def analyze_dist_1D(list_x,nstep,outdir,dtc):
"""Take the average over the D estimates in each of these x-arrays"""
# fit range = [0,nstep*dtc[
assert len(list_x) > 0
nfiles = len(list_x)
t = np.arange(0,nstep*dtc,dtc)
alldist2 = np.zeros((nstep,nfiles),float)
allD = np.zeros((nfiles),float)
print "="*5
print "Results"
print "making %i fits" %nfiles
print "fit from %i steps, i.e. time %f ps, actually time %f ps" %(nstep,nstep*dtc,(nstep-1)*dtc)
for i in range(nfiles):
dist2 = calc_dist_1D(np.array(list_x[i]))
# only fit the first nstep time steps (use all atoms)
average = np.mean(dist2[:nstep,:],1) # average over the atoms
alldist2[:,i] = average
p = np.polyfit(t,average,1)
allD[i] = p[0] # a_1 this is in angstrom**2/ps = 1e-20/1e-12 meter**2/second
# = 1e-8 meter**2/second = 1e-4 cm**2/s
# if I want many figures:
#fit_sqrt_vs_time(np.mean(dist_xy,1),dtc,outdir+"/fig_dist.xy.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_z,1), dtc,outdir+"/fig_dist.z.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_r,1), dtc,outdir+"/fig_dist.r.%i.png"%i,title=str(i))
# extra figures
m2 = np.mean(alldist2,axis=1)
#s2 = np.std(alldist2,axis=1)
fit_sqrt_vs_time(np.sqrt(m2),dtc,outdir+"/fig_dist.average.png",title="average of %i"%nfiles)
print_output_D_ave_1D(allD)
def analyze_matrixdist(list_x,
list_y,
list_z,
dn1,
outdir,
dtc,
dn2=None,
ddn=1,
unitcell=None,
npt=False):
"Take the average over the D estimates in each of these xyz trajectories" ""
assert len(list_x) > 0
assert len(list_x) == len(list_y)
assert len(list_x) == len(list_z)
nfiles = len(list_x)
"""Conditional Mean Square Distance
dn1 -- start shift
dn2 -- end shift
ddn -- range(dn1,dn2,ddn)
Example
dn1=3, dn2=7, ddn=2
then dns=[3,5,7], nlags=3, ntime should be 8 or more
"""
if dn2 is None:
dn2 = dn1
assert dn2 >= dn1
assert ddn >= 1
lt1 = dn1 * dtc
lt2 = dn2 * dtc # I will fit in region lagtime1 to lagtime2
dns = np.arange(dn1, dn2 + 1, ddn)
lagtimes = dns * dtc
nlags = len(lagtimes)
ntime = list_x[0].shape[0]
natom = list_x[0].shape[1]
if dns[-1] >= ntime:
raise ValueError("asking for too many shifts, max dn > ntime: %i,%i" %
(dns[-1], ntime))
#if nstep: lagtimes = np.arange(0,nstep*dtc,dtc)
alldist2 = np.zeros((nlags, nfiles, 6), float)
allD = np.zeros((nfiles, 6), float)
print "=" * 5
print "Results"
print "making %i fits" % nfiles, "(number of trajectories)"
print "fit from %i lagtimes, i.e. time %f ps to time %f ps, actually time %f ps" % (
nlags, lagtimes[0], lagtimes[-1], (dn2 - dn1) * dtc)
print "calculating..."
for i in range(nfiles):
print "file", i
# matrix A contains 6 components of correlation matrix
if unitcell is None:
A, weight = calc_dist(list_x[i],
list_y[i],
list_z[i],
shifts=dns,
matrix=True)
else:
A, weight = calc_dist_folded(list_x[i],
list_y[i],
list_z[i],
unitcell,
shifts=dns,
matrix=True,
npt=npt)
for k, dist2 in enumerate(A):
average = np.mean(dist2, 1) # average over the atoms
alldist2[:, i, k] = average
#print i,list_x[i],it,average[:10]
p = np.polyfit(lagtimes, average, 1)
allD[i, k] = p[
0] # a_1 this is in angstrom**2/ps = 1e-20/1e-12 meter**2/second
# = 1e-8 meter**2/second = 1e-4 cm**2/s
# if I want many figures:
#fit_sqrt_vs_time(np.mean(dist_xy,1),dtc,outdir+"/fig_dist.xy.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_z,1), dtc,outdir+"/fig_dist.z.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_r,1), dtc,outdir+"/fig_dist.r.%i.png"%i,title=str(i))
# extra figures
means = np.mean(alldist2, axis=1) # nlags x 6
stds = np.std(alldist2, axis=1) # nlags x 6
mean_r = np.mean(alldist2[:, :, 0] + alldist2[:, :, 1] + alldist2[:, :, 2],
axis=1)
std_r = np.std(alldist2[:, :, 0] + alldist2[:, :, 1] + alldist2[:, :, 2],
axis=1)
verbose = False
labs = ["xx", "yy", "zz", "xy", "xz", "yz"]
for i in range(6):
fit_sqrt_vs_time(means[:, i],
dtc * ddn,
outdir + "/fig_dist.%s.average.png" % (labs[i]),
title="average of %i" % nfiles,
t0=lagtimes[0],
std=stds[:, i],
verbose=verbose,
sqrt=False)
fit_sqrt_vs_time(mean_r,
dtc * ddn,
outdir + "/fig_dist.r.average.png",
title="average of %i" % nfiles,
t0=lagtimes[0],
std=std_r,
verbose=verbose,
sqrt=False)
# write files
for i in range(6):
store_msd(lagtimes,
means[:, i],
outdir + "/fig_dist.%s.average.txt" % (labs[i]),
error=stds[:, i])
store_msd(lagtimes,
mean_r,
outdir + "/fig_dist.%s.average.txt" % ("r"),
error=std_r)
print_output_matrixD_ave(allD)
def analyze_dist(list_x,list_y,list_z,dn1,outdir,dtc,dn2=None,ddn=1,unitcell=None):
"Take the average over the D estimates in each of these xyz trajectories"""
assert len(list_x) > 0
assert len(list_x) == len(list_y)
assert len(list_x) == len(list_z)
nfiles = len(list_x)
"""Conditional Mean Square Distance
dn1 -- start shift
dn2 -- end shift
ddn -- range(dn1,dn2,ddn)
Example
dn1=3, dn2=7, ddn=2
then dns=[3,5,7], nlags=3, ntime should be 8 or more
"""
if dn2 is None:
dn2 = dn1
assert dn2 >= dn1
assert ddn >= 1
lt1 = dn1*dtc
lt2 = dn2*dtc # I will fit in region lagtime1 to lagtime2
dns = np.arange(dn1,dn2+1,ddn)
lagtimes = dns*dtc
nlags = len(lagtimes)
ntime = list_x[0].shape[0]
natom = list_x[0].shape[1]
if dns[-1]>=ntime:
raise ValueError("asking for too many shifts, max dn > ntime: %i,%i"%(dns[-1],ntime))
#if nstep: lagtimes = np.arange(0,nstep*dtc,dtc)
alldist2 = np.zeros((nlags,nfiles,3),float)
allD = np.zeros((nfiles,3),float)
print "="*5
print "Results"
print "making %i fits" %nfiles, "(number of trajectories)"
print "fit from %i lagtimes, i.e. time %f ps to time %f ps, actually time %f ps" %(
nlags,lagtimes[0],lagtimes[-1],(dn2-dn1)*dtc)
print "calculating..."
for i in range(nfiles):
print "file",i
if unitcell is None:
dist2_xy,dist2_z,dist2_r,weight = calc_dist(list_x[i],list_y[i],list_z[i],shifts=dns)
else:
dist2_xy,dist2_z,dist2_r,weight = calc_dist_folded(list_x[i],list_y[i],list_z[i],unitcell,shifts=dns,)
for k,dist2 in enumerate([dist2_xy,dist2_z,dist2_r]):
average = np.mean(dist2,1) # average over the atoms
alldist2[:,i,k] = average
#print i,list_x[i],it,average[:10]
p = np.polyfit(lagtimes,average,1)
allD[i,k] = p[0] # a_1 this is in angstrom**2/ps = 1e-20/1e-12 meter**2/second
# = 1e-8 meter**2/second = 1e-4 cm**2/s
# if I want many figures:
#fit_sqrt_vs_time(np.mean(dist_xy,1),dtc,outdir+"/fig_dist.xy.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_z,1), dtc,outdir+"/fig_dist.z.%i.png"%i,title=str(i))
#fit_sqrt_vs_time(np.mean(dist_r,1), dtc,outdir+"/fig_dist.r.%i.png"%i,title=str(i))
# extra figures
m2_xy = np.mean(alldist2[:,:,0],1)
s_xy = np.std(alldist2[:,:,0],1)
m2_z = np.mean(alldist2[:,:,1],1)
s_z = np.std(alldist2[:,:,1],1)
m2_r = np.mean(alldist2[:,:,2],1)
s_r = np.std(alldist2[:,:,1],1)
verbose=False
fit_sqrt_vs_time(np.sqrt(m2_xy),dtc*ddn,
outdir+"/fig_dist.xy.average.png",title="average of %i"%nfiles,
t0=lagtimes[0],std=np.sqrt(s_xy),verbose=verbose)
fit_sqrt_vs_time(np.sqrt(m2_z), dtc*ddn,
outdir+"/fig_dist.z.average.png",title="average of %i"%nfiles,
t0=lagtimes[0],std=np.sqrt(s_z))
fit_sqrt_vs_time(np.sqrt(m2_r), dtc*ddn,
outdir+"/fig_dist.r.average.png",title="average of %i"%nfiles,
t0=lagtimes[0],std=np.sqrt(s_r))
store_msd(lagtimes,m2_xy,outdir+"/fig_dist.xy.average.txt",error=s_xy)
store_msd(lagtimes,m2_z ,outdir+"/fig_dist.z.average.txt", error=s_z)
store_msd(lagtimes,m2_r ,outdir+"/fig_dist.r.average.txt", error=s_r)
print_output_D_ave(allD)
