forked from IPKM/nmreval
some stuff done
This commit is contained in:
Dominik Demuth
parent
996b0b2ae2
commit
bc215ce32b
@ -1,41 +1,26 @@
|
||||
# Tool-Naranayaswamy-Moynihan-Hodge model
|
||||
# by Florian Pabst 2020
|
||||
|
||||
|
||||
import re
|
||||
import os
|
||||
import sys
|
||||
import time
|
||||
|
||||
import numpy as np
|
||||
import matplotlib.pyplot as plt
|
||||
|
||||
from scipy.integrate import quad, cumulative_trapezoid
|
||||
from scipy.integrate import cumulative_trapezoid
|
||||
from scipy.optimize import curve_fit, fsolve
|
||||
from scipy.stats import linregress
|
||||
|
||||
from nmreval.data import Points
|
||||
|
||||
|
||||
def slope(x, t, m):
|
||||
return t+x*m
|
||||
|
||||
|
||||
def tau_k(tau_0, deltaE, temp_k, xx, T_f_km1):
|
||||
return tau_0 * np.exp(xx*deltaE / (R*temp_k) + (1-xx) * deltaE / (R*T_f_km1))
|
||||
|
||||
|
||||
# Read data
|
||||
dataName = sys.argv[1]
|
||||
try:
|
||||
data = np.loadtxt(dataName, skiprows=0).T
|
||||
# print("Loading")
|
||||
temp = data[0]
|
||||
heat_capacity = data[1]
|
||||
except IndexError:
|
||||
# print("File not found")
|
||||
exit()
|
||||
|
||||
print(np.round(temp, decimals=3)[:30])
|
||||
|
||||
|
||||
def get_fictive_temperature(pts: Points, glass: tuple[float, float], liquid: tuple[float, float]):
|
||||
min_glass, max_glass = min(glass), max(glass)
|
||||
@ -51,8 +36,8 @@ def get_fictive_temperature(pts: Points, glass: tuple[float, float], liquid: tup
|
||||
region.y -= glass_extrapolation
|
||||
|
||||
liquid_regime = (min_liquid < region.x) & (region.x < max_liquid)
|
||||
regress = linregress(region.x[liquid_regime], region.y[liquid_regime])
|
||||
liquid_extrapolation = regress.slope * region.x + regress.intercept
|
||||
regress2 = linregress(region.x[liquid_regime], region.y[liquid_regime])
|
||||
liquid_extrapolation = regress2.slope * region.x + regress2.intercept
|
||||
|
||||
real_area = -cumulative_trapezoid(region.y, region.x, initial=0)
|
||||
real_area -= real_area[-1]
|
||||
@ -60,10 +45,12 @@ def get_fictive_temperature(pts: Points, glass: tuple[float, float], liquid: tup
|
||||
equivalent_area -= equivalent_area[-1]
|
||||
equivalent_area *= -1
|
||||
|
||||
return region.x, region.x[np.argmin(np.abs(real_area[:, None] - equivalent_area[None, :]), axis=1)]
|
||||
return region.x, np.round(region.x[np.argmin(np.abs(real_area[:, None] - equivalent_area[None, :]), axis=1)], decimals=4)
|
||||
|
||||
|
||||
curve = Points(x=temp, y=heat_capacity, name=dataName)
|
||||
# plt.plot(curve.x, curve.y)
|
||||
# plt.show()
|
||||
|
||||
# First step: Calculate fictive Temperature T_f(T) from Cp data, then dT_f/dT
|
||||
|
||||
@ -75,21 +62,19 @@ initial_Tf = np.mean(fictiveTemp[:20])
|
||||
rate = curve.value
|
||||
|
||||
# Calculate dTf/dT
|
||||
with np.errstate(all='ignore'):
|
||||
dTfdT = np.gradient(fictiveTemp, temperatures)
|
||||
nan_filter = ~np.isnan(dTfdT)
|
||||
temperatures = np.array(temperatures)[nan_filter]
|
||||
dTfdT = dTfdT[nan_filter]
|
||||
# with np.errstate(all='ignore'):
|
||||
# dTfdT = np.diff(fictiveTemp)/np.diff(temperatures) # np.gradient(fictiveTemp, temperatures)
|
||||
# nan_filter = ~np.isnan(dTfdT)
|
||||
# temperatures = 0.5 * (temperatures[:-1] + temperatures[1:])[nan_filter]
|
||||
# # fictiveTemp = fictiveTemp[nan_filter]
|
||||
# dTfdT = dTfdT[nan_filter]
|
||||
|
||||
print(fictiveTemp[:30])
|
||||
|
||||
exponents = []
|
||||
taus = []
|
||||
temps = [0]
|
||||
T_f_ns = []
|
||||
R = 8.314
|
||||
|
||||
|
||||
def tnmfunc(temperature, tau_g, x, beta, deltaE):
|
||||
def tnm_function(temperature, tau_g, x, beta, energy, tg):
|
||||
step = len(temperature)
|
||||
|
||||
Tf = np.empty(step)
|
||||
@ -103,7 +88,7 @@ def tnmfunc(temperature, tau_g, x, beta, deltaE):
|
||||
temp_0 = temperature[0]
|
||||
|
||||
for i in range(0, step-1):
|
||||
tau[i] = relax(temperature[i+1], Tf[i], tau_g, T_g, deltaE, x)
|
||||
tau[i] = relax(temperature[i+1], Tf[i], tau_g, tg, energy, x)
|
||||
dttau[:i] += delt[i] / tau[i]
|
||||
Tf[i+1] = np.sum(delT[:i] * (1-np.exp(-dttau[:i]**beta))) + temp_0
|
||||
|
||||
@ -115,147 +100,6 @@ def relax(t, tf, tau_g, t_glass, ea, x):
|
||||
return tau_g * np.exp((x*h / t) + ((1 - x) * h / tf) - h / t_glass)
|
||||
|
||||
|
||||
# TNMH Code (adapted from Badrinarayanan's PhD Thesis):
|
||||
def tnmfunc2(xdata, tau_0, xx, beta, deltaE):
|
||||
delhR = deltaE/8.314
|
||||
T = []
|
||||
Tf = []
|
||||
t = []
|
||||
delt = []
|
||||
tau = []
|
||||
dttau = []
|
||||
delT = []
|
||||
step = len(xdata)
|
||||
for k in range(0, step):
|
||||
dttau.append(0)
|
||||
T.append(240)
|
||||
Tf.append(240)
|
||||
t.append(0)
|
||||
delt.append(0)
|
||||
tau.append(0)
|
||||
dttau.append(0)
|
||||
delT.append(0)
|
||||
T[1] = xdata[1]
|
||||
Tf[1] = T[1]
|
||||
t[1] = 0
|
||||
delt[1] = 0
|
||||
tau[1] = tau_0 * np.exp((xx*delhR / T[1]) + ((1 - xx)*delhR / Tf[1])-delhR/T_g)
|
||||
dttau[1] = delt[1] / tau[1]
|
||||
delT[1] = 0
|
||||
|
||||
Tfinit = T[1]
|
||||
T_f_ns = [xdata[0]]
|
||||
for i in range(1, step):
|
||||
T[i] = xdata[i]
|
||||
delT[i] = xdata[i] - xdata[i-1]
|
||||
delt[i] = abs(delT[i]) / (rate/60)
|
||||
t[i] = t[i-1] + delt[i]
|
||||
tau[i] = tau_0 * np.exp((delhR*xx / T[i]) + ((1 - xx)*delhR / Tf[i-1])-delhR/T_g)
|
||||
# print(tau[i], T[i], Tf[i-1], dttau)
|
||||
for j in np.arange(2, i).reshape(-1):
|
||||
dttau[j] = dttau[j] + (delt[i] / tau[i])
|
||||
Tfinit = Tfinit + (delT[j] * (1 - np.exp(- (dttau[j] ** beta))))
|
||||
Tf[i] = Tfinit
|
||||
T_f_ns.append(Tfinit)
|
||||
Tfinit = Tf[1]
|
||||
|
||||
return T_f_ns
|
||||
|
||||
|
||||
def tnmfunc3(xdata, tau_0, xx, beta, deltaE):
|
||||
delhR = deltaE/8.314
|
||||
|
||||
step = len(xdata)
|
||||
|
||||
dttau = []
|
||||
T = []
|
||||
Tf = []
|
||||
t = []
|
||||
delT = []
|
||||
delt = []
|
||||
tau = []
|
||||
for _ in range(step):
|
||||
T.append(np.nan)
|
||||
Tf.append(np.nan)
|
||||
t.append(np.nan)
|
||||
delT.append(np.nan)
|
||||
delt.append(np.nan)
|
||||
tau.append(np.nan)
|
||||
dttau.append(0)
|
||||
|
||||
T[0] = xdata[0]
|
||||
Tf[0] = T[0]
|
||||
t[0] = 0
|
||||
delT[0] = 0
|
||||
delt[0] = 0
|
||||
tau[0] = tau_0 * np.exp(xx * delhR/T[0] + (1-xx)*delhR/Tf[0] - delhR/T_g)
|
||||
dttau[0] = delt[0]/tau[0]
|
||||
|
||||
Tfinit = T[0]
|
||||
|
||||
for i in range(1, step):
|
||||
T[i] = xdata[i]
|
||||
delT[i] = xdata[i] - xdata[i-1]
|
||||
delt[i] = abs(delT[i]) * 60 / rate
|
||||
t[i] = t[i-1] + delt[i]
|
||||
tau[i] = tau_0 * np.exp(xx * delhR/T[i] + (1-xx)*delhR/Tf[i-1] - delhR/T_g)
|
||||
|
||||
for j in range(1, i):
|
||||
dttau[j] += delt[i]/tau[i]
|
||||
Tfinit += delT[j] * (1-np.exp(-dttau[j]**beta))
|
||||
|
||||
Tf[i] = Tfinit
|
||||
Tfinit = T[0]
|
||||
|
||||
return Tf
|
||||
|
||||
|
||||
def tnmfunc2(xdata, tau_0, xx, beta, deltaE):
|
||||
delhR = deltaE/8.314
|
||||
T = []
|
||||
Tf = []
|
||||
t = []
|
||||
delt = []
|
||||
tau = []
|
||||
dttau = []
|
||||
delT = []
|
||||
step = len(xdata)
|
||||
for k in range(0, step):
|
||||
dttau.append(0)
|
||||
T.append(240)
|
||||
Tf.append(240)
|
||||
t.append(0)
|
||||
delt.append(0)
|
||||
tau.append(0)
|
||||
dttau.append(0)
|
||||
delT.append(0)
|
||||
T[1] = xdata[1]
|
||||
Tf[1] = T[1]
|
||||
t[1] = 0
|
||||
delt[1] = 0
|
||||
tau[1] = tau_0 * np.exp((xx*delhR / T[1]) + ((1 - xx)*delhR / Tf[1])-delhR/T_g)
|
||||
dttau[1] = delt[1] / tau[1]
|
||||
delT[1] = 0
|
||||
|
||||
Tfinit = T[1]
|
||||
T_f_ns = [xdata[0]]
|
||||
for i in range(1, step):
|
||||
T[i] = xdata[i]
|
||||
delT[i] = xdata[i] - xdata[i-1]
|
||||
delt[i] = abs(delT[i]) / (rate/60)
|
||||
t[i] = t[i-1] + delt[i]
|
||||
tau[i] = tau_0 * np.exp((delhR*xx / T[i]) + ((1 - xx)*delhR / Tf[i-1])-delhR/T_g)
|
||||
# print(tau[i], T[i], Tf[i-1], dttau)
|
||||
for j in np.arange(2, i).reshape(-1):
|
||||
dttau[j] = dttau[j] + (delt[i] / tau[i])
|
||||
Tfinit = Tfinit + (delT[j] * (1 - np.exp(- (dttau[j] ** beta))))
|
||||
Tf[i] = Tfinit
|
||||
T_f_ns.append(Tfinit)
|
||||
Tfinit = Tf[1]
|
||||
|
||||
return T_f_ns
|
||||
|
||||
|
||||
# Use T_g = T_fictive for tau determination
|
||||
T_g = initial_Tf
|
||||
|
||||
@ -263,48 +107,26 @@ T_g = initial_Tf
|
||||
p0 = [10, 0.5, 0.4, 225085]
|
||||
|
||||
|
||||
def fitTNMH(x, tau_0, xx, beta, deltaE):
|
||||
dTNMHdT = []
|
||||
modelTemp = np.append(x[::-1], x[1:])
|
||||
def tnmh_fit(tg):
|
||||
def wrap(x, tau_0, xx, beta, energy):
|
||||
modelTemp = np.r_[x[::-1], x[1:]]
|
||||
|
||||
TNMH = tnmfunc2(modelTemp, tau_0, xx, beta, deltaE)
|
||||
for i in range(len(modelTemp)-1):
|
||||
dTNMHdT.append((TNMH[i+1]-TNMH[i]) / (modelTemp[i+1]-modelTemp[i]))
|
||||
res = dTNMHdT[len(modelTemp)//2-1:]
|
||||
TNMH = tnm_function(modelTemp, tau_0, xx, beta, energy, tg)
|
||||
res = np.gradient(TNMH, modelTemp)
|
||||
|
||||
plt.plot(x, res)
|
||||
plt.show()
|
||||
return res[len(x)-1:]
|
||||
|
||||
return np.array(res)
|
||||
return wrap
|
||||
|
||||
|
||||
# Use only every 20th data point to reduce fitting time
|
||||
temperaturesInterpol = np.linspace(temperatures[0], temperatures[-1], 20)
|
||||
temperaturesInterpol = np.linspace(150, 200, 51)
|
||||
|
||||
temperaturesInterpol = np.append(temperaturesInterpol[::-1], temperaturesInterpol[1:])
|
||||
|
||||
start = time.time()
|
||||
tnmfunc2(temperaturesInterpol, *p0)
|
||||
print('Flo', time.time()-start)
|
||||
start = time.time()
|
||||
tnmfunc3(temperaturesInterpol, *p0)
|
||||
print('list', time.time()-start)
|
||||
start = time.time()
|
||||
tnmfunc(temperaturesInterpol, *p0)
|
||||
print('Array', time.time()-start)
|
||||
|
||||
temperaturesInterpol = np.linspace(temperatures[0], temperatures[-1], 200)
|
||||
dTfdTInterpol = np.interp(temperaturesInterpol, temperatures, dTfdT)
|
||||
|
||||
plt.plot(tnmfunc2(temperaturesInterpol, *p0), label='Flo')
|
||||
plt.plot(tnmfunc3(temperaturesInterpol, *p0), label='List')
|
||||
plt.plot(tnmfunc(temperaturesInterpol, *p0), label='array')
|
||||
plt.legend()
|
||||
plt.show()
|
||||
res = curve_fit(tnmh_fit(T_g), temperaturesInterpol, dTfdTInterpol, p0)
|
||||
|
||||
plt.plot(np.diff(tnmfunc2(temperaturesInterpol, *p0)), label='Flo')
|
||||
plt.plot(np.diff(tnmfunc3(temperaturesInterpol, *p0)), label='List')
|
||||
plt.plot(np.diff(tnmfunc(temperaturesInterpol, *p0)), label='array')
|
||||
|
||||
plt.plot(temperatures, dTfdT, label='dTf/dT')
|
||||
plt.plot(temperaturesInterpol, tnmh_fit(T_g)(temperaturesInterpol, *res[0]), label='fit')
|
||||
plt.legend()
|
||||
plt.show()
|
||||
# print(fitTNMH(temperaturesInterpol, *p0))
|
||||
|
||||
@ -26,7 +26,6 @@ except IndexError:
|
||||
|
||||
dataOutName = os.path.splitext(str(dataName))[0] + "_Tfict+TNMH.dat"
|
||||
|
||||
|
||||
# First step: Calculate fictive Temperature T_f(T) from Cp data, then dT_f/dT
|
||||
|
||||
# Find start and end point for glass Cp linear fit
|
||||
@ -111,7 +110,7 @@ res = curve_fit(slope, dataTemp[fitLimitLeft_Gl:fitLimitRight_Gl], dataCp[fitLim
|
||||
glassCpParam = res[0]
|
||||
|
||||
# Determine liquid Cp slope
|
||||
res = curve_fit(slope, dataTemp[fitLimitLeft_Lq:fitLimitRight_Lq], dataCp[fitLimitLeft_Lq:fitLimitRight_Lq])
|
||||
res = curve_fit(slope, dataTemp[fitLimitLeft_Lq:fitLimitRight_Lq], dataCp[fitLimitLeft_Lq:fitLimitRight_Lq])
|
||||
liquidCpParam = res[0]
|
||||
|
||||
xLin = np.linspace(coordsGl[0][0], coordsLq[1][0], len(dataTemp))
|
||||
@ -138,7 +137,7 @@ intStop = fitLimitRight_Lq
|
||||
|
||||
fictiveTemp = []
|
||||
temperatures = []
|
||||
for i,tempStart in enumerate(dataTemp):
|
||||
for i, tempStart in enumerate(dataTemp):
|
||||
if fitLimitLeft_Gl < i < fitLimitRight_Lq:
|
||||
intStart = i
|
||||
|
||||
@ -172,20 +171,15 @@ for i in range(len(temperatures)-1):
|
||||
dTfdT.append((fictiveTemp[i+1]-fictiveTemp[i]) / (temperatures[i+1]-temperatures[i]))
|
||||
temperaturesCut.append(temperatures[i+1])
|
||||
|
||||
"""
|
||||
fig, (ax1) = P.subplots(1)
|
||||
print(fictiveTemp[:30])
|
||||
|
||||
fig, ax1 = plt.subplots(1)
|
||||
ax1.set_xlabel('Temperature / K')
|
||||
ax1.set_ylabel('Fictive Temperature / K')
|
||||
#ax1.set_xlim(left=coordsGl[0][0])
|
||||
#ax1.set_xlim(right=coordsLq[1][0])
|
||||
#ax1.set_ylim(bottom=coordsGl[0][1]-0.1)
|
||||
#ax1.set_ylim(top=coordsLq[1][1]+0.1)
|
||||
ax1.plot(temperatures,fictiveTemp, 'ro')
|
||||
#ax1.plot(temperatures,(T_f_0+temperatures*0), 'b-')
|
||||
ax1.axhline(y=T_f_0[0], color='b', linestyle='-')
|
||||
#ax1.plot(temperatures[:-1],dTfdT, 'ro')
|
||||
P.show()
|
||||
"""
|
||||
# ax1.plot(temperatures, fictiveTemp, 'ro')
|
||||
ax1.plot(temperaturesCut, dTfdT)
|
||||
# ax1.axhline(y=T_f_0[0], color='b', linestyle='-')
|
||||
plt.show()
|
||||
|
||||
# Second step: Calcualte and Fit TNMH-model
|
||||
try:
|
||||
@ -207,6 +201,7 @@ T_f_ns = []
|
||||
markovs = [coordsLq[1][0]]
|
||||
R = 8.31
|
||||
|
||||
|
||||
# TNMH Code (adapted from Badrinarayanan's PhD Thesis):
|
||||
def tnmfunc2(xdata, tau_0, xx, beta, deltaE):
|
||||
delhR = deltaE/8.314
|
||||
@ -250,7 +245,6 @@ def tnmfunc2(xdata, tau_0, xx, beta, deltaE):
|
||||
T_f_ns.append(Tfinit)
|
||||
Tfinit = Tf[1]
|
||||
|
||||
print(T_f_ns[:10])
|
||||
return T_f_ns
|
||||
|
||||
|
||||
@ -297,25 +291,15 @@ tau_0, xx, beta, deltaE = p0
|
||||
|
||||
fitResult = fitTNMH(p0, temperaturesInterpol)
|
||||
|
||||
"""
|
||||
fig, (ax1) = P.subplots(1)
|
||||
|
||||
fig, ax1 = plt.subplots()
|
||||
ax1.set_xlabel('Temperature / K')
|
||||
ax1.set_ylabel('Fictive Temperature / K')
|
||||
#ax1.set_xlim(left=coordsGl[0][0])
|
||||
#ax1.set_xlim(right=coordsLq[1][0])
|
||||
#ax1.set_ylim(bottom=coordsGl[0][1]-0.1)
|
||||
#ax1.set_ylim(top=coordsLq[1][1]+0.1)
|
||||
#ax1.plot(temperatures,fictiveTemp, 'bo')
|
||||
#ax1.plot(temperatures,T_f_ns, 'r-')
|
||||
ax1.plot(temperaturesCut,dTfdT, 'bo')
|
||||
#ax1.plot(temperatures[0:5454],dTNMHdT[0:5454], 'bo')
|
||||
#ax1.plot(temperatures[5456:],dTNMHdT[5455:], 'ro')
|
||||
#ax1.plot(temperaturesModel[0:99],dTNMHdT[0:99], 'b-')
|
||||
#ax1.plot(temperaturesModel[100:199],dTNMHdT[100:199], 'r-')
|
||||
ax1.plot(temperaturesInterpol,dTfdTInterpol, 'yo')
|
||||
ax1.plot(temperaturesInterpol,fitResult, 'r-')
|
||||
P.show()
|
||||
"""
|
||||
ax1.plot(temperaturesCut, dTfdT, 'bo')
|
||||
ax1.plot(temperaturesInterpol, dTfdTInterpol, 'yo')
|
||||
ax1.plot(temperaturesInterpol, fitResult, 'r-')
|
||||
plt.show()
|
||||
|
||||
dataOutName = os.path.splitext(str(dataName))[0] + "_dTfdT_tau.fit"
|
||||
dataOut = np.column_stack((temperaturesInterpol, dTfdTInterpol, fitResult))
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user