def p_explore(dd):
''' plot the results of f_explore'''
nt = dd.shape[0]
nx = dd.shape[1]
xx = ()
yy = ()
for i in range(nx):
xx = xx + (dd[:, 0, i, 0], )
yy = yy + (dd[:, 1, i, 0], )
xx = xx + (dd[:, 0, i, 1], )
yy = yy + (dd[:, 1, i, 1], )
pd = pt.Paradraw()
pd.x_label = '$x_1$'
pd.y_label = '$x_2$'
pd.marks = ['k']
pt.multiplot2(xx, yy, pd)
return xx, yy
def plot_clusters(w1, w2):
ind0 = np.where(labels == 0)[0]
ind1 = np.where(labels == 1)[0]
ind2 = np.where(labels == 2)[0]
vocab_list = tfidf_vocab.tolist()
try:
w1 = int(w1)
indword1 = w1
except:
#w1 is a string, not an integerer
index1 = vocab_list.index(w1)
indword1 = np.where(tfidf_index_sorted_weights == index1)[0][0]
#
try:
w2 = int(w2)
indword2 = w2
except:
#w2 is a string, not an integerer
index2 = vocab_list.index(w2)
indword2 = np.where(tfidf_index_sorted_weights == index2)[0][0]
print(indword1, indword2)
#
pd = pt.Paradraw()
pd.title = tfidf_vocab[tfidf_index_sorted_weights[
indword1]] + ' ' + tfidf_vocab[tfidf_index_sorted_weights[indword2]]
pd.x_label = tfidf_vocab[tfidf_index_sorted_weights[indword1]]
pd.y_label = tfidf_vocab[tfidf_index_sorted_weights[indword2]]
pd.colors = ['r', 'g', 'b']
pd.markers = ['.', '.', '.']
pd.marks = ['', '', '']
pd.markeredge = True
pd.markerssize = [8, 8, 8]
pd.thickness = [18, 18, 18]
pd.legend = ['cluster 1', 'cluster 2', 'cluster 3']
datax = [
most_sig_array[:, indword1][ind0], most_sig_array[:, indword1][ind1],
most_sig_array[:, indword1][ind2]
]
datay = [
most_sig_array[:, indword2][ind0], most_sig_array[:, indword2][ind1],
most_sig_array[:, indword2][ind2]
]
pt.multiplot2(datax, datay, pd)
def graph_distrib(graph, plot=False, kmax=None):
if kmax is None:
k = np.arange(np.max(graph) + 1)
else:
k = np.arange(kmax + 1)
pk = np.zeros(len(k))
for g in graph:
if kmax is None:
pk[g] += 1
else:
if g < kmax + 1:
pk[g] += 1
k = k[1:]
pk = pk[1:]
pk = 1. * pk / np.sum(pk)
pk[pk == 0.] = np.min(pk[np.nonzero(pk)]) / 10.
A = np.ones((len(k) - 2, 2))
A[:, 1] = k[1:-1]
logpk = np.log(pk[1:-1])
mloga, lambda_ = -np.linalg.lstsq(A, logpk)[0][:]
a = np.exp(-mloga)
print('lambda approx', lambda_)
if plot is True:
import plot_tools
plot_tools.multiplot2(
(k, k), (pk, a * np.exp(-lambda_ * k)),
plot_tools.Paradraw(
marks=['-', '--'],
markers=['o', ''],
y_scale='log',
x_label='k',
y_label='P(k)',
ax_x_format='%.0f',
legend=['', '$\lambda = ' + '{0:.2f}'.format(lambda_) + '$']))
return k, pk, lambda_
def plot(self, style='z'):
import plot_tools as pt
pd = pt.Paradraw()
pd.colors = pt.get_colors(self.n_species)
pd.x_scale = 'symlog'
pd.y_scale = 'symlog'
pd.x_label = 't'
if style == 'rates':
pd.y_label = 'rates'
zs = [self.ts_rates(i)[1:] for i in range(self.n_species)]
else:
pd.y_label = 'z'
zs = [self.ts_Z(i)[1:] for i in range(self.n_species)]
time = (self.time[1:], ) * self.n_species
figs = pt.multiplot2(time, zs, pd)
return figs
def plot_mech(inputs,reaction=None,xlim=None):
ploc = inputs['ploc']
outputs = inputs['outputs']
#
if reaction is None:
if 'main_reaction' not in outputs.keys():order_observability(inputs)
reaction = outputs['main_reaction']
#time
tmin,tmax = ploc['tmin'],ploc['tmax']
n_s = reaction.n_species
tmini = int(np.floor(np.log10(tmin)))
tmaxi = int(np.ceil(np.log10(tmax)))
#main parameters
pd = pt.Paradraw()
if n_s<10:pd.legend = [reaction.list[i] for i in xrange(n_s)]
pd.x_scale = 'log'
if xlim is None:
pd.xlim = [tmin,tmax]
pd.x_tick_label = [1.*10**i for i in xrange(tmini+1,tmaxi+1,2)]
else: pd.xlim = xlim
pd.x_label = 'time'
import matplotlib.colors as colors
pd.colors = []
for ic,c in enumerate(colors.cnames):
if ic<n_s:pd.colors.append(str(c))
# pd.colors = ['tan','g','r','b','k','y','magenta','aqua']
cc = [reaction.time_series('concentrations',i) for i in xrange(n_s)]
tt = (reaction.time,)*n_s
pd.y_label = 'n'
pd.ylim = [np.min(cc),np.max(cc)]
pt.multiplot2(tt,cc,pd)
#
zz = [reaction.time_series('z',i) for i in xrange(n_s)]
pd.y_scale = 'log'
pd.y_label = 'z'
pd.ylim = [np.min(zz),np.max(zz)]
pt.multiplot2(tt,zz,pd)
#
pd.y_scale = 'symlog'
pd.y_label = 'p rate'
pp = [reaction.time_series('rates',i) for i in xrange(n_s)]
pd.ylim = [np.min(pp),np.max(pp)]
pt.multiplot2(tt,pp,pd)