def main(**kwargs):
v = {-1, +1, +2} # Set of votes.
n = 5000 # Number of iterations.
g = three_leaders(v)
measures = {j: [] for j in v.union({0})}
# Main cycle.
for i in range(n):
simulation.iteration(g, v)
for j in v.union({0}):
measures[j].append(formulae.occupation_measure(g, j))
# Log.
if 'log' in kwargs:
kwargs['log'](i, g, v)
# Plot.
if kwargs.get('plot', False):
colours = {-1: 'r', 0: 'g', +1: 'b', +2: 'y'}
cesaro = {x: formulae.cesaro(y) for x, y in measures.items()}
plot.lines(measures, colours)
plot.lines(cesaro, colours)
plot.show()
def plot_boot(data, step=10, tag=''):
N = min([len(boot[k][st]) for k in boot.keys()])
data = {k: {} for k in labels[5:]}
for j,k in enumerate(labels[5:]):
for ex, b in boot.items():
# data[k][ex] = [min(np.abs(b[i][j][2]-b[i][j][1]),1.) for i in range(N)]
data[k][ex] = [min(b[st][j],1.) for i in range(N)]
for k, v in data.items():
P.lines(v, 'Conv-%d-%s-%s'%(st,tag,k), step=step, xlabel='Simulation Time (in ns)')
def plotsep(boot, step=10, tag=''):
N = min([len(v) for v in boot.values()])
data = {k: {} for k in labels}
for j,k in enumerate(labels):
for ex, b in boot.items():
# data[k][ex] = [min(np.abs(b[i][j][2]-b[i][j][1]),1.) for i in range(N)]
data[k][ex] = [min(b[i][1][j],1.) for i in range(N)]
for k, v in data.items():
P.lines(v, 'Conv-%s-%s'%(tag,k), step=step, xlabel='Simulation Time (in ns)')
def graphexprB(elist, step=5, tw=True):
obsbin = {name: getobs(name) for name in elist}
if tw:
obslist = {k: obs2tw(v) for k, v in obsbin.items()}
else:
obslist = obsbin
convlist = {k: calc_conv(v, step=step*1000) for k,v in obslist.items()}
series = {}
for k, v in convlist.items():
for tw, data in v.items():
if tw not in series:
series[tw] = {}
series[tw][k] = data
for k, v in series.items():
P.lines(v, 'convB-'+k, step=step)
def plotm(boot, step=10, tag=''):
N = min([len(v) for v in boot.values()])
merge = {k: [np.mean([min(1., boot[k][i][1][f]) for f in range(10, 20)]) for i in range(N)] for k in boot.keys()}
P.lines(merge, 'Conv-%s-CI-MERGE'%tag, step=step, xlabel='Simulation Time (in ns)')
merge = {k: [np.mean([min(1., boot[k][i][2][f]) for f in range(10, 20)]) for i in range(N)] for k in boot.keys()}
P.lines(merge, 'Conv-%s-STD-MERGE'%tag, step=step, xlabel='Simulation Time (in ns)')
merge = {k: [np.mean([min(1., boot[k][i][3][f]) for f in range(10, 20)]) for i in range(N)] for k in boot.keys()}
P.lines(merge, 'Conv-%s-Err-MERGE'%tag, step=step, xlabel='Simulation Time (in ns)')
merge = {k: [np.mean([min(1., boot[k][i][4][f]) for f in range(10, 20)]) for i in range(N)] for k in boot.keys()}
P.lines(merge, 'Conv-%s-CiMu-MERGE'%tag, step=step, xlabel='Simulation Time (in ns)')
def graphexprC(elist, step=5, tw=True):
obsbin = {name: getobs(name) for name in elist}
if tw:
obslist = {k: obs2tw(v) for k, v in obsbin.items()}
else:
obslist = obsbin
seriesList = getdistinct(obslist[elist[0]])
series = {k: {} for k in seriesList}
for name in elist:
print('Bootstrapping on ', name)
for k in seriesList:
series[k][name] = []
N = len(obslist[name])
for n in range(0, N, step*1000):
c = bootstrap_iter(obslist[name][:min(n,N)], size=step)
for k in seriesList:
if k in c.keys():
series[k][name].append(min(c[k][1]/c[k][0], 1.))
else:
series[k][name].append(1.)
for k, v in series.items():
P.lines(v, 'convC-'+k, step=step)
LB={}
LB['Serial'] = np.load(home+'/work/results/serial_labels.npy').reshape(1, 270402)
LB['Parallel'] = np.array([i[:15000] for i in np.load(home+'/work/results/denparallel_labels.npy')])
LB['Biased'] = np.load(home+'/work/results/denbias_labels.npy')
LB['Uniform'] = np.load(home+'/work/results/denuniform_labels.npy')
LB['Lattice'] = np.load(home+'/work/results/denlattice_labels.npy')
# Process and make graph
fsize = (TEXT_WIDTH, STD_HGT)
LBA = {k: [activity1(t, y+2, 50) for t in LB[k]] for y, k in enumerate(EXP_NAMES)}
idxlist = {'Serial': [0], 'Parallel': [2,3,4], 'Biased': [0,3,11,17,23,28,31,32,33,35], 'Uniform':list(range(10)), 'Lattice': [12,16,27,34,41,42,45,46,52,57]}
X = {k: np.array(list(it.chain(*[LBA[k][i] for i in v]))) for k,v in idxlist.items()}; xmax = min([len(v) for v in X.values()])
imp.reload(P)
P.lines({k: v[:xmax] for k,v in X.items()}, showlegend=False, labels=EXP_NAMES[::-1], lw=1, yticks=EXP_NAMES + [' '], \
fname='denovo_activity', ylim=(.85, 6.25), figsize=fsize, no_ytick=True, colors=EXP_COLORS[::-1], \
xlabel='Simulation Time (normalized)', xticks=['%d'%i for i in range(0, 101, 10)], latex=True)
# idxlist = {'Serial': [0], 'Parallel': [1], 'Biased': [0,3,11,17,23,28,31,32,33,35], 'Uniform':[10,16,18,21,23,24,25,28,33,35], 'Lattice': [1,15,18,29,31,32,33,40,41,42]}
################
# TRADEOFF
# data,totc = [],[]; expname='biased6'
# with open(home+'/work/results/{0}_prov.log'.format(expname)) as provfile:
# for line in provfile.read().strip().split('\n'):
# if line.startswith('BASIN'):
elif plot_type == 'lp':
i=1
while True:
aux = [point for point in input if point.l==i]
if not len(aux):
break
aux.sort(key=lambda x:x.p)
aux = [point.file for point in aux]
flist.append(aux)
i=i+1
for l in flist:
av,e = generate_data(l, field, conf)
ylist.append(av)
error.append(e)
# ylist,error=generate_data(input, field, conf)
# plot.line(ylist, error, output, title, xlabel, ylabel)
#elif plot_type == 'lp':
if not legend:
warning("labels for lines not entered. Default labels will be used.")
if len(ylist) == 1:
plot.line(ylist[0], error[0], output, title, xlabel, ylabel)
elif len(ylist) > 1:
plot.lines(ylist, error, output, title, ylabel, legend)
else:
fail(plot_type+"is not a valid plot type", 1)
#verificar se grafico foi gerado
