def main(M, z, output_filename):
numpy.random.seed(31415)
x_label = "$R$ [R$_\odot$]"
y_label = "$\\rho$ [g/cm$^{3}$]"
fig, ax = figure_frame(x_label, y_label, xsize=12, ysize=8)
cols = get_distinct(3)
r, rho = get_density_profile(EVtwin, M, z)
pyplot.plot(r.value_in(units.RSun),
rho.value_in(units.g/units.cm**3),
label="EVtwin", c=cols[0])
r, rho = get_density_profile(MESA, M, z)
pyplot.plot(r.value_in(units.RSun),
rho.value_in(units.g/units.cm**3),
label="MESA", c=cols[1])
# Run the merger code.
r, rho = merge_two_stars(0.5*M, 0.5*M, 1|units.yr)
pyplot.plot(r.value_in(units.RSun),
rho.value_in(units.g/units.cm**3),
label="MESA", c=cols[2])
pyplot.semilogy()
if output_filename is not None:
pyplot.savefig(output_filename)
print '\nSaved figure in file', output_filename,'\n'
else:
output_filename = 'merger_stellar_density_profile.png'
pyplot.savefig(output_filename)
print '\nSaved figure in file', output_filename,'\n'
pyplot.show()
def point_mass():
GM = (G * (1.*u.M_sun)).decompose([u.au,u.M_sun,u.year,u.radian]).value
def F(t,x):
x,y,px,py = x.T
a = -GM/(x*x+y*y)**1.5
return np.array([px, py, x*a, y*a]).T
x0 = np.array([1.0, 0.0, 0.0, 2*np.pi]) # Earth
integrator = DOPRI853Integrator(F)
nsteps = 10000
dt = 0.01
nsteps_per_pullback = 10
d0 = 1e-5
LEs, xs = lyapunov(x0, integrator, dt, nsteps,
d0=d0, nsteps_per_pullback=nsteps_per_pullback)
print("Lyapunov exponent computed")
plt.clf()
plt.semilogy(LEs, marker=None)
plt.savefig("point_mass_le.png")
plt.clf()
plt.plot(xs[:,0],xs[:,1],marker=None)
plt.savefig(os.path.join(plot_path,"point_mass.png"))
def plotPeriodogram(fvec, pvec, axes=None, thinning=None):
'''
**plotPeriodogram(fvec, pvec, axes=None, thinning=None)**
Function that plot a periodogram
Parameters
----------
**fvec** : vector containing the frequencies resulting from fftdem() \n
**pvec** : vector containing the power values resulting from fftdem() \n
**axes** : string indicating what type of axes to use. Possible options are: \n\t
- "loglog" \n\t
- "semilogx"\n\t
- "semilogy" \n\t
- None (default option)\n
**thinning** : parameter to thin the data o plot as vectors can be ver large. It will plot only the number of dots indicated by thinning '''
# Wvec=1/fvec
if thinning == None:
thinning = thinningCheckup(fvec)
plt.figure()
if axes == "loglog":
plt.loglog(fvec[range(0,fvec.size,thinning)],pvec[range(0,pvec.size,thinning)])
elif axes == "semilogx":
plt.semilogx(fvec[range(0,fvec.size,thinning)],pvec[range(0,pvec.size,thinning)])
elif axes == "semilogy":
plt.semilogy(fvec[range(0,fvec.size,thinning)],pvec[range(0,pvec.size,thinning)])
else:
plt.plot(fvec[range(0,fvec.size,thinning)],pvec[range(0,pvec.size,thinning)])
plt.title("Periodogram")
plt.ylabel("DFT mean square amplitude")
plt.xlabel("Frequency (1/m)")
plt.show()
def draw_log_hist(X):
X = X.tocsc().tocoo() # collapse multiple records. I don't think it is needed
# we are interested only in existence of a token in user posts, not it's quantity
vf = np.vectorize(lambda x: 1 if x > 0 else 0)
X_data_booleaned = vf(X.data)
X = coo_matrix((X_data_booleaned, (X.row, X.col)), shape=X.shape)
# now we will calculate (1, 1, ... 1) * X to sum up rows
features_counts = np.ones(X.shape[0]) * X
features_counts_sorted = np.sort(features_counts)
features_counts_sorted = features_counts_sorted[::-1] # this is how decreasing sort looks like in numpy
ranks = np.arange(features_counts_sorted.size)
plt.figure()
plt.semilogy(ranks, features_counts_sorted,
color='red',
linewidth=2)
plt.title('For each feature (word), how many users has it at least once?')
plt.ylabel("number of users which has this word at least once")
plt.xlabel("rank")
# plt.show()
return features_counts
def pal5():
from streams import usys
r0 = np.array([[8.312877511,0.242593717,16.811943627]])
v0 = ([[-52.429087,-96.697363,-8.156130]]*u.km/u.s).decompose(usys).value
x0 = np.append(r0,v0)
acc = np.zeros((2,3))
potential = LawMajewski2010()
def F(t,X):
x,y,z,px,py,pz = X.T
dH_dq = potential._acceleration_at(X[...,:3], 2, acc)
return np.hstack((np.array([px, py, pz]).T, dH_dq))
integrator = DOPRI853Integrator(F)
nsteps = 100000
dt = 0.1
print(nsteps*dt*u.Myr)
nsteps_per_pullback = 10
d0 = 1e-5
LEs, xs = lyapunov(x0, integrator, dt, nsteps,
d0=d0, nsteps_per_pullback=nsteps_per_pullback)
print("Lyapunov exponent computed")
plt.clf()
plt.semilogy(LEs, marker=None)
plt.savefig('pal5_le.png')
plt.clf()
plt.plot(np.sqrt(xs[:,0]**2+xs[:,1]**2), xs[:,2], marker=None)
plt.savefig('pal5_orbit.png')
def __call__(self,u,v,w,bx,by,bz):
q = 4
d = (self.sim.zmx, self.sim.ymx, self.sim.xmx)
vspec = spec(u,v,w,dims=d)
bspec = spec(bx,by,bz,dims=d)
plt.subplot(221)
plt.plot(vspec)
plt.semilogy()
plt.title('v-spec')
plt.axis("tight")
plt.ylim([1e-12,1e-2])
plt.subplot(222)
plt.plot(vspec)
plt.loglog()
plt.title('v-spec')
plt.axis("tight")
plt.ylim([1e-12,1e-2])
plt.subplot(223)
plt.plot(bspec)
plt.semilogy()
plt.title('b-spec')
plt.axis("tight")
plt.ylim([1e-12,1e-2])
plt.subplot(224)
plt.plot(bspec)
plt.loglog()
plt.title('b-spec')
plt.axis("tight")
plt.ylim([1e-12,1e-2])
def main():
conn = krpc.connect()
vessel = conn.space_center.active_vessel
streams = init_streams(conn,vessel)
print vessel.control.throttle
plt.axis([0, 100, 0, .1])
plt.ion()
plt.show()
t0 = time.time()
timeSeries = []
vessel.control.abort = False
while not vessel.control.abort:
t_now = time.time()-t0
tel = Telemetry(streams,t_now)
timeSeries.append(tel)
timeSeriesRecent = timeSeries[-40:]
plt.cla()
plt.semilogy([tel.t for tel in timeSeriesRecent], [norm(tel.angular_velocity) for tel in timeSeriesRecent])
#plt.semilogy([tel.t for tel in timeSeriesRecent[1:]], [quat_diff_test(t1,t2) for t1,t2 in zip(timeSeriesRecent,timeSeriesRecent[1:])])
#plt.axis([t_now-6, t_now, 0, .1])
plt.draw()
plt.pause(0.0000001)
#time.sleep(0.0001)
with open('log.json','w') as f:
f.write(json.dumps([tel.__dict__ for tel in timeSeries],indent=4))
print 'The End'
def plot_magnif_from_times(times, magnifs, fmt="--ro", magnif_log_scale=False):
plt.xlabel("time ({})".format(times.unit))
plt.ylabel("magnification")
if magnif_log_scale:
plt.semilogy(times, magnifs, fmt)
else:
plt.plot(times, magnifs, fmt)
def __call__(self,u,v,w,iteration):
q = 4
plt.cool()
if self.x == None:
ny = v.shape[1]
nz = v.shape[0]
self.x,self.y = np.meshgrid(range(ny),range(nz))
x,y = self.x,self.y
if self.iterations == None:
self.iterations = self.sim.bulk_calc(getIteration())
all_itr = self.iterations
if self.xvar == None:
class temp(sim_operation):
def get_params(self):
return ["u"]
def __call__(self,u):
return np.max(self.sim.ddx(u))
self.xvar = self.sim.bulk_calc(temp())
xvar_series = self.xvar
min = np.min(xvar_series)
max = np.max(xvar_series)
if min <= 0:
min = 0.000001
if max <= min:
max = 0.00001
avgu = np.average(u,2)
avgv = np.average(v,2)
avgw = -np.average(w,2)
xd = self.sim.ddx(u)
xd2d = np.max(xd,2)
xd1d = np.max(xd2d,1)
plt.subplot(221)
plt.imshow(avgu)
plt.quiver(x[::q,::q],y[::q,::q],avgv[::q,::q],avgw[::q,::q])
plt.title('Avg u')
plt.axis("tight")
plt.subplot(222)
plt.imshow(xd2d)
plt.title('Max x Variation (y-z)')
plt.axis("tight")
plt.subplot(223)
plt.plot(xd1d)
plt.title('Max x Variation (z)')
plt.axis("tight")
plt.subplot(224)
plt.plot(all_itr,xvar_series, '--')
plt.plot([iteration,iteration],[min,max])
plt.semilogy()
plt.title('Max x Variation (t)')
plt.axis("tight")
def compute_lyapunov(orbit_name, halo_shape='oblate'):
if halo_shape == "prolate":
params = prolate_params
raise NotImplementedError()
elif halo_shape == "oblate":
params = oblate_params
else:
raise ValueError("Invalid halo shape.")
integrator = DOPRI853Integrator(F, func_args=params)
nsteps = 10000
dt = 1.
print(nsteps*dt*u.Myr)
nsteps_per_pullback = 10
d0 = 1e-5
LEs, xs = lyapunov(x0, integrator, dt, nsteps,
d0=d0, nsteps_per_pullback=nsteps_per_pullback)
print("Lyapunov exponent computed")
plt.figure(figsize=(10,10))
plt.clf()
plt.semilogy(LEs, marker=None)
plt.savefig('zotos_le_{}.png'.format(orbit_type))
plt.clf()
plt.plot(xs[:,0], xs[:,1], marker=None)
plt.xlim(0,15)
plt.ylim(-15,15)
plt.savefig('zotos_orbit_{}.png'.format(orbit_type))
def plot_magnif_from_time_terms(time_terms, magnifs, fmt="--ro", magnif_log_scale=False):
plt.xlabel("time term, (t - t_max)/t_E")
plt.ylabel("magnification")
if magnif_log_scale:
plt.semilogy(time_terms, magnifs, fmt)
else:
plt.plot(time_terms, magnifs, fmt)
def advance(self, t, plotresult=False):
y0 = self.concs * self.molWeight
y0 = append(y0, self.thickness)
yt = odeint(self.rightSideofODE, y0, t)
if (plotresult):
import matplotlib.pyplot as plt
plt.figure()
plt.axes([0.1, 0.1, 0.6, 0.85])
plt.semilogy(t, yt)
plt.ylabel('mass concentrations (kg/m3)')
plt.xlabel('time(s)')
#plt.legend(self.speciesnames)
for i in range(len(self.speciesnames)):
plt.annotate(
self.speciesnames[i], (t[-1], yt[-1, i]),
xytext=(20, -5),
textcoords='offset points',
arrowprops=dict(arrowstyle="-"))
plt.show()
self.thickness = yt[-1][-1]
ytt = yt[-1][:-1]
# for iii in range(len(ytt)):
# if ytt[iii]<0:
# ytt[iii]=0.
molDens = ytt / self.molWeight
self.concs = molDens
self.molFrac = molDens / sum(molDens)
self.massFrac = ytt / sum(ytt)
def _main():
args = _parse_input_arguments()
# read the file
handle = open(args.filename)
data = yaml.load(handle)
handle.close()
# Plot relresvecsi.
#pp.subplot(121)
# Mind that the last Newton datum only contains the final ||F||.
num_newton_steps = len(data['Newton results']) - 1
for k in xrange(num_newton_steps):
pp.semilogy(
data['Newton results'][k]['relresvec'],
color=str(1.0 - float(k+1)/num_newton_steps)
)
pp.xlabel('Krylov step')
pp.ylabel('||r||/||b||')
pp.title('Krylov: %s Prec: %r ix-defl: %r extra defl: %r ExpRes: %r Newton iters: %d' %
(data['krylov'], data['preconditioner type'], data['ix deflation'],
data['extra deflation'], data['explicit residual'], num_newton_steps)
)
if args.xmax:
pp.xlim([0, args.xmax])
pp.ylim([1e-10, 10])
# Write the info out to files.
if args.imgfile:
pp.savefig(args.imgfile)
if args.tikzfile:
matplotlib2tikz.save(args.tikzfile)
return
def do_halomass_plots(fosc):
"""Plot halo mass, distance to the halo and the relationship between eq. width and column density"""
ahalos = {
4:CIVPlottingSpectra(5, myname.get_name(4, box=25), None, None, savefile="rand_civ_spectra.hdf5", spec_res=5.,label=labels[4]+" 25"),
9:CIVPlottingSpectra(5, myname.get_name(9, box=25), None, None, savefile="rand_civ_spectra.hdf5", spec_res=5.,label=labels[9]+" 25"),
7:CIVPlottingSpectra(5, myname.get_name(7, box=25), None, None, savefile="rand_civ_spectra.hdf5", spec_res=5.,label=labels[7]+" 25")}
for (ll, ahalo) in ahalos.items():
ahalo.plot_eqw_mass("C",4,1548,color=colors[ll])
plt.legend(loc="upper left")
save_figure(path.join(outdir,"civ_eqwvsmass"))
plt.clf()
for (ll, ahalo) in ahalos.items():
ahalo.plot_eqw_dist("C",4,1548,color=colors[ll])
plt.legend(loc="upper left")
save_figure(path.join(outdir,"civ_eqwvsdist"))
plt.clf()
ahalos['I']=CIVPlottingSpectra(68, path.expanduser("~/data/Illustris"), None, None, savefile="rand_civ_spectra.hdf5", spec_res=5.,label=labels['I']+" 75")
ccc = {1e12: "yellow", 1e15:"red"}
for tag in ('I', 4, 7):
for nmin in (1e12, 1e15):
nminstr = str(np.round(np.log10(nmin),1))
#for (ll, ahalo) in ahalos.iteritems():
ahalos[tag].plot_mass_hist(elem="C",ion=4,color=ccc[nmin], ls=lss[tag],nmin=nmin, label=ahalos[tag].label+" "+nminstr)
plt.legend(loc="upper left",ncol=1)
save_figure(path.join(outdir,"civ_halos_hist"))
plt.clf()
ahalos['I'].plot_eq_width_vs_col_den("C",4,1548)
eqw = np.linspace(-3, 0.5,50)
plt.semilogy(eqw, linear_cog_col(10**eqw, 1548, fosc), '-',color="black")
plt.ylim(1e12,1e16)
plt.xlim(-2.5,0.5)
save_figure(path.join(outdir,"civ_eqwvscolden"))
plt.clf()
def plot_gens(images, rowlabels, losses):
'''
From great jupyter notebook by Tim Sainburg:
http://github.com/timsainb/Tensorflow-MultiGPU-VAE-GAN
'''
examples = 8
fig, ax = plt.subplots(nrows=len(images), ncols=examples, figsize=(18, 8))
for i in range(examples):
for j in range(len(images)):
ax[(j, i)].imshow(create_image(images[j][i]), cmap=plt.cm.gray,
interpolation='nearest')
ax[(j, i)].axis('off')
title = ''
for i in rowlabels:
title += ' {}, '.format(i)
fig.suptitle('Top to Bottom: {}'.format(title))
plt.show()
#fig.savefig(''.join(['imgs/test_',str(epoch).zfill(4),'.png']),dpi=100)
fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(20, 10), linewidth = 4)
D_plt, = plt.semilogy((losses['discriminator']), linewidth=4, ls='-',
color='b', alpha=.5, label='D')
G_plt, = plt.semilogy((losses['generator']), linewidth=4, ls='-',
color='k', alpha=.5, label='G')
plt.gca()
leg = plt.legend(handles=[D_plt, G_plt],
fontsize=20)
leg.get_frame().set_alpha(0.5)
plt.show()
def test_airy_2d(display=False):
""" Test 2D airy function vs 1D function; both
should yield the exact same results for a 1D cut across the 2d function.
And we've already tested the 1D above...
"""
fn2d = airy_2d(diameter=1.0, wavelength=1e-6, shape=(511, 511), pixelscale=0.010)
r, fn1d = airy_1d(diameter=1.0, wavelength=1e-6, length=256, pixelscale=0.010)
cut = fn2d[255, 255:].flatten()
print(cut.shape)
if display:
plt.subplot(211)
plt.semilogy(r, fn1d, label='1D')
plt.semilogy(r, cut, label='2D', color='black', ls='--')
plt.legend(loc='upper right')
plt.axvline(0.251643, color='red', ls='--')
plt.ylabel('Intensity relative to peak')
plt.xlabel('Separation in $\lambda/D$')
ax=plt.subplot(212)
plt.plot(r, cut-fn1d)
ax.set_ylim(-1e-8, 1e-8)
plt.ylabel('Difference')
plt.xlabel('Separation in $\lambda/D$')
#print fn1d[0], cut[0]
#print np.abs(fn1d-cut) #< 1e-9
assert np.all( np.abs(fn1d-cut) < 1e-9)
def test_airy_1d(display=False):
""" Compare analytic airy function results to the expected locations
for the first three dark rings and the FWHM of the PSF."""
lam = 1.0e-6
D = 1.0
r, airyprofile = airy_1d(wavelength=lam, diameter=D, length=20480, pixelscale=0.0001)
# convert to units of lambda/D
r_norm = r*_ARCSECtoRAD / (lam/D)
if display:
plt.semilogy(r_norm,airyprofile)
plt.axvline(1.028/2, color='k', ls=':')
plt.axhline(0.5, color='k', ls=':')
plt.ylabel('Intensity relative to peak')
plt.xlabel('Separation in $\lambda/D$')
for rad in airy_zeros:
plt.axvline(rad, color='red', ls='--')
airyfn = scipy.interpolate.interp1d(r_norm, airyprofile)
# test FWHM occurs at 1.028 lam/D, i.e. HWHM is at 0.514
assert (airyfn(0.5144938) - 0.5) < 1e-5
# test first minima occur near 1.22 lam/D, 2.23, 3.24 lam/D
# TODO investigate/improve numerical precision here?
for rad in airy_zeros:
#print(rad, airyfn(rad), airyfn(rad+0.005))
assert airyfn(rad) < airyfn(rad+0.0003)
assert airyfn(rad) < airyfn(rad-0.0003)
def make_accuracy_plot():
from numpy import linspace, mean
from math import pi
from matplotlib import pyplot as pl
pl.rc('text', usetex=True)
pl.rc('font', family='serif')
pl.axvline(x=1,color='k')
pl.xlabel(r'$$\lambda$$',fontsize=16)
pl.ylabel(r'$$\epsilon$$',fontsize=16)
print "Testing M=0"
it_list = list()
#We test for zero revolutions
err,it,la,x,T = test_standard(M=100000,gooding=False,
house=True,tau=False,iter_max=15,eps=1e-5, rnd_seed=4562,N=0)
it_list.append(mean(it))
pl.semilogy(la,err,'k.')
#We test for multiple revolutions
for rev in range(50):
print "Testing M=" + str(rev)
err,it,la,x,T = test_standard(M=10000,gooding=False,
house=True,tau=False,iter_max=15,eps=1e-8, rnd_seed=4562+rev+1,N=rev+1)
pl.semilogy(la,err,'k.')
it_list.append(mean(it))
pl.figure()
pl.plot(it_list)
def semilogy(x_vals, y_vals, x_label, y_label, figsize=(3.5, 2.5)):
"""绘图（y取对数）。"""
set_fig_size(mpl, figsize)
plt.semilogy(x_vals, y_vals)
plt.xlabel(x_label)
plt.ylabel(y_label)
plt.show()
def plot(self):
from matplotlib import pyplot
pyplot.figure()
it_ls = [] #ls = lineseach
y_ls = []
it_ga = [] #gradient approximation
y_ga = []
gradApprox = False
for i in range(len(self.x)):
y = norm( self.f_x[i] ) + 10**-9
if self.notes.has_key(i):
if self.notes[i] == 'starting gradient approximation':
gradApprox = True
if self.notes[i] == 'finished gradient approximation':
gradApprox = False
if gradApprox:
it_ga.append( i )
y_ga.append( y )
else:
it_ls.append( i )
y_ls.append( y )
pyplot.semilogy( it_ls, y_ls, 'go')
pyplot.semilogy( it_ga, y_ga, 'bx')
pyplot.xlabel('function evaluation')
pyplot.ylabel('norm(f(x)) + 10**-9')
pyplot.legend(['line searches', 'gradient approx' ])
pyplot.show()
def _double_hit_check(x):
# first PSD
W, fr = PSD(x, dt=dt)
# second PSD: look for oscillations in PSD
W2, fr2 = PSD(W, dt=fr[1])
upto = int(0.01 * len(x))
max_impact = np.max(W2[:upto])
max_after_impact = np.max(W2[upto:])
if plot_figure:
import matplotlib.pyplot as plt
plt.subplot(121)
l = int(0.002*len(x))
plt.plot(1000*dt*np.arange(l), x[:l])
plt.xlabel('t [ms]')
plt.ylabel('F [N]')
plt.subplot(122)
plt.semilogy((W2/np.max(W2))[:5*upto])
plt.axhline(limit, color='r')
plt.axvline(upto, color='g')
plt.xlabel('Double freq')
plt.ylabel('')
plt.show()
if max_after_impact / max_impact > limit:
return True
else:
return False
def create_plot(freq_mat, data_mat, length_vec, color_keys, plot_name, output_dir=""):
print("Creating plot: {0:s}".format(plot_name) )
pl.figure(1)
pl.clf()
pl.hold(True)
for idx, freq in enumerate(freq_mat):
data = data_mat[idx]
length = length_vec[idx]
num_pos = 0
num_neg = 0
for el in data:
if el > 0:
num_pos += 1
if el < 0:
num_neg +=1
if num_pos > 0:
pl.semilogy(freq/1e9, data, color_keys[length])
if num_neg > 0:
pl.semilogy(freq/1e9, -data, color_keys[length] + ":")
pl.xlabel('Frequency (GHz)')
pl.ylabel("PUL R ($\Omega$/m)")
output_name = os.path.join(output_dir, plot_name)
pl.savefig(output_name)
def plot_trunc_gr_model(aval, bval, min_mag, max_mag, dmag, catalogue=None,
completeness=None, figure_size=None, filename=None, filetype='png',
dpi=300):
"""
Plots a Gutenberg-Richter model
"""
input_model = TruncatedGRMFD(min_mag, max_mag, dmag, aval, bval)
if not catalogue:
# Plot only the modelled recurrence
annual_rates, cumulative_rates = _get_recurrence_model(input_model)
plt.semilogy(annual_rates[:, 0], annual_rates[:, 1], 'b-')
plt.semilogy(annual_rates[:, 0], cumulative_rates, 'r-')
plt.xlabel('Magnitude', fontsize='large')
plt.ylabel('Annual Rate', fontsize='large')
plt.legend(['Incremental Rate', 'Cumulative Rate'])
_save_image(filename, filetype, dpi)
else:
completeness = _check_completeness_table(completeness, catalogue)
plot_recurrence_model(input_model,
catalogue,
completeness,
input_model.bin_width,
figure_size,
filename,
filetype,
dpi)
def plot_prob_sums(prob_sums, folder):
"""Plot the probability sums of the original motif against iterations."""
# Create the folder if it does not exist
if not os.path.exists("figures/%s" % folder):
os.makedirs("figures/%s" % folder)
# Plot on automatic axis
figure_file = "figures/%s/probability_mass.png" % folder
plt.title("%s | Prob Mass of Original Motif" % folder)
plt.ylabel("Probability Mass")
plt.xlabel("Iterations")
plt.grid(True)
plt.plot(prob_sums)
pylab.savefig(figure_file, format="png")
plt.clf()
# Plot on semi-log axis
figure_file = "figures/%s/probability_mass_semilog.png" % folder
plt.title("%s | Prob Mass of Original Motif" % folder)
plt.xlabel("Iterations")
plt.ylabel("Probability Mass (semi-log)")
plt.grid(True)
plt.semilogy(prob_sums)
pylab.savefig(figure_file, format="png")
plt.clf()
def plot_KLs(KLs, folder, filename, title):
if not os.path.exists("figures/%s" % folder):
os.makedirs("figures/%s" % folder)
# Plot on automatic axis
figure_file = "figures/%s/%s.png" % (folder, filename)
plt.title("%s | %s" % (folder, title))
plt.ylabel("KL")
plt.xlabel("Iterations")
plt.grid(True)
plt.plot(KLs)
pylab.savefig(figure_file, format="png")
plt.clf()
# Plot on semi-log axis
try:
figure_file = "figures/%s/%s_semilog.png" % (folder, filename)
plt.title("%s | %s" % (folder, title))
plt.xlabel("Iterations")
plt.ylabel("KL (semi-log)")
plt.grid(True)
plt.semilogy(KLs)
pylab.savefig(figure_file, format="png")
except:
debug("Could not generate semilog plot.")
plt.clf()
def plot_cmf_with_arbitrary_input(catalog, cmf_input, bins=20, hist_range=(4, 7),
mass_column_name='mass', **kwargs):
fig = plt.figure()
number_in_bin_q2, bin_edges, ch = plt.hist(np.log10(catalog[mass_column_name]), cumulative=-1, log=True, bins=bins, range=hist_range)
bin_centers_q2 = (bin_edges[1:] + bin_edges[:-1])/2.
plt.clf()
plt.plot(bin_centers_q2, number_in_bin_q2, 'ko' )
plt.semilogy()
plt.xlabel(r"log$_{10}$ (M$_{GMC}$ / M$_\odot$)")
plt.ylabel("n(M > M')")
M_0, N_0, gamma = cmf_input
m_array = np.linspace(min(bin_edges), max(bin_edges), 50)
n_array = truncated_cloudmass_function([M_0, N_0, gamma], 10**m_array)
plt.plot(m_array, n_array, label="$\\gamma = {0:.2f}$,\n$M_0={1:.2e}$,\n$N_0={2:.1f}$".format(gamma, M_0, N_0))
text_string = r"$N(M' > M) = N_0 \left [ \left ( \frac{M}{M_0} \right )^{\gamma+1} - 1 \right ]$"
plt.text(4.1, 3, text_string, fontsize=18)
plt.xlim(*hist_range)
plt.ylim(0.7, 1e3)
plt.legend(loc='upper right')
return fig
def on_off_experiment2(num_motifs=100,filename="gini-vs-mi-correlation-in-on-off-spoofs.pdf"):
"""compare MI vs Gini on biological_motifs"""
bio_motifs = [getattr(tfdf,tf) for tf in tfdf.tfs]
Ns = map(len, bio_motifs)
spoofses = [spoof_on_off_motif(motif,num_motifs=num_motifs,trials=1) for motif in bio_motifs]
spoof_ginises = mmap(motif_gini,tqdm(spoofses))
spoof_mises = mmap(total_motif_mi,tqdm(spoofses))
cors, ps = [],[]
for ginis, mis in zip(ginises, mises):
cor, p = pearsonr(ginis,mis)
cors.append(cor)
ps.append(p)
q = fdr(ps)
plt.scatter(cors,ps,filename="gini-vs-mi-correlation-in-on-off-spoofs.pdf")
plt.plot([-1,1],[q,q],linestyle='--',label="FDR-Adjusted Significance Level")
plt.semilogy()
plt.legend()
plt.xlabel("Pearson Correlation Coefficient")
plt.ylabel("P value")
plt.xlim([-1,1])
plt.ylim([10**-4,1+1])
cor_ps = zip(cors,ps)
sig_negs = [(c,p) for (c,p) in cor_ps if c < 0 and p < q]
sig_poses = [(c,p) for (c,p) in cor_ps if c > 0 and p < q]
insigs = [(c,p) for (c,p) in cor_ps if p > q]
def weighted_correlation(cor_p_Ns):
cors,ps,Ns = transpose(cor_p_Ns)
return sum([cor*N for (cor,N) in zip (cors,Ns)])/sum(Ns)
plt.title("Gini-MI Correlation Coefficient vs. P-value for On-Off Simulations from Prokaryotic Motifs")
maybesave(filename)
def plota_teste5(arqsaida):
n, c, t = np.loadtxt(arqsaida, unpack=True)
# Calcula os coeficientes de um ajuste a um polinômio de grau 2 usando
# o método dos mínimos quadrados
coefs = np.polyfit(n, c, 2)
p = np.poly1d(coefs)
# set_yscale('log')
# set_yscale('log')
plt.semilogy(n, p(n), label='$n^2$')
plt.semilogy(n, c, 'ro', label='bubble sort')
# Posiciona a legenda
plt.legend(loc='upper left')
# Posiciona o título
plt.title('Análise da complexidade de \ntempo do método da bolha')
# Rotula os eixos
plt.xlabel('Tamanho do vetor (n)')
plt.ylabel('Número de comparações')
plt.savefig('bubble5.png')
plt.show()
def throughputs(output='throughputs.pdf'):
"""
Plot throughputs, compares to HST WFC3 UVIS F600LP
:param output: name of the output file
:type output: str
:return: None
"""
#comparison
bp1 = S.ObsBandpass('wfc3,uvis2,f600lp')
#VIS
bp, bpEoL = _VISbandpass()
#ghost
bpG, bpEoLG = _VISbandpassGhost()
#plot
plt.semilogy(bp1.wave/10., bp1.throughput, 'r-', label='WFC3 F600LP')
plt.semilogy(bp.wave/10., bp.throughput, 'b-', label='VIS Best Estimate')
plt.semilogy(bpEoL.wave/10., bpEoL.throughput, 'g--', label='VIS EoL Req.')
plt.semilogy(bpG.wave/10., bpG.throughput, 'm-', label='VIS Ghost')
plt.semilogy(bpEoLG.wave/10., bpEoLG.throughput, 'y-.', label='VIS Ghost EoL')
plt.xlim(230, 1100)
plt.xlabel(r'Wavelength [nm]')
plt.ylabel(r'Total System Throughput')
plt.legend(shadow=True, fancybox=True, loc='best')
plt.savefig(output)
plt.close()
def runAlgorithm(data, d, k):
W = None
alphas = 10.0 ** linspace(-4, 1, 6)
#alphas = 10.0 ** array([-2,-1,0,-2,-3,-2])
expNum = len(alphas)
allStates = []
for (alpha, i) in zip(alphas, count(1)):
states = jointlyPenalizedMultiplePCA(data, d, alpha, W, k=k)
allStates.append(states)
weightVectors = [s.weights for s in states]
W = array(weightVectors)
figure()
for (states, alpha, i) in zip(allStates, alphas, count(1)):
subplot(expNum, 1, i)
weightVectors = [s.weights for s in states]
W = array(weightVectors)
plot(W.T)
title(r'Run with $\alpha $ = %f' % alpha)
figure()
for (states, alpha, i) in zip(allStates, alphas, count(1)):
subplot(expNum, 1, i)
lossVectors = [s.squaredLosses() for s in states]
L = array(lossVectors)
semilogy(L.T)
title(r'Run with $\alpha $ = %f' % alpha)
show()
def norm_plot(self, loss_name='loss', acc_name='acc'):
self.norm = [float(i) for i in self.norm]
x = [
np.linspace(0,
self.data[i].shape[0],
self.data[i].shape[0],
endpoint=False) * self.norm[i]
for i in range(len(self.data))
]
plt.figure(figsize=(15, 7))
plt.subplot(121)
for i, d in enumerate(self.data):
loss_name_tmp = loss_name
if loss_name not in self.header[i]:
for h in self.header[i]:
if 'loss' in h:
loss_name_tmp = h
print('%d: use %s\n' % (i, loss_name_tmp))
break
if len(self.name) == len(self.labels):
plt.semilogy(x[i],
d[:, self.header[i].index(loss_name_tmp)],
label='train_loss_' + self.labels[i])
plt.semilogy(x[i],
d[:,
self.header[i].index('val_' + loss_name_tmp)],
label='val_loss_' + self.labels[i])
else:
plt.semilogy(x[i],
d[:, self.header[i].index(loss_name_tmp)],
label='train_loss_' + str(i))
plt.semilogy(x[i],
d[:,
self.header[i].index('val_' + loss_name_tmp)],
label='val_loss_' + str(i))
plt.legend(loc=0)
plt.xlabel('Training time (h)')
plt.ylabel('Loss')
plt.grid()
plt.subplot(122)
for i, d in enumerate(self.data):
acc_name_tmp = acc_name
if acc_name not in self.header[i]:
for h in self.header[i]:
if 'acc' in h:
acc_name_tmp = h
print('%d: use %s' % (i, acc_name_tmp))
break
if len(self.name) == len(self.labels):
plt.plot(x[i],
d[:, self.header[i].index(acc_name_tmp)],
label='train_acc_' + self.labels[i])
plt.plot(x[i],
d[:, self.header[i].index('val_' + acc_name_tmp)],
label='val_acc_' + self.labels[i])
else:
plt.plot(x[i],
d[:, self.header[i].index(acc_name_tmp)],
label='train_acc_' + str(i))
plt.plot(x[i],
d[:, self.header[i].index('val_' + acc_name_tmp)],
label='val_acc_' + str(i))
plt.legend(loc=0)
plt.xlabel('Training time (h)')
plt.ylabel('Accuracy')
plt.grid()
mp.figure('Filter', facecolor='lightgray')
mp.subplot(221)
mp.title('Time Domain', fontsize=16)
mp.ylabel('Signal', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
mp.plot(times[:178], noised_sigs[:178],
c='orangered', label='Noised')
mp.legend()
mp.subplot(222)
mp.title('Frequency Domain', fontsize=16)
mp.ylabel('Power', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
mp.semilogy(freqs[freqs >= 0],
noised_pows[freqs >= 0],
c='limegreen', label='Noised')
mp.legend()
mp.subplot(223)
mp.xlabel('Time', fontsize=12)
mp.ylabel('Signal', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
mp.plot(times[:178], filter_sigs[:178],
c='hotpink', label='Filter')
mp.legend()
mp.subplot(224)
mp.xlabel('Frequency', fontsize=12)
mp.ylabel('Power', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
def main():
Nz = []
for zm, zw in [(0.5, 0.1), (0.8, 0.2), (1.1, 0.25)]:
zarr = np.linspace(zm - 3 * zw, zm + 3 * zw, 1000)
Nzarr = np.exp(-(zm - zarr)**2 / (2 * zw**2))
Nz.append((zarr, Nzarr))
Nt = len(Nz)
t = GSKY_Theory(Nz)
larr = np.linspace(100, 2000, 20)
import matplotlib.pyplot as plt
plt.figure(figsize=(10, 6))
for count in range(2):
if (count == 0):
style = '-'
lf = lambda x: x
else:
style = '--'
lf = lambda x: None
t.set_cosmology(
ccl.Cosmology(Omega_c=0.3,
Omega_b=0.045,
h=0.67,
sigma8=0.83,
n_s=0.96))
t.set_params({'mmin': 11.5})
plt.subplot(2, 4, 1)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
for j in range(i, Nt):
plt.plot(larr,
t.getCls('gg', larr, i, j),
style,
label='GG %i x %i' % (i, j))
plt.subplot(2, 4, 2)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
for j in range(Nt):
plt.plot(larr,
t.getCls('gs', larr, i, j),
style,
label=lf('GS %i x %i' % (i, j)))
plt.subplot(2, 4, 3)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
for j in range(i, Nt):
plt.plot(larr,
t.getCls('ss', larr, i, j),
style,
label=lf('SS %i x %i' % (i, j)))
plt.subplot(2, 4, 4)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
plt.plot(larr,
t.getCls('gk', larr, i),
style,
label=lf('GK %i' % i))
plt.subplot(2, 4, 5)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
plt.plot(larr,
t.getCls('sk', larr, i),
style,
label=lf('SK %i' % i))
plt.subplot(2, 4, 6)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
plt.plot(larr,
t.getCls('gy', larr, i),
style,
label=lf('GY %i' % i))
plt.subplot(2, 4, 7)
plt.gca().set_prop_cycle(None)
for i in range(Nt):
plt.plot(larr,
t.getCls('sy', larr, i),
style,
label=lf('SY %i' % i))
plt.subplot(2, 4, 8)
plt.gca().set_prop_cycle(None)
plt.plot(larr, t.getCls('kk', larr), style, label=lf('KK'))
for c in range(1, 9):
plt.subplot(2, 4, c)
plt.xlabel('$\ell$')
plt.ylabel('$C_\ell$')
plt.legend(fontsize=6)
plt.semilogy()
plt.tight_layout()
plt.show()
batch_size=batch_size,
epochs=epochs,
verbose=2,
validation_split=vsplit,
callbacks=callbacks_list)
# score = model.evaluate(x_test, y_test,
# batch_size=batch_size, verbose=1)
# print('Test score:', score[0])
# print('Test accuracy:', score[1])
if vsplit:
# summarize history for loss
fig = plt.figure()
# plt.plot(history.history['loss'])
# plt.plot(history.history['val_loss'])
plt.semilogy(history.history['loss'])
plt.semilogy(history.history['val_loss'])
plt.title('mse')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()
# visualisation
# 1. analytical solution
x_test = np.zeros((ny * nx, 2))
surface = np.zeros((ny, nx))
for i, x in enumerate(x_space):
for j, y in enumerate(y_space):
x_test[i * nx + j] = [x, y]
surface[i][j] = undecorated(analytic_solution)([x, y])
if BER_vector[-1] < BER_go_on_in_smaller_steps:
EbN0_dB_vector = np.append(
EbN0_dB_vector,
EbN0_dB_vector[-1] + EbN0_dB_small_stepwidth)
else:
EbN0_dB_vector = np.append(
EbN0_dB_vector,
EbN0_dB_vector[-1] + EbN0_dB_normal_stepwidth)
BER_vector = np.append(BER_vector, 0)
else:
ready = True
#Plot
plt.figure()
plt.semilogy(EbN0_dB_vector, BER_vector)
plt.xlabel('Eb/N0')
plt.ylabel('Bit Error Rate ')
plt.grid(True)
plt.show()
np.savez(os.path.join(pathname, 'BER_results'),
EbN0_dB_vector=EbN0_dB_vector,
BER_vector=BER_vector)
plt.savefig(os.path.join(pathname, 'BER_figure.pgf'))
plt.savefig(os.path.join(pathname, 'BER_figure.pdf'))
res_dict = {
"EbN0_dB_vector": EbN0_dB_vector,
"BER_vector": BER_vector,
for i in range(int(len(max_index[0]))):
if i == 0:
#aux = np.concatenate((np.array(list(('{0:04b}'.format(int(max_index[0][0])-1)).zfill(4)), dtype=int),np.array(list(('{0:04b}'.format(int(max_index[0][1])-1)).zfill(4)), dtype=int)), axis=0)
aux = np.concatenate((np.array(list(('{0:04b}'.format(int(max_index[0][0]))).zfill(4)), dtype=int),np.array(list(('{0:04b}'.format(int(max_index[0][1]))).zfill(4)), dtype=int)), axis=0)
elif i > 1:
aux = np.concatenate((aux,np.array(list(('{0:04b}'.format(int(float(max_index[0][i])))).zfill(4)), dtype=int)), axis=0)
ipHat_soft = aux
#counting the errors
#nErr_hard = np.zeros((len(Eb_N0_dB)))
nErr_hard[yy] = len(np.where((ip - ipHat_hard[0]) != 0)[0])
#nErr_soft = np.zeros((len(Eb_N0_dB)))
nErr_soft[yy] = len(np.where((ip - ipHat_soft) != 0)[0])
theoryBer = 0.5*special.erfc(np.sqrt(10**(Eb_N0_dB/10)))
simBer_hard = nErr_hard/N
simBer_soft = nErr_soft/N
plt.semilogy(Eb_N0_dB, theoryBer,'bo',Eb_N0_dB, theoryBer,'b')
plt.semilogy(Eb_N0_dB, theoryBer,'bo',label='theory - uncoded')
plt.semilogy(Eb_N0_dB, simBer_hard, 'mo',Eb_N0_dB, simBer_hard, 'm')
plt.semilogy(Eb_N0_dB, simBer_hard, 'mo',label="simulation - Hamming 7,4 (hard)")
plt.semilogy(Eb_N0_dB, simBer_soft,'ro',Eb_N0_dB, simBer_soft, 'r')
plt.semilogy(Eb_N0_dB, simBer_soft,'ro',label='simulation - Hamming 7,4 (soft)')
plt.xlabel("Eb/No, dB")
plt.ylabel("Bit Error Rate")
#plt.legend(borderaxespad=0)
plt.legend()
plt.show()
def opt_lf_num_bits(lf_params,
min_bits,
max_bits,
rms_filt_error=0.1,
noise_figure=1,
sim_steps=1000,
sim_runs=10,
fpoints=512,
mode="tdc",
sigma_ph=0.1,
tdc_in_stdev=1,
plot=False):
""" optimize number of bits for a digital direct form-I implementation using two's complement
representation fixed point words with all parts of the data path with same data representation
args:
noise_figure: the maximum dB increase in noise due to loop filter quantization
rms_filt_error : RMS value in dB for allowable filter error
"""
print("\n********************************************************")
print("Optimizing loop filter digital direct form-I implementation for")
print("number of bits in fixed point data words utilized")
sign_bits = 1
# fint number of integer bits needed
int_bits = n_int_bits(lf_params)
print("\n* Integer bits = %d" % int_bits)
""" Optimization for quantization noise
"""
print("\n* Optimizing for quantization noise:")
# find optimal number of bits for quantization noise
# generate white noise signal w simulating "regular" activity
if lf_params["mode"] == "tdc":
# w = np.floor(np.random.normal(0, 0.1*lf_params["m"], sim_steps))
w = np.floor(np.random.normal(0, tdc_in_stdev, sim_steps))
else: # BBPD mode, test sequence is random +/- 1
# w = np.random.normal(0, np.sqrt((1-2/np.pi)), sim_steps)
w = np.random.choice((-1, 1), sim_steps) # ate
pow_npd_post_lf = var_npd_post_lf(
lf_params, mode=mode) # variance of TDC noise at loop filter
lf_ideal = LoopFilterPIPhase(ignore_clk=True, **lf_params)
x_ideal = np.zeros(sim_steps)
for n in range(sim_steps):
x_ideal[n] = lf_ideal.update(w[n], 0)
mses = []
bit_range = range(min_bits - int_bits - 1, max_bits - int_bits)
for frac_bits in bit_range:
# use a large number of int bits to avoid overflow. Tuning here is with frac bits as
runs = np.zeros(sim_runs)
for m in range(sim_runs):
lf_quant = LoopFilterPIPhase(ignore_clk=True,
int_bits=32,
frac_bits=frac_bits,
quant_filt=False,
**lf_params)
x_quant = np.zeros(sim_steps)
for n in range(sim_steps):
x_quant[n] = lf_quant.update(w[n], 0)
runs[m] = np.var(x_ideal - x_quant)
mse = np.average(runs)
print(
"\tBits = %d,\t #(sign,int,frac) = (%d,%d,%d), \tQuant noise power = %E LSB^2"
% ((sign_bits + int_bits + frac_bits), sign_bits, int_bits,
frac_bits, mse))
mses.append(mse)
threshold = (10**(noise_figure / 10.0) - 1) * pow_npd_post_lf
print("Threshold=%E, PD noise post-LF=%E" % (threshold, pow_npd_post_lf))
for n, v in enumerate(mses):
if v < threshold:
break
opt_frac_bits_qn = bit_range[n]
print(
"* Optimum int bits = %d, frac bits = %d, sign bits = 1, quant noise = %.3f LSB^2"
% (int_bits, opt_frac_bits_qn, mses[n]))
if plot:
plt.figure(1)
plt.clf()
plt.semilogy(np.arange(min_bits, max_bits + 1), mses)
plt.title("RMS Quantization Noise versus Filter Coefficient Bits")
plt.xlabel("Total bits")
plt.grid()
plt.ylabel("DAC LSB$^2$")
razavify()
ticks = plt.yticks()[0]
plt.yticks(ticks,
["10$^{%d}$" % int(round(np.log10(x))) for x in ticks])
plt.xlim(min_bits, max_bits)
ticks = plt.xticks()[0]
plt.xticks(ticks, ["%d" % x for x in ticks])
#////////////////////////////////////////////////////////////////////////////////////
""" Optimization for filter accuracy
"""
print("\n* Optimizing for filter design accuracy:")
fmin = 1e2
fref = lf_params["fref"]
b = [lf_params["b0"], lf_params["b1"]]
a = [
1,
-1,
]
f, h_ideal = scipy.signal.freqz(b,
a,
np.geomspace(fmin, fref / 2, fpoints),
fs=fref)
s = 2j * np.pi * f
l = 2 * np.pi * lf_params["kpd"] * lf_params["kdco"] * h_ideal / s
g = l / (1 + l)
bit_range = range(min_bits - int_bits - 1, max_bits - int_bits)
mses = []
# print(lf_params["b0"], lf_params["b1"])
for frac_bits in bit_range:
_lf_params = quant_lf_params(lf_params, int_bits, frac_bits)
b = [_lf_params["b0"], _lf_params["b1"]]
# print(_lf_params["b0"], _lf_params["b1"])
f, h = scipy.signal.freqz(b,
a,
np.geomspace(fmin, fref / 2, fpoints),
fs=fref)
s = 2j * np.pi * f
l = 2 * np.pi * lf_params["kpd"] * lf_params["kdco"] * h / s
_g = l / (1 + l)
# mses.append(np.var(20*np.log10(np.abs(h[1:]))-20*np.log10(np.abs(h_ideal[1:]))))
mses.append(
np.var(20 * np.log10(np.abs(g[1:])) -
20 * np.log10(np.abs(_g[1:]))))
# print("\tN bits = %d\tMSE = %E dB^2"%(frac_bits+int_bits+sign_bits, mses[-1]))
print(
"\tBits = %d,\t #(sign,int,frac) = (%d,%d,%d), \tMSE = %E LSB^2" %
((sign_bits + int_bits + frac_bits), sign_bits, int_bits,
frac_bits, mses[-1]))
n = len(mses) - 1
for n, v in enumerate(mses):
if v < rms_filt_error**2:
break
opt_frac_bits_filt_acc = bit_range[n]
print(
"* Optimum int bits = %d, frac bits = %d, sign_bits=1, quant noise = %E LSB^2"
% (int_bits, opt_frac_bits_filt_acc, mses[n]))
if plot:
plt.figure(2)
plt.clf()
plt.semilogy(np.arange(min_bits, max_bits + 1), mses)
plt.title("MSE Filter Error (dB) versus Filter Coefficient Bits")
plt.xlabel("Total bits")
plt.ylabel("MSE [dB$^2$]")
plt.grid()
razavify()
ticks = plt.yticks()[0]
plt.yticks(ticks,
["10$^{%d}$" % int(round(np.log10(x))) for x in ticks])
plt.xlim(min_bits, max_bits)
ticks = plt.xticks()[0]
plt.xticks(ticks, ["%d" % x for x in ticks])
frac_bits = max(opt_frac_bits_qn, opt_frac_bits_filt_acc)
print("\n* Optimization complete:")
print("\tInt bits = %d, frac bits = %d, sign bits = 1" %
(int_bits, frac_bits))
print("\tTotal number bits = %d" % (int_bits + frac_bits + sign_bits))
return int_bits, frac_bits
previousMAE = float('inf')
print(2)
for i in range(1000):
ANN.fit(featTrain, labelsTrain)
epochs.append(ANN.n_iter_)
yPredict = ANN.predict(featTrain)
MAEtrain.append(np.mean(np.abs(yPredict - labelsTrain)))
yPredict = ANN.predict(featVal)
MAEval.append(np.mean(np.abs(yPredict - labelsVal)))
if MAEval[-1] >= previousMAE and ANN.n_iter_ % 10 == 0:
break
if ANN.n_iter_ % 10 == 0:
previousMAE = MAEval[-1]
# Graph training and validation error vs training time
plt.figure(2, figsize=(12, 4), dpi=100)
plt.clf()
plt.semilogy(epochs, MAEval, label='validation')
plt.semilogy(epochs, MAEtrain, label='training')
plt.legend()
plt.xlabel('Epochs')
plt.ylabel('MAE in ANN output')
plt.savefig('figure2.png')
# Pickle ANN for later use
pickle.dump(ANN, open("ANN.pkl", "wb"))
pickle.dump(trans, open("normalizationScalar.pkl", "wb"))
t2 = time.time()
print('Run time (s):', round(t2 - t, 2))
testerr=[]
trainerr=[]
err = np.zeros(len(d))
for i,k in enumerate(n_learners):
J = np.zeros(nFolds)
for iFold in range(nFolds):
Xti, Xvi, Yti, Yvi = ml.crossValidate(Xtr, Ytr, nFolds, iFold);
gbm0 = GradientBoostingClassifier(n_estimators=k)
gbm0.fit(Xti, Yti)
mean_acc = gbm0.score(Xti, Yti)
trainerr.append(1 - mean_acc);
mean_acc = gbm0.score(Xvi, Yvi)
J[iFold] = 1 - mean_acc
testerr.append(1 - mean_acc);
plt.semilogy(depth,testerr,marker='o',color = 'r');
plt.semilogy(depth,trainerr,marker='o',color = 'b');
plt.xlabel(Number of learners)
plt.ylabel('MSE')
plt.show()
testAUC = []
trainAUC = []
err = np.zeros(len(d))
for i,k in enumerate(depth):
J = np.zeros(nFolds)
for iFold in range(nFolds):
Xti, Xvi, Yti, Yvi = ml.crossValidate(Xtr, Ytr, nFolds, iFold);
gbm0 = GradientBoostingClassifier(n_estimators=1000,max_depth=k)
gbm0.fit(Xti, Yti)
mean_acc = gbm0.score(Xti, Yti)
slice1 = np.array(df1, dtype=float)
slice1.shape
freqs, times, Sx = signal.spectrogram(slice1, fs=sampling_frequency, nperseg=1024, noverlap=512, window='hanning', return_onesided=True)
freqs.shape
plt.pcolormesh(times, freqs, Sx, shading='gouraud')
plt.ylabel('Frequency [Hz]')
plt.xlabel('Time [sec]')
plt.show()
"""##### Power Spectral Density """
f, Pxx_den = signal.welch(slice1, sampling_frequency, nperseg=1024)
plt.semilogy(f, Pxx_den)
plt.ylim([0.5e-3, 1])
plt.xlabel('frequency [Hz]')
plt.ylabel('PSD [V**2/Hz]')
plt.show()
"""#### FFT on ADC_02
##### DFT
"""
y2 = df2['var_names']
c2 = fftpack.fft(y2) # returns DFT of real or complex sequences
f2 = np.fft.fftfreq(len(y2),time_step)
print(c2.shape)
alpha_is_correct = tf.equal(tf.argmax(alpha, 1), tf.argmax(label, 1))
accuracy = tf.reduce_mean(tf.cast(alpha_is_correct, tf.float32))
alpha_values = alpha.eval({inp: test_images, label: test_labels})
index_first_mistaken_number = alpha_is_correct.eval({
inp: test_images,
label: test_labels
}).tolist().index(False)
mistaken_alpha = tf.argmax(alpha_values[index_first_mistaken_number, :])
mistaken_label = tf.argmax(test_labels[index_first_mistaken_number, :])
print('Accuracy:', accuracy.eval({inp: test_images, label: test_labels}))
print('The first mistaken number')
print('-------------------------')
print('Alpha:', mistaken_alpha.eval())
print('Label:', mistaken_label.eval())
mistaken_number = test_images[index_first_mistaken_number, :].reshape(
(28, 28))
plt.imshow(mistaken_number, cmap='gray')
with tf.Session() as sess:
alpha, E_1 = train_neural_network(sess)
evaluate_model(sess, alpha)
plt.figure('Test error')
plt.semilogy(np.linspace(0, T, len(E_1)), E_1, label='$E_1$')
plt.legend()
plt.show()
p = np.zeros(n)
for i in trange(N,desc="1"):
for j in range(n):
if random_walks[i,1] > 0:
if random_walks[i,j] <0:
p[j] += 1
break
if random_walks[i,1] < 0:
if random_walks[i,j] >0:
p[j] += 1
break
p = p / np.linalg.norm(p)
logt = np.log(t)[2:]
alpha = np.log(p[2:])
coeffs = np.polyfit(logt, alpha,1)
print(coeffs)
plt.figure()
plt.subplot(121)
plt.semilogy(p[2:], label = "Probability")
plt.legend()
plt.grid()
plt.subplot(122)
plt.plot(alpha, label = r"$\alpha$")
plt.plot(logt*coeffs[0] + coeffs[1], label = r"Linear fit. $\alpha = ${:.3f}".format(coeffs[0]))
plt.grid()
plt.legend()
plt.show()
s = StandardScaler().fit_transform(activity)
normalize = pd.DataFrame(data=s)
print(normalize.head())
#do the PCA
pca = PCA(n_components=3)
prin_Comp = pca.fit_transform(s)
prin_CompDf = pd.DataFrame(data=prin_Comp,
columns=['prin_comp1', 'prin_comp2', 'prin_comp3'])
prin_CompDf.head()
# Join the label to the data and un-comment 'for label data below'
# pca_data = pd.concat([prin_CompDf, activity[['0']]], axis = 1)
# print(pca_data.head(5))
# for no-label data
# normalize.to_csv('./datas/normalize.csv')
prin_CompDf.to_csv('./evaluate/thirtydays_feature.csv')
plt.semilogy(prin_CompDf, '--o')
plt.title('Feature after PCA')
plt.show()
# For label data
# # normalize.to_csv('./datas/normalize.csv')
# pca_data.to_csv('./data/cane_data/pca_dis.csv')
# plt.semilogy(pca_data, '--o')
# plt.title('Feature after PCA')
# plt.show()
means = []
low = []
up = []
for i in range(len(time)):
s = []
for j in range(len(count)):
s.append(size[j][i])
s_work = np.array(s)
means.append(np.mean(s_work))
low.append(np.percentile(s_work,5))
up.append(np.percentile(s_work,95))
time = np.arange(0,40+1,1)
plt.semilogy(time,means[0:len(time)],'o',color='C1')
plt.semilogy(t,I_plot,color='C1',linewidth=3)
plt.semilogy(t,I2_plot,color='C0',linewidth=3)
#plt.plot(t_det,inf_det/p_surv,linewidth=3)
#plt.plot(t,inf,linewidth = 3, color ='C0')
#plt.plot(time,low,color='C0',linewidth=3,alpha=0.5)/p_surv
#plt.plot(time,up,color='C0',linewidth=3,alpha=0.5)
plt.ylim((0,3*10000))
plt.xlim((0,40))
plt.fill_between(time,low[0:len(time)],up[0:len(time)],alpha=0.25,color='C1')
plt.tick_params(axis='both', which='major', labelsize=15, width=1, length=10)
plt.tick_params(axis='both', which='minor', labelsize=10, width=1, length=5)
plt.show()
mpl.legend(["data","psedo-interp"])#,"galerk-interp","pseudo-points"]) mpl.grid(True) mpl.show() mpl.plot(xs, funcExpanding ,xs,ypseudo)#, xs,ys,cpoints,pseudopoints) mpl.legend(["data","psedo-interp"])#,"galerk-interp","pseudo-points"]) mpl.grid(True) mpl.show() mpl.plot(xs, funcExpanding - ypseudo)#, xs,ys,cpoints,pseudopoints) mpl.legend(["Diff between data & pseudo-interp"])#,"galerk-interp","pseudo-points"]) mpl.grid(True) mpl.show() # mpl.plot(xs, d2pseudo, cpoints, d2pseudopoints )#, xs,ys,cpoints,pseudopoints) # mpl.legend(["data","d2psedo-interp"])#,"galerk-interp","pseudo-points"]) # #mpl.show() # #mpl.semilogx(rhoFunc.points,rhoFunc.data) # mpl.grid(True) # mpl.show() mpl.plot(cpoints,dpseudopoints,xs,dypseudo,realxs,fdeos) mpl.legend(["dpseudo-pts","dpseudo-interp","finite-difference"]) mpl.grid(True) mpl.show() mpl.semilogy(range(N),absolute(pseudofks),range(N), absolute(fks)) mpl.show()
for value in values:
allTransitions.add(value)
for index in allTransitions: # for example 0,1,2,5,6,9
possibles = []
percent = []
temp = []
for values in resultsAe:
try:
possibles.append(values[index][1]) # read possibles from slot 1
percent.append(values[index][0]) # read ratio from slot 0
temp.append(values[index][2]) # read temperature from slot 2
except KeyError:
pass
kbT = 1/kb/np.array(temp)
plt.semilogy(kbT, possibles, "-x", label=index)
# try to fit
if len(possibles) > 1:
try:
popt = curve_fit(f.exp, kbT, possibles, p0=[10e5, -0.01])
a = popt[0][0]
b = popt[0][1]
minusEnergy = b
plt.semilogy(kbT, f.exp(kbT, a,b), label="fit {0:.4g}e^{1:.4f}".format(a,b))
percentMean = np.mean(percent)
# index is the process, from x to y
x,y = mkl.getXy(index, len(energies))
if (verbose):
print("/",index, mkl.getXy(index, len(energies)))
print(percentMean,energies[x][y],minusEnergy)
# Use the same formulas above
N = x.shape[0]
for n in range(1, N):
w0 = np.append(w0, w)
x_predict[n, :] = x[n-1, :] * w
error[n, :] = x[n, :] - x_predict[n, :]
w = w + eta * error[n, :].T * x[n-1, :]
if iteration == 0:
error_full = np.copy(error)
else:
error_full = np.concatenate((error_full, np.copy(error)), axis=0)
# Reshape the final error
error_full = error_full.reshape((100, 5000))
# LMS learning curve: formula (3.63) in textbook where the learning-rate parameter eta is small
Jn = sigu2*(1-a**2)*(1+(eta/2)*sigu2) + sigu2*(a**2+(eta/2)*(a**2)*sigu2-0.5*eta*sigu2)*(1-eta*sigu2)**(2*t)
# LMS mean square error: formula (under formula (3.62) in textbook)
J_mean = np.mean(np.square(error_full), axis=0)
# Plot the desired results
plt.semilogy(Jn, 'r--', label='Theory')
plt.semilogy(J_mean, 'k-', label='Experiment', linewidth=0.8)
plt.legend(loc="upper right")
plt.title('Learning-rate parameter eta = ' + str(eta))
plt.xlabel("Number of iterations, n")
plt.ylabel("MSE")
plt.show()
############################################################################
# %% SAVE MODELS
############################################################################
g_model.save('02-results/gen_model.h5')
d_model.save('02-results/dis_model.h5')
acgan_model.save('02-results/gan_model.h5')
e_model.save('02-results/enc_model.h5')
ae_model.save('02-results/ae_model.h5')
############################################################################
# %% PLOT LOSS CURVES
############################################################################
fig = mp.figure(figsize=(10,8))
mp.semilogy(np.array(loss)[:, [1, 2, 4, 5, 6]])
mp.xlabel('batch')
mp.ylabel('loss')
mp.legend(['D(w) loss', 'D(y) loss', 'D(G(w)) loss', 'D(G(y)) loss', 'E(X) loss'])
mp.savefig('02-results/loss.png')
mp.close()
np.save('02-results/loss.npy', np.array(loss))
############################################################################
# %% PLOT ACCURACY CURVES
############################################################################
fig = mp.figure(figsize=(10,8))
#mp.plot(np.array(acc))
N = 16
mp.plot(np.convolve(np.array(acc)[:,0], np.ones((N,))/N)[(N-1):-N])
def ptest(methods,
ivps,
tols=[1.e-1, 1.e-2, 1.e-4, 1.e-6],
verbosity=0,
parallel=False):
"""
Runs a performance test, integrating a set of problems with a set
of methods using a sequence of error tolerances. Creates a plot
of the error achieved versus the amount of work done (number of
function evaluations) for each method.
**Input**:
* methods -- a list of ODEsolver instances
Note that all methods must have error estimators.
* ivps -- a list of IVP instances
* tols -- a specified list of error tolerances (optional)
* parallel -- to exploit possible parallelization (optional)
**Example**::
>>> import nodepy.runge_kutta_method as rk
>>> from nodepy.ivp import load_ivp
>>> bs5=rk.loadRKM('BS5')
>>> myivp=load_ivp('nlsin')
>>> work,err=ptest(bs5,myivp)
"""
import matplotlib.pyplot as pl
pl.clf()
pl.draw()
# In case just one method is passed in (and not as a list):
if not isinstance(methods, list): methods = [methods]
if not isinstance(ivps, list): ivps = [ivps]
err = np.ones([len(methods), len(tols)])
work = np.zeros([len(methods), len(tols)])
for ivp in ivps:
if verbosity > 0: print("solving problem {}".format(ivp))
try:
exsol = ivp.exact(ivp.T)
except:
bs5 = rk.loadRKM('BS5') #Use Bogacki-Shampine RK for fine solution
lowtol = min(tols) / 100.
if verbosity > 0:
print(
'solving for "exact" solution with tol= {}'.format(lowtol))
t, u = bs5(ivp, errtol=lowtol, dt=ivp.dt0)
if verbosity > 0: print('done')
exsol = u[-1] + 0.
for imeth, method in enumerate(methods):
if verbosity > 0:
print('Solving with method {}'.format(method.name))
if parallel:
speedup = len(method) / float(method.num_seq_dep_stages())
else:
speedup = 1.
workperstep = len(method) - method.is_FSAL()
for jtol, tol in enumerate(tols):
t, u, rej, dt, errhist = method(ivp,
errtol=tol,
dt=ivp.dt0,
diagnostics=True,
controllertype='P')
if verbosity > 1: print('{} rejected steps'.format(rej))
err[imeth, jtol] *= np.max(np.abs(u[-1] - exsol))
#FSAL methods save on accepted steps, but not on rejected:
work[imeth, jtol] += (len(t) * workperstep +
rej * len(method)) / speedup
for imeth, method in enumerate(methods):
for jtol, tol in enumerate(tols):
err[imeth, jtol] = err[imeth, jtol]**(1. / len(ivps))
for imeth, method in enumerate(methods):
pl.semilogy(work[imeth, :],
err[imeth, :],
label=method.name,
linewidth=3)
pl.xlabel('Function evaluations')
pl.ylabel('Error at $t_{final}$')
pl.legend(loc='best')
pl.draw()
return work, err
infile = name
names.append(os.path.basename(name).replace(".txt", ""))
print('plotting ' + infile)
with open(infile, 'r') as csvfile:
Ns = []
Times = []
reader = csv.DictReader(csvfile, delimiter='\t')
for row in reader:
Ns.append(int(row['N']))
Times.append(float(row['time']))
plt.semilogy(Ns, Times)
pp = PdfPages('time.pdf')
#plt.axis([0, 2048, .01, 24])
plt.legend(names, loc='upper left', fontsize='small')
plt.title('Euclidean Norm Computation')
plt.xlabel('Problem Size')
plt.ylabel('Execution Time')
plt.xticks([22, 23, 24, 25, 26, 27], [22, 23, 24, 25, 26, 27])
pp.savefig()
pp.close()
plt.show()
return result[0]
plt.figure
for j in range(14):
v = v0
if (j != 0):
v[j - 1] = [0]
v[j] = [1]
#print(v)
# compute their product
u = np.dot(A, v)
# diff records the difference of Av and v
diff = [abssum(u, v)]
i = 0
while diff[i] > 1e-5:
v = u # record A^kv
u = np.dot(A, v) # compute A^{k+1}v
diff.append(abssum(u, v)) # append add
i = i + 1
print(vecsum(u))
#plot the differences with iterations
#plt.subplot(7,2,j+1)
plt.semilogy(range(len(diff)), diff, label='vector ' + str(j))
plt.xlabel('iterations')
plt.ylabel('difference')
plt.title('Convergence')
plt.legend()
plt.show()
def get_completeness_adjusted_table(catalogue, completeness, dmag,
offset=1.0E-5, end_year=None, plot=False,
figure_size=(8, 6), filename=None,
filetype='png', dpi=300, ax=None):
"""
Counts the number of earthquakes in each magnitude bin and normalises
the rate to annual rates, taking into account the completeness
"""
if not end_year:
end_year = catalogue.end_year
# Find the natural bin limits
mag_bins = _get_catalogue_bin_limits(catalogue, dmag)
obs_time = end_year - completeness[:, 0] + 1.
obs_rates = np.zeros_like(mag_bins)
durations = np.zeros_like(mag_bins)
n_comp = np.shape(completeness)[0]
for iloc in range(n_comp):
low_mag = completeness[iloc, 1]
comp_year = completeness[iloc, 0]
if iloc == (n_comp - 1):
idx = np.logical_and(
catalogue.data['magnitude'] >= low_mag - offset,
catalogue.data['year'] >= comp_year)
high_mag = mag_bins[-1]
obs_idx = mag_bins >= (low_mag - offset)
else:
high_mag = completeness[iloc + 1, 1]
mag_idx = np.logical_and(
catalogue.data['magnitude'] >= low_mag - offset,
catalogue.data['magnitude'] < (high_mag - offset))
idx = np.logical_and(mag_idx,
catalogue.data['year'] >= (comp_year - offset))
obs_idx = np.logical_and(mag_bins >= (low_mag - offset),
mag_bins < (high_mag + offset))
temp_rates = np.histogram(catalogue.data['magnitude'][idx],
mag_bins[obs_idx])[0]
temp_rates = temp_rates.astype(float) / obs_time[iloc]
obs_rates[obs_idx[:-1]] = temp_rates
durations[obs_idx[:-1]] = obs_time[iloc]
selector = np.where(obs_rates > 0.)[0]
mag_bins = mag_bins[selector]
obs_rates = obs_rates[selector]
durations = durations[selector]
# Get cumulative rates
cum_rates = np.array([sum(obs_rates[iloc:])
for iloc in range(0, len(obs_rates))])
if plot:
plt.figure(figsize=figure_size)
plt.semilogy(mag_bins + dmag / 2., obs_rates, "bo",
label="Incremental")
plt.semilogy(mag_bins + dmag / 2., cum_rates, "rs",
label="Cumulative")
plt.xlabel("Magnitude (M)", fontsize=16)
plt.ylabel("Annual Rate", fontsize=16)
plt.grid(True)
plt.legend(fontsize=16)
if filename:
plt.savefig(filename, format=filetype, dpi=dpi,
bbox_inches="tight")
return np.column_stack([mag_bins, durations, obs_rates, cum_rates,
np.log10(cum_rates)])
Hyee = h.initial(xx)
Hyee[0] = 0
Hyee[-1] = 0
t0yee = time.time()
for j in range(m):
Hyee[1:-1] = Hyee[1:-1] + s * (Eyee[2:] - Eyee[:-2])
Hyee[0] = Hyee[1]
Hyee[-1] = Hyee[-2]
Eyee[1:-1] = Eyee[1:-1] + s * (Hyee[2:] - Hyee[:-2])
tout[2, k] = time.time() - t0yee
#print(dxs)
#print(tout.T)
plt.plot(dxs, tout.T)
plt.legend(('RK45', 'DOP853', 'FDTD'))
plt.xlabel('dx')
plt.ylabel('computation time [s]')
plt.title('T=' + str(T))
plt.savefig('testresults/timing/MOL2yeespacing.pdf')
plt.clf()
plt.semilogy(dxs, tout.T, '.')
plt.legend(('RK45', 'DOP853', 'FDTD'))
plt.xlabel('dx')
plt.ylabel('computation time [s]')
plt.title('T=' + str(T))
plt.savefig('testresults/timing/MOL2yeespacingsemilog.pdf')
plt.clf()
ram_prices = pd.read_csv("data/ram_price.csv")
# use historical data to forecast prices after the year 2000
data_train = ram_prices[ram_prices.date < 2000]
data_test = ram_prices[ram_prices.date >= 2000]
# predict prices based on date
X_train = data_train.date[:, np.newaxis]
# we use a log-transform to get a simpler relationship of data to target
y_train = np.log(data_train.price)
tree = DecisionTreeRegressor().fit(X_train, y_train)
linear_reg = LinearRegression().fit(X_train, y_train)
# predict on all data
X_all = ram_prices.date[:, np.newaxis]
pred_tree = tree.predict(X_all)
pred_lr = linear_reg.predict(X_all)
# undo log-transform
price_tree = np.exp(pred_tree)
price_lr = np.exp(pred_lr)
plt.semilogy(data_train.date, data_train.price, label="Training data")
plt.semilogy(data_test.date, data_test.price, label="Test data")
plt.semilogy(ram_prices.date, price_tree, label="Tree prediction")
plt.semilogy(ram_prices.date, price_lr, label="Linear prediction")
plt.legend()
plt.show()
#pl.show()
pl.clf()
pl.yscale('log')
pl.hist(BinVar,
BinNumb,
log=True,
histtype='stepfilled',
color=hex2color('#0000FF'),
linewidth=0.2)
pl.ylim([-0.75, (np.max(BinHists)) * 10**0.35])
if type(FitParam) is dict and len(FitRange) > 0:
pl.semilogy(
BinCents[FitRange],
FitGauss,
color=hex2color('#FF0000'),
linewidth=3,
linestyle='solid'
) # -- You simply need to use semilogy instead of plot -- http://stackoverflow.com/questions/773814/plot-logarithmic-axes-with-matplotlib-in-python
pl.text(FitParam['mu'] + 1.0 * FitParam['sigma'],
FitParam['A'],
'sigma = %.10g' % (FitParam['sigma']),
color=hex2color('#FF0000'),
fontsize=18)
pl.xlabel("Pixel Value")
pl.ylabel("log N")
#
# Save eps
#
def test_case():
np.random.seed(42)
print(
"Length of Nvals = number of iternations, with 100 sample points added per step. i.e. setting up the function for 10 iterations of 100 sample points per iteration"
)
Nvals = np.array([100, 100, 100, 100, 100, 100, 100, 100, 100, 100])
num_Ns = len(Nvals)
print(
"Length of Nvals becomes also the number of iteration steps (= 10) used in the model"
)
max_errs = np.inf * np.ones((num_Ns, 3))
max_errs_gp = np.inf * np.ones(num_Ns)
avg_errs_gp = np.inf * np.ones(num_Ns)
print(
"construct test points, setting N = 100 test points and D = 8 dimensions"
)
X_test = randOmega(100, 8)
f_test = test_function(X_test.T)
for i, N in enumerate(Nvals):
# training points
X = randOmega(int(N), 8)
V = test_function(X.T)
G = dtest_function(X.T)
CN = (G.T @ G) / N
vals, vecs = linalg.eigh(CN)
for d in range(3, 0, -1):
# find active subspace of dimension d
W = vecs[:, -d:]
Y = X @ W
# fit GP on active subspace
gp_as = GaussianProcessRegressor(RBF(), n_restarts_optimizer=8)
gp_as.fit(Y.reshape(N, d), V)
m_tilde = gp_as.predict((X_test @ W).reshape(100, d))
max_errs[i, d - 1] = np.max(np.abs(f_test - m_tilde))
gp = GaussianProcessRegressor(RBF())
gp.fit(X, V)
m_tilde = gp.predict(X_test)
# Calculating the max error for the GP
max_errs_gp[i] = np.max(np.abs(f_test - m_tilde))
#Calculating the same for avg error according to the formula in part b), and using these values for part b)
avg_errs_gp[i] = (1 / N) * np.sum(np.abs(f_test - m_tilde))
#Adjusting the plot to fit 8 dimensions
fig, ax = plt.subplots(1, 2, figsize=(9, 5))
x = np.arange(1, 9)
ax[0].semilogy(x, vals, ".", ms=15)
ax[0].set_xlabel("Eigenvalues")
ax[0].set_xticks(x)
ax[0].set_ylabel("$\lambda$")
ax[1].semilogy(Nvals, max_errs, ".-", ms=15)
ax[1].semilogy(Nvals, max_errs_gp, ".-", ms=15)
ax[1].legend([str(i) + "d AS" for i in range(1, 4)] + ["Full GP"])
ax[1].set_xlabel("# of points")
fig.tight_layout()
plt.show()
print(
"b) Plotting the average and maximum errors according to the formulae in part b:"
)
plt.semilogy(Nvals, max_errs_gp, ".-", ms=15)
plt.title("Maximum errors")
plt.xlabel("# of points")
plt.show()
plt.semilogy(Nvals, avg_errs_gp, ".-", ms=15)
plt.title("Average errors")
plt.xlabel("# of points")
plt.show()
print(
"As expected we get a vertical line. This line shows us the difference in errors generated by the model when we run the model 10 times over the same amount of sample points"
)
print(
"is the gradient vector, hence, we plot and put out the value for the gradient at the 10th iteration step"
)
#
plt.plot(x, G[10], ".", ms=15)
plt.xlabel("Dimension")
plt.ylabel("Partial derivative")
plt.show()
print("The gradient vector for the 10th generation: ")
print(G[10])
print(
"For this exercise we utilise the partial derivative at the 10th iteration, where G is the matrix of partial derivatives\
Then we construct a matrix C whose eigenvalues can help us decided the active subspaces in the 10th iteration. We see\
From our plot that the active subspace is present at the 2nd and 3rd dimension for the 10th iteration."
)
return fig, max_errs
import matplotlib.pyplot as plt
import numpy as np
import sys
import seaborn as sns
sns.set()
if __name__ == '__main__':
depth = int(sys.argv[1])
g = 1.4
h = 0.9045
for idx in range(1, 21):
try:
T = idx * 0.5
plt.semilogy(np.load(
'data/1d_TFI_g%.4f_h%.4f/L31/T%.1f/circuit_depth%d_Niter100000_1st_error.npy'
% (g, h, T, depth)),
label='T=%.1f' % T)
print(
'plot T %.1f' % T,
np.load(
'data/1d_TFI_g%.4f_h%.4f/L31/T%.1f/circuit_depth%d_Niter100000_1st_error.npy'
% (g, h, T, depth))[0])
except Exception as e:
print(e)
plt.legend()
plt.show()
# T = 0.2
# T = float(sys.argv[1])
# for idx in range(6):
# for mag in [7,8,9,10]:
# for bmv in [1.0, 1.4]:
cols = ['violet', 'b', 'r']
mags = [4.7, 6.53, 10.65]
bmvs = [0.77, 0.68, 0.996]
for i in range(0, len(mags)):
uncs = np.linspace(1, 10)
mag = mags[i]
bmv = bmvs[i]
if bmv > 1.2:
stype = "M"
i2counts = ds.getI2_M(uncs)
else:
stype = "K"
i2counts = ds.getI2_K(uncs)
exp_counts = ds.getEXPMeter(i2counts, bmv)
exp_time = ds.getEXPTime(i2counts, mag, bmv, STAR_EL, AVG_FWHM)
# print "%s %4.1f %3.1f %7.0f %.1g" % (stype,mag,precision,exp_time,exp_counts)
plt.semilogy(uncs, exp_time, c=cols[i])
plt.ylabel("Exposure Time (sec)")
plt.xlabel("Desired Precision m/s")
plt.ylim(1, 2500)
plt.show()
iterAMP2 = 4
xhat2 = AMP(y, H, sigma2, sigmas2, iterAMP2, m, n)
# APPROXIMATE MESSAGE 2
errors['AMP2'].append(np.sum(x != xhat2))
x_mmse = np.dot(
inv(sigma2 / sigmas2 * np.eye(n) + np.dot(prim(H), H)),
np.dot(prim(H), y))
x_mmse = np.sign(np.real(x_mmse)) + 1.j * np.sign(np.imag(x_mmse))
errors['MMSE'].append(np.sum(x != x_mmse))
count = count + 1
meta_errors['AMP1'][0].append(np.mean(errors['AMP1']))
meta_errors['AMP1'][1].append(np.std(errors['AMP1']))
meta_errors['AMP2'][0].append(np.mean(errors['AMP2']))
meta_errors['AMP2'][1].append(np.std(errors['AMP2']))
meta_errors['MMSE'][0].append(np.mean(errors['MMSE']))
meta_errors['MMSE'][1].append(np.std(errors['MMSE']))
import matplotlib.pyplot as plt
plt.semilogy(SNRrange, meta_errors['AMP1'][0], 'r')
plt.semilogy(SNRrange, meta_errors['AMP2'][0], 'b')
plt.semilogy(SNRrange, meta_errors['MMSE'][0], 'k')
plt.title('Approximate Message Passing')
plt.xlabel('Signal to Noise [db]')
plt.ylabel('Mean meta error')
plt.show()
Y10[i] *= 1000
Y11[i] *= 1000
# Normalize by number of cells
Y8[i] /= Y1[i]
Y9[i] /= Y1[i]
Y10[i] /= Y1[i]
Y11[i] /= Y1[i]
#print (X1[i]*pow(2,i+1) * X1[i]*pow(2,i+1))/Y1[i]
fig = pyplot.figure()
fig.patch.set_facecolor('white')
pyplot.xlabel("Compressibility", fontdict={'fontsize': 16})
pyplot.ylabel("Runtime/numcells (millisecs/cell)", fontdict={'fontsize': 16})
pyplot.semilogy()
pyplot.plot(Y2,
Y8,
color='green',
linestyle='-',
linewidth=1.0,
marker='D',
label='Full Perfect on GPU')
pyplot.plot(Y2,
Y9,
color='orange',
linestyle='-',
linewidth=1.0,
marker='D',
label='7 write, 1 read on GPU')
