g = numpy.ones(2)
print "statistical inefficiencies taken from input options and are %.3f and %.3f steps" % (
options.efficiency[0], options.efficiency[1])
if (options.useg == 'subsample'):
readmdfiles.subsample(N_k, U_kn, V_kn, N_kn, g, type)
else:
print "statistical efficiencies used to scale the statistical uncertained determined from all data"
figname = options.figname
title = options.figname
checkensemble.ProbabilityAnalysis(N_k,
type=analysis_type,
T_k=T_k,
P_k=P_k,
mu_k=mu_k,
U_kn=U_kn,
V_kn=V_kn,
N_kn=N_kn,
title=title,
figname=figname,
nbins=nbins,
reptype='bootstrap',
g=g,
nboots=nboots,
bMaxwell=(type == 'kinetic'),
bLinearFit=bLinearFit,
bNonLinearFit=bNonLinearFit,
bMaxLikelihood=bMaxLikelihood,
seed=options.seed,
kB=kB)
p = numpy.exp(beta_k[k] * mu_k[k]) * numpy.sqrt(
2 * numpy.pi / beta_k[k] * K_k[k])
N_kn[k, n] = scipy.stats.geom.rvs(1 - p, size=1)
# now generate random distances
x_i = numpy.random.normal(O_k[k], sigma_k[k], N_kn[k, n])
# compute potential energy of all samples in all potentials
U_kn[k, n] = 0.5 * (K_k[k] * numpy.sum(x_i**2))
addrep = [U_kn.copy(), N_kn.copy()]
reps.append(addrep)
checkensemble.ProbabilityAnalysis(N_k,
type=analysis_type,
T_k=T_k,
mu_k=mu_k,
U_kn=U_kn,
N_kn=N_kn,
kB=1.0,
title=title,
figname=options.figname,
nbins=options.nbins,
reptype=reptype,
nboots=options.nboots,
reps=reps,
cuttails=options.cuttails,
eunits='kT',
seed=options.seed)
else:
g = options.efficiency*numpy.ones(K)
print "statistical inefficiency taken from input options is %f" % (options.efficiency)
if (options.useg == 'subsample'):
readmdfiles.subsample(N_k,U_kn,V_kn,N_kn,g,type)
else:
print "statistical efficiencies used to scale the statistical uncertained determined from all data"
figname = options.figname
title = options.figname
for k in range(K-1):
print 'Now analyzing replicas %d and %d' % (k,k+1)
twoN = numpy.array([N_k[k],N_k[k+1]])
if (type in onlyE) or (type == 'enthalpy') or (type == 'jointEV'):
twoT = numpy.array([T_k[k],T_k[k+1]])
else:
twoT = None
if type in requireV:
twoP = numpy.array([P_k[k],P_k[k+1]])
else:
twoP = None
twoU = U_kn[k:k+2,:]
twoV = V_kn[k:k+2,:]
checkensemble.ProbabilityAnalysis(twoN,type=analysis_type,T_k=twoT,P_k=twoP,U_kn=twoU,V_kn=twoV,nbins=nbins,
reptype='bootstrap',g=g,nboots=nboots,bMaxwell=(type=='kinetic'),figname='replica',bLinearFit=bLinearFit,bNonLinearFit=bNonLinearFit,bMaxLikelihood=bMaxLikelihood,seed=options.seed)
# now, we construct a graph with all of the lines. We could write
# the probability analysis to do it, but better to do new specially designed plot here.