def load_range_data(infiles, start, end):
minval = float('inf')
maxval = float('-inf')
for f in infiles:
ds = dr.loadPhi(f)
# Min, max for this dataset
tmpmax = ds.max(start=start, end=end)['$\~N$']
tmpmin = ds.min(start=start, end=end)['$\~N$']
minval = tmpmin if minval > tmpmin else minval
maxval = tmpmax if maxval < tmpmax else maxval
return minval, maxval
def loadRangeData(infiles, start, end):
minval = float('inf')
maxval = float('-inf')
for f in infiles:
ds = dr.loadPhi(f)
# Min, max for this dataset
tmpmax = ds.max(start=start, end=end)['$\~N$']
tmpmin = ds.min(start=start, end=end)['$\~N$']
minval = tmpmin if minval>tmpmin else minval
maxval = tmpmax if maxval<tmpmax else maxval
return minval, maxval
def phiAnalyze(infiles, show, start, end, outfile, conv, S, myrange, nbins):
phi_vals = np.zeros((len(infiles), 8), dtype=np.float32)
prev_phi = 0.0
prev_n = 0
n_0 = 0
n_arr = np.empty((len(infiles), 2))
for i, infile in enumerate(infiles):
log.debug('loading file: {}'.format(infile))
ds = dr.loadPhi(infile)
'''
with open(infile) as f:
# Ugh
for li, line in enumerate(f):
if (li == 2):
phi = float(line.split()[3])
break
'''
phi = ds.phi
bphi = phi * conv #Could be in kT or kJ/mol
if (args.plotDist):
#fig = pyplot.figure()
#mu = ds.data[start:].mean()[0]
rng = ds.getRange(start=start, end=end)
mu = ds.getMean(start=start, end=end, bphi=bphi)
#var = ds.data[start:].var()[0]
var = ds.getVar(start=start, end=end, bphi=bphi)
txtstr = "$\mu={:.3f}$\n$\sigma^2={:.3f}$\n$F={:.2f}$".format(
mu, var, var / mu)
#print(txtstr)
hist, bounds = np.histogram(np.array(ds.data[start:end]['$\~N$']),
bins=50)
ctrs = np.diff(bounds) / 2.0 + bounds[:-1]
pyplot.bar(ctrs, hist, width=np.diff(ctrs)[0], alpha=0.6)
#pyplot.annotate(txtstr, xy=(0.2,0.75), xytext=(0.2, 0.75),
# xycoords='figure fraction', textcoords='figure fraction')
#pyplot.suptitle(r'$\beta \phi ={:.2f}$'.format(bphi), fontsize=42)
#pyplot.legend()
if outfile:
stuff = np.zeros((ctrs.shape[0], 2), dtype=float)
stuff[:, 0] = ctrs
stuff[:, 1] = hist
np.savetxt(outfile, stuff)
log.debug("Beta*Phi: {:.3f}".format(bphi))
n = ds.getMean(start=start, end=end, bphi=bphi)
#n = ds.data[start:]['$\~N$'].mean()
log.debug(" N: {:.2f}".format(n))
lg_n = np.log(n) + bphi
log.debug(" lg(N) + beta*phi: {:.2f}".format(lg_n))
secondCum = ds.getVar(start=start, end=end, bphi=bphi)
if i == 0:
dndphi_neg = 0
n_0 = n
delta_phi = 0
try:
dndphi_neg = (prev_n - n) / (bphi - prev_phi)
except ZeroDivisionError:
dndphi_neg = 0
# Suggested delta phi (adaptive method)
try:
delta_phi = (n_0 / S) / secondCum
except ZeroDivisionError:
delta_phi = 0
try:
lg_n_negslope = (dndphi_neg / n) - 1
except ZeroDivisionError:
lg_n_negslope = -1
phi_vals[
i] = bphi, n, 0, dndphi_neg, lg_n, lg_n_negslope, delta_phi, secondCum
prev_phi = bphi
prev_n = n
for i in range(phi_vals.shape[0]):
phi_vals[i, 2] = np.trapz(phi_vals[:i + 1, 1], phi_vals[:i + 1, 0])
# sort by phi
phi_vals = phi_vals[phi_vals[:, 0].argsort()]
# Fill in deriv of lg_beta graph
# phi_vals[1:,5] = - np.diff(phi_vals[:, 2]) / np.diff(phi_vals[:, 0])
log.debug("Phi Vals: {}".format(phi_vals))
if outfile and not args.plotDist:
np.savetxt(outfile, phi_vals, fmt='%.2f')
if show:
if (args.plotDistAll):
total_stuff = dr.plotHistAll(start=start, end=end, nbins=nbins)
if outfile:
np.savetxt(outfile, total_stuff)
#for ds in dr.datasets.values():
# hist, binbounds = np.histogram(ds.data[start:end]['N'], normed=True)
# plt.plot(binbounds[:-1]+np.diff(binbounds)/2.0, -np.log(hist))
#plt.show()
if (args.plotN):
title = r"$\langle{N_V}\rangle_\phi$"
num = 1
pyplot.plot(phi_vals[:, 0],
phi_vals[:, num],
'o-',
markersize=12,
linewidth=4)
#pyplot.fill_between(phi_vals[:,0], phi_vals[:,num], color='none', hatch='\\', edgecolor='b')
if (args.plotInteg):
#title = r'$\int{\langle{N}\rangle_\phi}d\phi$'
title = r'$G(\phi)$'
num = 2
pyplot.plot(phi_vals[:, 0],
phi_vals[:, num],
'o-',
markersize=12,
linewidth=4)
if (args.plotLogN):
title = r'$\ln{\langle{N}\rangle_\phi} + \beta\phi$'
num = 4
pyplot.plot(phi_vals[:, 0], phi_vals[:, num], 'o-')
if (args.plotNegSlope):
title = r'$\langle{\delta N^2}\rangle_\phi / \langle{N}\rangle_\phi - 1$'
num = 5
pyplot.plot(phi_vals[:, 0], phi_vals[:, num], 'o-')
if (args.plotSus):
title = r"$\langle{\delta N^2}\rangle_\phi$"
num = 7
pyplot.plot(phi_vals[:, 0],
phi_vals[:, num],
'o-',
markersize=12,
linewidth=4)
if (args.plotBoth):
pyplot.close('all')
f, axarr = pyplot.subplots(2, sharex=True)
axarr[0].plot(phi_vals[:, 0], phi_vals[:, 1], 'o-')
axarr[0].set_ylabel(r'$\langle{N}\rangle_\phi$')
ticks = axarr[0].get_yticks()
axarr[0].set_yticks(ticks[::2])
axarr[1].plot(phi_vals[:, 0], phi_vals[:, 7], 'o-')
axarr[1].set_ylabel(r'$\langle{\delta N^2}\rangle_\phi$')
ticks = axarr[0].get_yticks()
axarr[0].set_yticks(ticks[::2])
if conv == 1:
pyplot.xlabel(r'$\phi$ (kJ/mol)')
else:
pyplot.xlabel(r'$\beta\phi$ ($k_B T$)')
pyplot.tight_layout()
pyplot.show()
if (args.plotDist or args.plotDistAll):
pyplot.tight_layout()
pyplot.show()
else:
#pyplot.plot(phi_vals[:, 0],phi_vals[:, num])
#pyplot.title(title + r' vs $\beta\phi$')
if conv == 1:
pyplot.xlabel(r'$\phi$ (kJ/mol)')
else:
pyplot.xlabel(r'$\beta\phi$ ($k_B T$)')
pyplot.ylabel(title)
if myrange:
pyplot.ylim(myrange)
pyplot.show()
end = args.end
outfile = args.output
conv = 1
if args.T:
conv /= (args.T * 0.008314)
S = args.S or len(
infiles
) # Number of phi values - used to suggest delta_phi using adaptive method
myrange = None
if args.range:
myrange = parseRange(args.range)
if len(infiles) == 1 and args.blockAvg:
ds = dr.loadPhi(infiles[0], corr_len=1)
block_vals = ds.blockAvg(start, end=end)
data_len = np.array(ds.data[start:end]['$\~N$']).shape[0]
print("Data length:{}".format(data_len))
pyplot.plot(block_vals[:, 0], np.sqrt(block_vals[:, 2]), 'ro')
pyplot.xlim(0, block_vals.shape[0] / 2)
pyplot.xlabel("Block size")
pyplot.ylabel(r'$\sigma_{{\langle{\~N}\rangle}}$')
pyplot.show()
if outfile:
block = input("Input block size for output: ")
ds.printOut(outfile, start=start, block=block)
elif len(infiles) == 1 and not args.plotDist:
dr.loadPhi(infiles[0])
dr.plot(start=start, end=end, ylim=myrange)
dr.show()
import glob, os, sys
import numpy as np
from matplotlib import pyplot as plt
import matplotlib as mpl
from mdtools import dr
from constants import k
import MDAnalysis
import argparse
from scipy.spatial import cKDTree
ds = dr.loadPhi('phiout.dat')
dat = ds.data[500:]['$\~N$']
mean = dat.mean()
var = dat.var(ddof=1)
std = dat.std(ddof=1)
kt = 300 * k
alpha = 3
kappa = alpha * kt / var
dn = int(np.ceil(4 * (np.sqrt(1 + alpha) / alpha) * std))
neg_nstar = mean / alpha
print('Mean: {:0.2f} var: {:0.2f} std: {:0.2f}'.format(mean, var, std))
print(' kappa: {:0.4f} dn: {:02d} neg_nstar: {:0.2f}'.format(
parser.add_argument(
'--alpha',
default=3,
type=float,
help='Alpha val by which to choose kappa (default: %(default)s)')
parser.add_argument(
'--temp',
default=300,
type=float,
help='Operating temp for determining kT (default: %(default)s K)')
args = parser.parse_args()
kt = args.temp * k
ds = dr.loadPhi(args.f)
start = args.b
alpha = args.alpha
dat = ds.data[start:]['$\~N$']
mean = dat.mean()
var = dat.var()
std = dat.std()
print('\navg ntwid: {:0.2f} var ntwid: {:0.2f} (std: {:0.2f})'.format(
mean, var, std))
print(' alpha: {:0.1f}'.format(alpha))
kappa = np.round((alpha * kt / var), 3)
dn = int(np.ceil(4 * (np.sqrt(1 + alpha) / alpha) * std))
psi_stars = []
n_samples = []
uncorr_n_samples = []
nreg_dat = []
ntwid_dat = []
uncorr_nreg_dat = []
uncorr_ntwid_dat = []
start = 500
this_mu = None
for fname in fnames:
rama = np.loadtxt(fname, usecols=(0,1), comments=['@','#'])
dirname = os.path.dirname(fname)
ds = dr.loadPhi('{}/phiout.dat'.format(dirname))
if this_mu is None:
this_mu = ds.phi
else:
assert this_mu == ds.phi
n_dat = ds.data[start::10]
this_nreg_dat = np.array(n_dat['N'])
this_ntwid_dat = np.array(n_dat['$\~N$'])
with open("{}/topol.top".format(dirname), "r") as f:
lines = f.readlines()
phi_line = lines[-34].strip().split()
psi_line = lines[-33].strip().split()
phi_kappas.append(float(phi_line[-1]))
phi_stars.append(float(phi_line[-2]))
parser.add_argument('--fmt', type=str, choices=['phi', 'xvg'], default='phi',
help='Format of input data files: \'phi\' for phiout.dat;' \
'\'xvg\' for XVG type files (i.e. from alchemical GROMACS sims)')
args = parser.parse_args()
if args.verbose:
log.setLevel(logging.INFO)
if args.debug:
log.setLevel(logging.DEBUG)
log.info("Loading input files")
if args.fmt == 'phi':
for infile in args.input:
dr.loadPhi(infile)
elif args.fmt == 'xvg':
for infile in args.input:
dr.loadXVG(infile)
log.info(" ...Done")
start = args.start
end = args.end
logweights = numpy.loadtxt(args.logweights)
weights = numpy.exp(logweights)
logP = numpy.loadtxt(args.logpdist)
logP[:,1] = -logP[:,1]
probDist = logP.copy()
)
parser.add_argument('input',
metavar='INPUT',
type=str,
nargs='+',
help='Input file names')
parser.add_argument("--phi",
type=float,
default=20.0,
help="This phi (kJ/mol) (default: 20.0 kJ/mol)")
args = parser.parse_args()
fnames = args.input
ds_0 = dr.loadPhi(fnames[0])
times = np.array(ds_0.data.index)
n_dat = times.size
biases = np.zeros((n_dat, len(fnames) + 1))
biases[:, 0] = times
for i, fname in enumerate(fnames):
ds = dr.loadPhi(fname)
assert np.allclose(ds.data.index, ds_0.data.index)
biases[:, i + 1] = args.phi * ds.data['$\~N$']
np.savetxt("phi_biases.dat",
biases,
fmt='%02.6f',
header='Time(ps) [biases (kJ/mol)...]')
def phiAnalyze(infiles, show, start, end, outfile, conv, S, myrange, nbins):
phi_vals = np.zeros((len(infiles), 8), dtype=np.float32)
prev_phi = 0.0
prev_n = 0
n_0 = 0
n_arr = np.empty((len(infiles),2))
for i, infile in enumerate(infiles):
log.debug('loading file: {}'.format(infile))
ds = dr.loadPhi(infile)
'''
with open(infile) as f:
# Ugh
for li, line in enumerate(f):
if (li == 2):
phi = float(line.split()[3])
break
'''
phi = ds.phi
bphi = phi*conv #Could be in kT or kJ/mol
if (args.plotDist):
#fig = pyplot.figure()
#mu = ds.data[start:].mean()[0]
rng = ds.getRange(start=start, end=end)
mu = ds.getMean(start=start, end=end, bphi=bphi)
#var = ds.data[start:].var()[0]
var = ds.getVar(start=start, end=end, bphi=bphi)
txtstr = "$\mu={:.3f}$\n$\sigma^2={:.3f}$\n$F={:.2f}$".format(mu, var, var/mu)
#print(txtstr)
hist, bounds = np.histogram(np.array(ds.data[start:end]['$\~N$']), bins=50, normed=1)
ctrs = np.diff(bounds)/2.0 + bounds[:-1]
pyplot.bar(ctrs, hist, width=np.diff(ctrs)[0])
#pyplot.annotate(txtstr, xy=(0.2,0.75), xytext=(0.2, 0.75),
# xycoords='figure fraction', textcoords='figure fraction')
#pyplot.suptitle(r'$\beta \phi ={:.2f}$'.format(bphi), fontsize=42)
#pyplot.legend()
if outfile:
stuff = np.zeros((ctrs.shape[0], 2), dtype=float)
stuff[:,0] = ctrs
stuff[:,1] = hist
np.savetxt(outfile, stuff)
log.debug("Beta*Phi: {:.3f}".format(bphi))
n = ds.getMean(start=start, end=end, bphi=bphi)
#n = ds.data[start:]['$\~N$'].mean()
log.debug(" N: {:.2f}".format(n))
lg_n = np.log(n)+bphi
log.debug(" lg(N) + beta*phi: {:.2f}".format(lg_n))
secondCum = ds.getVar(start=start, end=end, bphi=bphi)
if i == 0:
dndphi_neg = 0
n_0 = n
delta_phi = 0
try:
dndphi_neg = (prev_n - n) / (bphi - prev_phi)
except ZeroDivisionError:
dndphi_neg = 0
# Suggested delta phi (adaptive method)
try:
delta_phi = (n_0/S)/secondCum
except ZeroDivisionError:
delta_phi = 0
try:
lg_n_negslope = (dndphi_neg/n) - 1
except ZeroDivisionError:
lg_n_negslope = -1
phi_vals[i] = bphi, n, 0, dndphi_neg, lg_n, lg_n_negslope, delta_phi, secondCum
prev_phi = bphi
prev_n = n
for i in xrange(phi_vals.shape[0]):
phi_vals[i, 2] = np.trapz(phi_vals[:i+1, 1], phi_vals[:i+1, 0])
# sort by phi
phi_vals = phi_vals[phi_vals[:,0].argsort()]
# Fill in deriv of lg_beta graph
# phi_vals[1:,5] = - np.diff(phi_vals[:, 2]) / np.diff(phi_vals[:, 0])
log.debug("Phi Vals: {}".format(phi_vals))
if outfile and not args.plotDist:
np.savetxt(outfile, phi_vals, fmt='%.2f')
if show:
if (args.plotDistAll):
total_stuff = dr.plotHistAll(start=start, end=end, nbins=nbins)
if outfile:
np.savetxt(outfile, total_stuff)
if (args.plotN):
title = r"$\langle{\~N}\rangle_\phi$"
num = 1
pyplot.plot(phi_vals[:, 0],phi_vals[:, num], 'o-', markersize=12, linewidth=4)
#pyplot.fill_between(phi_vals[:,0], phi_vals[:,num], color='none', hatch='\\', edgecolor='b')
if (args.plotInteg):
title = r'$\int{\langle{N}\rangle_\phi}d\phi$'
num = 2
pyplot.plot(phi_vals[:, 0], phi_vals[:, num], 'o-')
if (args.plotLogN):
title = r'$\ln{\langle{N}\rangle_\phi} + \beta\phi$'
num = 4
pyplot.plot(phi_vals[:, 0],phi_vals[:, num], 'o-')
if (args.plotNegSlope):
title = r'$\langle{\delta N^2}\rangle_\phi / \langle{N}\rangle_\phi - 1$'
num = 5
pyplot.plot(phi_vals[:, 0],phi_vals[:, num], 'o-')
if (args.plotSus):
title = r"$\langle{\delta \~N^2}\rangle_\phi$"
num = 7
pyplot.plot(phi_vals[:, 0], phi_vals[:, num], 'o-', markersize=12, linewidth=4)
if (args.plotBoth):
pyplot.close('all')
f, axarr = pyplot.subplots(2, sharex=True)
axarr[0].plot(phi_vals[:, 0],phi_vals[:, 1], 'o-')
axarr[0].set_ylabel(r'$\langle{N}\rangle_\phi$')
ticks = axarr[0].get_yticks()
axarr[0].set_yticks(ticks[::2])
axarr[1].plot(phi_vals[:, 0], phi_vals[:, 7], 'o-')
axarr[1].set_ylabel(r'$\langle{\delta N^2}\rangle_\phi$')
ticks = axarr[0].get_yticks()
axarr[0].set_yticks(ticks[::2])
if conv == 1:
pyplot.xlabel(r'$\phi$ (kJ/mol)')
else:
pyplot.xlabel(r'$\beta\phi$ ($k_B T$)')
pyplot.show()
if (args.plotDist or args.plotDistAll):
pyplot.show()
else:
#pyplot.plot(phi_vals[:, 0],phi_vals[:, num])
#pyplot.title(title + r' vs $\beta\phi$')
if conv == 1:
pyplot.xlabel(r'$\phi$ (kJ/mol)')
else:
pyplot.xlabel(r'$\beta\phi$ ($k_B T$)')
pyplot.ylabel(title)
if myrange:
pyplot.ylim(myrange)
pyplot.show()
log.debug("{} input files".format(len(infiles)))
start = args.start
end = args.end
outfile = args.output
conv = 1
if args.T:
conv /= (args.T * 0.008314)
S = args.S or len(infiles) # Number of phi values - used to suggest delta_phi using adaptive method
myrange = None
if args.range:
myrange = parseRange(args.range)
if len(infiles) == 1 and args.blockAvg:
ds = dr.loadPhi(infiles[0], corr_len=1)
block_vals = ds.blockAvg(start,end=end)
data_len = np.array(ds.data[start:end]['$\~N$']).shape[0]
print "Data length:{}".format(data_len)
pyplot.plot(block_vals[:,0], np.sqrt(block_vals[:,2]), 'ro')
pyplot.xlim(0,block_vals.shape[0]/2)
pyplot.xlabel("Block size")
pyplot.ylabel(r'$\sigma_{{\langle{\~N}\rangle}}$')
pyplot.show()
if outfile:
block = input("Input block size for output: ")
ds.printOut(outfile, start=start, block=block)
elif len(infiles) == 1 and not args.plotDist:
dr.loadPhi(infiles[0])
dr.plot(start=start, end=end, ylim=myrange)
dr.show()
start = args.start
end = args.end
n_windows = len(args.input)
all_dat = np.array([])
n_samples = []
beta = 1
if args.T:
beta /= (args.T * k)
ts = None
for infile in args.input:
log.info("Loading data {}".format(infile))
ds = dr.loadPhi(infile)
if ts is None:
ts = ds.ts
else:
assert ts == ds.ts
dat = np.array(ds.data[start:end]['$\~N$'])
all_dat = np.append(all_dat, dat)
n_samples.append(dat.size)
if args.autocorr is not None:
autocorr_nsteps = (1 + 2 *
(np.loadtxt(args.autocorr) / ts)).astype(int)
else:
autocorr_nsteps = np.ones(n_windows, dtype=int)
n_samples = np.array(n_samples)