def Emeanvar(q, mode_nums=None):
"""
returns 2 dictionaries:
key, value = modeNo, mean{E}
key, value = modeNo, stdv{E}
"""
if not mode_nums:
mode_nums = range(len(q))
E = nm_s.compute_E(q)
n = len(E[0])
Em = [1.0*sum(Ei)/n for Ei in E]
Es = [(1.0*sum((Ei-Eim)**2)/(n-1))**0.5 for Ei, Eim in zip(E, Em)]
return dict( (modeNo, Em[modeNo]) for modeNo in mode_nums ), dict( (modeNo, Es[modeNo]) for modeNo in mode_nums )
def plot(t_P, q, system, gens, mode_nums=None, verbose=False, tc="x", xmin=False, xmax=False, ymin=False, ymax=False):
""" a wrapper for plotting routines we'll use a lot """
n_l_m = system.network.nlm
ans = []
### amp-time
if verbose:
print "amp_time"
fig, ax = nmp.amp_plot(t_P, q, n_l_m=n_l_m, mode_nums=mode_nums)
ax.set_ylabel(r"$|" + tc + "_i|$")
ax.set_xlabel(r"$t/P_{\mathrm{orb}}$")
if ymin:
ax.set_ylim(ymin=ymin)
if ymax:
ax.set_ylim(ymax=ymax)
if xmin:
ax.set_xlim(xmin=xmin)
if xmax:
ax.set_xlim(xmax=xmax)
ans.append((fig, ax))
### multi-gen-amp-time
if verbose:
print "multi-gen_amp_time"
fig, axs = nmp.multi_gen_amp_plot(t_P, q, gens, n_l_m=n_l_m, mode_nums=mode_nums)
for ax in axs:
ax.set_ylabel(r"$|" + tc + "|$")
ax.set_xlabel(r"$t/P_{\mathrm{orb}}$")
if ymin:
ax.set_ylim(ymin=ymin)
if ymax:
ax.set_ylim(ymax=ymax)
if xmin:
ax.set_xlim(xmin=xmin)
if xmax:
ax.set_xlim(xmax=xmax)
ans.append((fig, axs))
### coupling diagrams
colormap = nmd.plt.get_cmap("jet")
### nl-placement disp
if verbose:
print "nl-placmenet disp"
myE = -np.array([np.mean(_) for _ in nms.viscous_disp(q, network, Eo=1.0)[-1]])
mode_colors = [colormap(_) for _ in myE]
mode_order = myE.argsort()
fig = nmd.plt.figure()
ax = fig.add_axes([0.1, 0.1, 0.750, 0.8])
cbax = fig.add_axes([0.875, 0.1, 0.025, 0.8])
fig, ax = nmd.coupling_tree_nl(
system,
tree_type="placement",
genNos=[],
verbose=verbose,
mode_colors=mode_colors,
mode_order=mode_order,
fig_ax=(fig, ax),
mode_nums=mode_nums,
)
# colorbar = nmd.matplotlib.colorbar.ColorbarBase(cbax, cmap=colormap, orientation='vertical', norm=nmd.matplotlib.colors.LogNorm())
colorbar = nmd.matplotlib.colorbar.ColorbarBase(cbax, cmap=colormap, orientation="vertical")
colorbar.ax.set_ylabel(r"$\gamma_i A_i^2/\mathrm{max}\{\gamma_i A_i^2\}$")
ans.append((fig, ax, colorbar))
### nl-placement E
if verbose:
print "nl-placement E"
mE = np.array([np.mean(_) for _ in nms.compute_E(q, Eo=1.0)])
mode_colors = [colormap(_) for _ in mE / max(mE)]
mode_order = mE.argsort()
fig = nmd.plt.figure()
ax = fig.add_axes([0.1, 0.1, 0.750, 0.8])
cbax = fig.add_axes([0.875, 0.1, 0.025, 0.8])
fig, ax = nmd.coupling_tree_nl(
system,
tree_type="placement",
genNos=[],
verbose=verbose,
mode_colors=mode_colors,
mode_order=mode_order,
fig_ax=(fig, ax),
mode_nums=mode_nums,
)
# colorbar = nmd.matplotlib.colorbar.ColorbarBase(cbax, cmap=colormap, orientation='vertical', norm=nmd.matplotlib.colors.LogNorm())
colorbar = nmd.matplotlib.colorbar.ColorbarBase(cbax, cmap=colormap, orientation="vertical")
colorbar.ax.set_ylabel(r"$A_i^2/\mathrm{max}\{A_i^2\}$")
ans.append((fig, ax, colorbar))
### nl-siblings gens
for genNo in xrange(1, len(gens)):
if verbose:
print "nl-siblings gen%d" % genNo
fig, ax = nmd.coupling_tree_nl(
system, tree_type="siblings", genNos=[genNo], verbose=verbose, mode_nums=mode_nums
)
ans.append((fig, ax))
return ans
for collE in opts.collE:
if opts.verbose:
print "\tcollE:", collE
collE_modeNos[collE] = [
[network.modeNoD[nlms] for nlms in _]
for _ in pn.downselect_collE(collE, system, freqs=None, mode_nums=None)
]
# =================================================
### plot data!
# =================================================
### compute stuff for statistics
myE = -np.array([np.mean(_) for _ in nms.viscous_disp(q, network, Eo=1.0)[-1]])
myE /= sum(myE)
mE = np.array([np.mean(_) for _ in nms.compute_E(q, Eo=1.0)])
mE /= sum(mE)
pkl_obj = {"myE": myE, "mE": mE}
# === total data set
if opts.plot:
if opts.verbose:
print "total data"
tag = "" + opts.tag
savefig(
plot(t_P, q, system, gens, mode_nums=None, verbose=opts.vverbose, tc=opts.tc),
opts.outfilename,
opts.logfilename,
tag=tag,
no_legend=opts.no_legend,
if multi_gen:
if opts.verbose: print "determining generational structure"
gens, _ = system.network.gens()
####################################################################################################
#
#
# stacked histograms
#
#
####################################################################################################
if opts.stacked_hist:
if opts.verbose: print "\tstacked_hist"
### prepare data
E = nms.compute_E(q, Eo=1.)
mE = [nms.sample_mean(e) for e in E]
# sE = [nms.sample_var(e, xo=mE[ind])**0.5 for ind, e in enumerate(E)]
mdE = [2*network.wyU[ind][1]*e for ind, e in enumerate(mE)]
# sdE = [2*network.wyU[ind][1]*e for ind, e in enumerate(sE)]
data = [(1, len(network.K[ind]), mE[ind], mdE[ind]) for ind in range(N_m)] # number of modes, number of connections, mean Energy, mean dissipation
### make and save each plot
for xvar, bin_width in opts.stacked_hist:
#==========
if xvar == "w":
if opts.verbose: print "\t\tw"
if not bin_width:
if not len(t_P):
raise ValueError, "no data loaded from "+opts.filename
if opts.amplitude:
if opts.verbose: print "computing mode amplitudes"
x = nm_s.compute_A(q, Eo=1.)
mA = []
sA = []
lA = []
for A in x:
mA.append( nm_s.sample_mean(A) )
sA.append( nm_s.sample_var(A, xo=mA[-1])**0.5 )
lA.append( A[-1] )
del x
x = nm_s.compute_E(q, Eo=1.)
mE = []
sE = []
lE = []
totE = [sum([x[modeNo][i] for modeNo in range(N_m)]) for i in range(len(t_P))]
mean_tot_E = nm_s.sample_mean(totE)
var_tot_E = nm_s.sample_var(totE, xo=mean_tot_E)
for E in x:
mE.append( nm_s.sample_mean(E) )
sE.append( nm_s.sample_var(E, xo=mA[-1])**0.5 )
lE.append( E[-1] )
del x, totE
if opts.conc_fit:
if opts.verbose: print "\tattempting to fit concentration diagram"
mE_fp, mE_cvr, mE_rchi2 = nm_s.broken_PowLaw_fitter(range(opts.conc_fit_start+1, opts.conc_fit_stop+1), np.array(sorted(mE)[opts.conc_fit_start:opts.conc_fit_stop])/mean_tot_E, np.ones((opts.conc_fit_stop-opts.conc_fit_start)), max_iters=opts.conc_fit_iters, rtol=opts.conc_fit_rtol, verbose=opts.conc_verbose)
