def plot_fft2(samp_t, ROWLIST, SETTINGS):
from scipy import signal
from scipy.ndimage.filters import maximum_filter
from scipy.ndimage.morphology import generate_binary_structure, binary_erosion, iterate_structure
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
mag = lambda F: (F * cj(F)).real
T, Npr, rows = fi_rows.shape
for ROW in ROWLIST:
fig, ax = plt.subplots(1, 1)
print(ROW)
matrix = fi_rows[:, :, ROW].T
Fmatrix = np.fft.fftshift(mag(np.fft.fftn(matrix)))
# Fmatrix = mag( np.fft.fftn(matrix) )
shape = Fmatrix.shape
F = Fmatrix.flatten()
F[np.argmax(F)] = 0
newmax = F[np.argmax(F)]
Fmatrix = F.reshape(shape)
thresh = Fmatrix > 0.5 * newmax
samp_e = 1. # vecinos
t_freq = np.fft.fftshift(np.fft.fftfreq(len(t), d=1. / samp_t))
e_freq = np.fft.fftshift(np.fft.fftfreq(Npr, d=1. / samp_e))
# print(e_freq)
defreq = e_freq[1] - e_freq[0]
dtfreq = t_freq[1] - t_freq[0]
logF = np.log(Fmatrix)
filt = logF > np.max(logF) * .7
# print( logF.shape, filt.shape )
# # Peak finder
# neighborhood = np.zeros((5,5)); neighborhood[2,2]=1
# neighborhood = neighborhood.astype(bool)
# local_max = maximum_filter(Fmatrix, footprint=neighborhood)==Fmatrix
# background = (Fmatrix <= np.max(Fmatrix)*.9)
# eroded_background = binary_erosion(background, structure=neighborhood, iterations=1, border_value=1)
# detected_peaks = local_max ^ eroded_background
# peaks_index = np.transpose( np.nonzero( detected_peaks ) )
# for ipeak in peaks_index:
# espacial = "i={0}, k={1:f} 1/vec, la={2:f} vec".format(ipeak[0],e_freq[ipeak[0]],1/e_freq[ipeak[0]])
# temporal = "i={0}, f={1:f} 1/, T={2:f} corrida".format(ipeak[1],t_freq[ipeak[1]],1/t_freq[ipeak[1]])
# print( espacial )
# print( temporal )
# ax.imshow( np.log( Fmatrix*thresh +1 ) ,aspect='auto', extent=[ t_freq[0], t_freq[-1], e_freq[0]+defreq, e_freq[-1]+defreq ] )
ax.imshow(logF * filt,
aspect='auto',
extent=[
t_freq[0], t_freq[-1], e_freq[0] + defreq,
e_freq[-1] + defreq
])
ax.axhline(0, c='r')
ax.axvline(0, c='r')
# plt.show()
return fig, ax
def angle_dist(ROWLIST):
X, V, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = loadargs(sys.argv[2:])
r2 = (R * cj(R)).real
Rang = np.angle(R) #+ np.pi
print(Rang.min(), Rang.max())
print(fi_rows.min(), fi_rows.max())
Tsize, Npr, rows = fi_rows.shape
DAT = [RUNS, ini, end, rows, Name, Tsize, Npr]
ed2cen = lambda ed: .5 * (ed[1:] + ed[:-1])
def bound(A):
for i in range(len(A)):
if A[i] > np.pi:
A[i] = -A[i] + 2 * np.pi
return A
def bound2(A):
for i in range(len(A)):
if A[i] < 0:
A[i] = A[i] + 2 * np.pi
return A
def bound3(A):
for i in range(len(A)):
if A[i] > np.pi * .5:
A[i] = A[i] - np.pi
return A
fi_rel = np.empty_like(fi_rows)
for c in range(Npr):
fi_rel[:, c, :] = fi_rows[:, c, :] - Rang
fig, ax = plt.subplots(1, 1)
for ROW in ROWLIST:
fi_rel_flat = fi_rel[:, :, ROW].flatten()
fi_rel_flat_bound = bound(fi_rel_flat)
# fi_rel_flat_bound = fi_rel_flat
# ax.plot( t, np.angle(R[:,ROW]), 'k' )
H, ed = np.histogram(abs(fi_rel_flat_bound), bins=100, density=True)
bins = ed2cen(ed)
ax.plot(bins, H, 'o-')
plt.show()
def r2_stat(SETTINGS):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
r2 = (R * cj(R)).real
mean = np.mean(r2, axis=0)
std = np.std(r2, axis=0)
print(" ".join(["ROW", "MEAN", "STD"]))
for i in range(len(mean)):
# text = "{0}\t{1:.3f}\t{2:.3f}".format( i, mean[i], std[i])
text = "{0} {1:.3f} {2:.3f}".format(i, mean[i], std[i])
print(text)
print("-" * len(text))
print("\t".join(["<MEAN>", "<STD>/sqrt(size)"]))
print("{0:.3f}\t{1:.3f}".format(np.mean(mean),
np.mean(std) / len(std)**.5))
def stripes_ROW(ROWLIST, SETTINGS):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
r2 = R * cj(R)
T, Npr, rows = fi_rows.shape
DAT = [RUNS, ini, end, rows, Name, T, Npr]
for ROW in ROWLIST:
matrix = fi_rows[:, :, ROW].T # (osc, time)
labeled, n_obj = find_stripes(matrix)
pendientes = []
tiempos = []
# Taking means
centers = np.empty((labeled.shape[0], n_obj))
centers[:, :] = np.nan
for obj in range(1, n_obj + 1): # objects start at 1
filt = labeled == obj
obj_i = np.where(filt)
length = len(set(np.transpose(obj_i)[:, 0]))
if length > int(Npr * .6): # stripe-length % of system
for fila in range(centers.shape[0]):
centers[fila, obj - 1] = np.mean(t[filt[fila, :]])
center_obj = centers[:, obj - 1]
nanfilt = ~np.isnan(center_obj)
if any(nanfilt):
cent = center_obj[nanfilt]
vec = (np.arange(0, Npr) + .5)[nanfilt]
P = np.polyfit(vec, cent, 1)
pendientes.append(P[0])
tiempos.append(np.mean(cent))
block_name = "block_" + str(ROW) + "_" + Name
np.savez(block_name,
pendientes=pendientes,
tiempos=tiempos,
DAT=DAT,
centers=centers,
r2=r2)
def plot_r2(SETTINGS, ROWLIST):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
fig, axROWS = plt.subplots(len(ROWLIST), 1)
cmap = plt.cm.tab10(range(len(ROWLIST)))
r2 = R * cj(R)
if type(axROWS) == np.ndarray:
for i, ROW in enumerate(ROWLIST):
axROWS[i].plot(t, r2[:, ROW].real, color=cmap[i])
axROWS[int(len(axROWS) * .5)].set_ylabel(
r'$\left\Vert R\right\Vert^{2}$', fontsize=13)
axROWS[-1].set_xlabel(r'$Nº\;Corrida\;[10^6 \Delta t]$', fontsize=13)
else:
for ROW in ROWLIST:
axROWS.plot(t, r2[:, ROW].real)
axROWS.set_ylabel(r'$\left\Vert R\right\Vert^{2}$', fontsize=13)
axROWS.set_xlabel(r'$Nº\;Corrida\;[10^6 \Delta t]$', fontsize=13)
# ax.plot(t,r2[:,0].real)
print(np.mean(r2.real, axis=0))
return fig, axROWS
def plot_polar(ROWLIST, SETTINGS):
X, V, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
r2 = (R * cj(R)).real
Tsize, Npr, rows = fi_rows.shape
DAT = [RUNS, ini, end, rows, Name, Tsize, Npr]
# fig_R, ax_R = plt.subplots(1,1,figsize=fsize)
cm2in = lambda w, h: (w / 2.54, h / 2.54)
fig, ax = plt.subplots(2, 3, subplot_kw=dict(polar=True))
rowi = np.asarray([[0, 0], [0, 1], [0, 2], [1, 0], [1, 1], [1, 2]])
p = 1.8
fig.set_size_inches(cm2in(p * 12, p * 8), forward=True)
def bound2(A):
if A < 0:
A += 2 * np.pi
return A
T = 0000
for ROW in ROWLIST:
for c in range(Npr):
i, j = rowi[ROW]
# fi_rel = bound2( fi_rows[T,c,ROW]-np.angle(R[T,ROW]) )
fi_rel = fi_rows[T, c, ROW] - (np.angle(R[T, ROW]))
ax[i, j].scatter(0, np.absolute(R[T, ROW]), 50, color='xkcd:red')
ax[i, j].scatter(fi_rel, 1, 30, color='xkcd:blue')
# ax_R.plot(t,np.absolute(R),zorder=-1)
# ax_R.set_xlabel(r'$Corrida \: [10^6 \: \Delta t ] $')
# fig_R.tight_layout()
fig.tight_layout()
plt.show()
# ax.plot(fi[0,:],r_0[0,:])
return [fig, fig_R], [ax, ax_R]
#!/usr/bin/env python3
import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
from numpy import conjugate as cj
import sys, argparse, pickle
padlength = lambda size: int(2**np.ceil(np.log2(size))-size)
Mag = lambda f: (f*cj(f)).real
cm2in = lambda w,h: (w/2.54,h/2.54)
get_par = lambda d: "{0:.2f}".format(float(d.split('=')[1].strip('$')))
def plot_par(sim,par,ROWS):
par = int(par)
datos = np.load('corr_err_rows_'+sim+'.npz')
corr, cerr, labels = datos['CORR'],datos['CERR'],datos['labels']
ch=np.linspace(0,10,11); N=ch.shape[0]
if sim == 'mec' : sim_label = r'$\rho ='; sim_unit = r'\;1/\sigma$'
if sim == 'ela' : sim_label = r'$ k ='; sim_unit = r'\;\epsilon/\sigma^2$'
fig, ax = plt.subplots(1,1)
# colormap = plt.cm.nipy_spectral #I suggest to use nipy_spectral, Set1,Paired
# ax.set_color_cycle([colormap(i) for i in np.linspace(0., 0.95,4 )])
for ROW in ROWS:
# for d, label in enumerate( labels ):
# label = labels[n]
# corr = C_ce[n,:,0]
# err = C_ce[n,:,1]
ax.errorbar(ch[1:], corr[par,ROW,1:], yerr=cerr[par,ROW,1:],
fmt='-o', ms=3, capsize=5, lw=1.,
def corr(ROWLIST, SETTINGS):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
r2 = (R * cj(R)).real
mag = lambda F: (F * cj(F)).real
T, Npr, rows = fi_rows.shape
def rot(v, j):
# posiciona el indice j primero, preservando orden ciclico
v2 = np.copy(v)
n = len(v)
i = int(n / 2 + 1 + j)
if i > n - 1:
i = int(i - n)
v2[:n - i] = v[i:]
v2[n - i:] = v[:i]
return v2
def plegado(A, j):
# Pliega la matriz de correlacion sobre la col j
# A[i, j+1] -> (A[i,j+1] + A[i,j-1])/2
col = (A.shape[0], 1)
flip = np.copy(A[:, :j])
for i in range(j):
flip[:, i] = A[:, j - 1 - i]
if A.shape[1] % 2 == 0:
B = np.concatenate(
((A[:, j].reshape(col)), ((A[:, j + 1:-1] + flip) * .5),
(A[:, -1].reshape(col))),
axis=1)
else:
B = np.concatenate(
((A[:, j].reshape(col)), ((A[:, j + 1:] + flip) * .5)), axis=1)
return B
def CrossCorr(ROW):
# fi = fi_rows[:,:,ROW] # time, chain
c = xcm[:, :, ROW] # time, chain
C = 0 * np.empty((Npr, Npr))
for i in range(Npr):
sig_i = np.std(c[:, i])
c_i = c[:, i] - np.mean(c[:, i])
for j in range(i + 1):
sig_j = np.std(c[:, j])
c_j = c[:, j] - np.mean(c[:, j])
C[i, j] = np.mean(c_i * c_j) / (sig_i * sig_j)
C[j, i] = C[i, j]
C_cent = np.empty((Npr, Npr))
for ch in range(Npr):
C_cent[ch, :] = rot(C[ch, :], ch)
return C_cent
if Npr % 2 == 0:
pleg_col = int(Npr / 2. - 1)
else:
pleg_col = int((Npr + 1) / 2.)
Cross = 0 * np.empty((Npr, Npr, rows))
r2mean = np.mean(r2, axis=0)
text_start = r"$\left\langle r^{{2}}\right\rangle_"
fig, ax = plt.subplots(1, 1)
for ROW in ROWLIST:
Cross_ROW = CrossCorr(ROW)
CrossPleg = plegado(Cross_ROW, pleg_col)
CrossMean = np.mean(CrossPleg, axis=0)
CrossErr = np.std(CrossPleg, axis=0) / (CrossMean.shape[0])**.5
vecinos = range(1, CrossMean[1:].size + 1)
text = text_start + str(ROW) + r"={0:.2f}$".format(r2mean[ROW])
# ax.plot(vecinos, CrossMean[1:],'-o', ms=3, label=text)
ax.errorbar(vecinos,
CrossMean[1:],
yerr=CrossErr[1:],
capsize=5,
fmt='-o',
ms=3,
label=text)
ax.axhline(0, color='k')
Cross[:, :, ROW] = Cross_ROW
ax.legend()
return ax, Cross
def plot_stripes_ROW(ROWLIST, SETTINGS):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS
r2 = R * cj(R)
T, Npr, rows = fi_rows.shape
DAT = [RUNS, ini, end, rows, Name, T, Npr]
cm2in = lambda w, h: (w / 2.54, h / 2.54)
print(len(ROWLIST))
fig_pend, ax_pend = plt.subplots(len(ROWLIST), 1)
cmap = create_custom_cmap()
# np.save("matrix", matrix )
for index, ROW in enumerate(ROWLIST):
matrix = fi_rows[:, :, ROW].T # (osc, time)
# matrix[0,:] = 0*matrix[0,:]
labeled, n_obj = find_stripes(matrix)
pendientes = []
tiempos = []
color_bar_TRUE = True
if color_bar_TRUE:
fig, ax = plt.subplots(1, 2, gridspec_kw={'width_ratios': [9, 1]})
axim = ax[0]
axcb = ax[1]
else:
fig, ax = plt.subplots(1, 1)
axim = ax
p = 1.
fig.set_size_inches(cm2in(p * 12, p * 12), forward=True)
extent = [t[0], t[-1], 0, Npr]
implot = axim.imshow(
matrix,
interpolation='none',
cmap=cmap, # cmap="gray",
vmin=0,
vmax=2 * np.pi,
extent=extent,
aspect='auto',
origin='lower')
ytick_labels = np.arange(0, Npr, 10)
ytick_pos = np.arange(0, Npr, 10) + .5
axim.set_xlabel(r'$Nº\;Corrida \: [10^6 \: \Delta t ] $', fontsize=13)
axim.set_ylabel('Oscilador', fontsize=13)
axim.set_ylim([0, Npr])
axim.set_xlim([t[0], t[-1]])
axim.set_yticks(ytick_pos)
axim.set_yticklabels(ytick_labels)
# Colorbar
if color_bar_TRUE:
axcb.set_axis_off()
cbar = fig.colorbar(implot, ax=axcb, fraction=1)
ticks = (np.arange(0, 2.5, .5)) * np.pi
cbar.set_ticks(ticks)
cbar.set_ticklabels([
r"$0$", r"$\frac{\pi}{2}$", r"$\pi$", r"$\frac{3\pi}{2}$",
r"$2\pi$"
])
fig.tight_layout()
# Taking means
centers = np.empty((labeled.shape[0], n_obj))
centers[:, :] = np.nan
for obj in range(1, n_obj + 1): # objects start at 1
filt = labeled == obj
obj_i = np.where(filt)
length = len(set(np.transpose(obj_i)[:, 0]))
if length > int(Npr * .6): # stripe-length % of system
for fila in range(centers.shape[0]):
centers[fila, obj - 1] = np.mean(t[filt[fila, :]])
center_obj = centers[:, obj - 1]
nanfilt = ~np.isnan(center_obj)
if any(nanfilt):
cent = center_obj[nanfilt]
vec = (np.arange(0, Npr) + .5)[nanfilt]
P = np.polyfit(vec, cent, 1)
pendientes.append(P[0] * 10**3)
tiempos.append(np.mean(cent))
# axim.plot( cent, vec, color='xkcd:blue', marker='o', ms=3, ls='', lw=3, zorder=2 )
# axim.plot( np.polyval(P, np.arange(0,Npr+1) ), np.arange(0,Npr+1), color='xkcd:orange', lw=4, zorder=1 )
# for m in range(11):
# axim.axhline( 18.2*m, color='xkcd:green' )
# axim.axhline( 100, color='xkcd:green' )
size_ = (11.2, 4.8)
fig_pend.set_size_inches(cm2in(*size_), forward=True)
if type(ax_pend) == np.ndarray:
ax_pend_current_plot = ax_pend[index]
ax_pend[int(len(ROWLIST) *
.5)].set_ylabel(r'$a/v\;[10^6 \Delta t]$')
ax_pend[-1].set_xlabel(r'$Nº\;Corrida \: [10^6 \: \Delta t ] $')
else:
ax_pend_current_plot = ax_pend
ax_pend.set_ylabel(r'$a/v\;[10^6 \Delta t]$')
ax_pend.set_xlabel(r'$Nº\;Corrida \: [10^6 \: \Delta t ] $')
ax_pend_current_plot.plot(tiempos, pendientes, 'o', ms=1)
ax_pend_current_plot.set_xlim([t[0], t[-1]])
ax_pend_current_plot.text(-0.10,
0.99,
"$\\times 10^{-" + str(3) + "}$",
fontsize=9,
transform=plt.gca().transAxes)
# ax_pend[1].plot(t,r2[:,ROW].real, lw=.5)
# ax_pend[1].set_xlabel(r'$Corrida \: [10^6 \: \Delta t ] $')
# ax_pend[1].set_ylabel(r'$r^2$')
# ax_pend[1].set_xlim([t[0],t[-1]])
fig_pend.tight_layout()
if not color_bar_TRUE:
p = 1
fig.set_size_inches(cm2in(p * 16, p * 11))
fig.tight_layout()
plots = ([fig, ax], [fig_pend, ax_pend])
# fig_pend.savefig('pendientes.png')
# md.add_meta('pendientes.png', meta)
return plots
def plot_fft_x(sample, SETTINGS):
xcm, vcm, fi_rows, R, t, imap, rmap = slicer(SETTINGS)
RUNS, ini, end, rows, Name = SETTINGS #loadargs( sys.argv[2:] )
T, Npr, rows = fi_rows.shape
r2 = (R * cj(R)).real
label_bp = r'$\left\langle r^{{2}}\right\rangle _{{{0}}}={1:.2f}$'
r2mean = np.mean(r2, axis=0)
fig, ax = plt.subplots(1, 1)
# size_zoom = (12.3, 10) #(11,10)
# size_ = size_zoom
size_ = (16, 8)
fig.set_size_inches(cm2in(*size_), forward=True)
for ROW in range(rows):
x = xcm[:, :, ROW]
v = vcm[:, :, ROW]
F = np.fft.rfft(x, axis=0)
Fv = np.fft.rfft(v, axis=0)
E = (F * np.conjugate(F)).real
Ev = (Fv * np.conjugate(Fv)).real
f = np.fft.rfftfreq(x.shape[0], d=1. / sample)
# print( x.shape[0], f.shape )
meanE = np.mean(E, axis=1)
meanEv = np.mean(Ev, axis=1)
# ax.plot(f, meanE, label = "{0}".format(ROW))
# ax.plot(f, meanE, label = label_bp.format(ROW, r2mean[ROW]))
# ax.set_xlim([27.25,30])
# ax.set_ylim([10**5, 10**9.5])
ax.set_xlabel(r'$f\;[\frac{10^{-6}}{\Delta t}]$', size=15)
ax.set_ylabel(r'$\mathscr{F}\cdot\mathscr{F}^{\mathscr{*}}$', size=15)
# print(ROW, f[ np.argmax(meanE) ] )
try:
meanRow += meanE
meanRowv += meanEv
except:
meanRow = meanE
meanRowv = meanEv
ax.plot(
f,
meanRow / rows,
label=
r"$\left\langle \left\Vert \mathscr{F}[X_{CM}]\right\Vert ^{2}\right\rangle$",
zorder=1)
ax.plot(
f,
meanRowv / rows,
label=
r"$\left\langle \left\Vert \mathscr{F}[V_{CM}]\right\Vert ^{2}\right\rangle$",
zorder=0)
ax.legend(fontsize=11)
ax.set_yscale('log')
fig.tight_layout()
dat = [f, meanRow / ROW]
return ax, dat