgname, movname, indexsz='04', framerate=15)
